...

Related documents

Expand
title Click here to expand the list of related documents (RD1-RD10)
Reference ID Document Anchor LK_RD_1 LK_RD_1 RD.1 Global Climate Observing System (2016) THE GLOBAL OBSERVING SYSTEM FOR CLIMATE: IMPLEMENTATION NEEDS, GCOS-200, https://library.wmo.int/doc_num.php?explnum_id=3417 Anchor LK_RD_2 LK_RD_2 RD.2 Merchant, C. J., Paul, F., Popp, T., Ablain, M., Bontemps, S., Defourny, P., Hollmann, R., Lavergne, T., Laeng, A., de Leeuw, G., Mittaz, J., Poulsen, C., Povey, A. C., Reuter, M., Sathyendranath, S., Sandven, S., Sofeiva, V. F. and Wagner, W. (2017) Uncertainty information in climate data records from Earth observation. Earth System Science Data, 9 (2). pp. 511-527. ISSN 1866-3516 doi: https://doi.org/10.5194/essd-9-511-2017 Anchor LK_RD_3 LK_RD_3 RD.3 Group for High Resolution Seas Surface Temperature Data Specification (GDS) v2, Casey and Donlon (eds.), 2012,  https://www.ghrsst.org/wp-content/uploads/ 2016 2021/ 10 04/GDS20r5.pdf Image Removed Anchor SM_RD_4 SM_RD_4 RD.4 W. Dorigo, T. Scanlon, P. Buttinger, R. Kidd, 2019. C3S D312b Lot 4, D3.SM.5-v1.0, Product User Guide and Specification (PUGS): Soil Moisture (v201812). Anchor SM_RD_5 SM_RD_5 RD.5 R. van der Schalie, R. De Jeu, C. Paulik, W. Dorigo, T. Scanlon, C. Reimer, R. Kidd, 2019. C3S D312b Lot 4 D1.SM.2-v1.0,.Algorithm Theoretical Basis Document (ATBD): Soil Moisture (v201812). Anchor SM_RD_6 SM_RD_6 RD.6 W. Dorigo, T. Scanlon, W. Preimesberger, P. Buttinger R. Kidd, 2019. C3S D312b Lot 4 D2.SM.1_v1.0 Product Quality Assurance Document (PQAD): Soil Moisture. Anchor SM_RD_7 SM_RD_7 RD.7 T. Scanlon, W. Dorigo , P. Buttinger, W. Preimesberger, R. Kidd, 2018. C3S D312b Lot 4, D2.SM.2-v1.0, Product Quality Assessment Report (PQAR): Soil Moisture (v201812) (to be issued in April 2019). RD.8 RGI v5.0 Technical Documentation (glims.org/RGI/00_rgi50_TechnicalNote.pdf) RD.9 WGMS FoG Version, incl. attribute description (wgms.ch/data_databaseversions) RD.10 RGI v6.0 Technical Documentation (glims.org/RGI/00_rgi60_TechnicalNote.pdf)

Acronyms

General definitions

Level 2 pre-processed (L2P): this is a designation of satellite data processing level. "Level 2" means geophysical variables derived from Level 1 source data on the same grid (typically the satellite swath projection). "Pre-processed" means ancillary data and metadata added following GHRSST Data Specification, adopted in the case of LSWT.
Level 3 /uncollated/collated/super-collated (L3U/L3C/L3S): this is a designation of satellite data processing level. "Level 3" indicates that the satellite data is a geophysical quantity (retrieval) that has been averaged where data are available to a regular grid in time and space. "Uncollated" means L2 data granules remapped to a regular latitude/longitude grid without combining observations from multiple source files. L3U files will typically be "sparse" corresponding to a single satellite orbit. "Collated" means observations from multiple images/orbits from a single instrument combined into a space-time grid. A typical L3C file may contain all the observations from a single instrument in a 24-hour period. "Super-collated" indicates that (for those periods where more than one satellite data stream delivering the geophysical quantity has been available) the data from more than one satellite have been gridded together into a single grid-cell estimate, where relevant.
Target requirement: ideal requirement which would result in a significant improvement for the target application.
Threshold requirement: minimum requirement to be met to ensure data are useful.

Scope of the document

This document aims to provide users with the relevant information on requirements and gaps for each of the given products within the Land Hydrology and Cryosphere service. The gaps in this context refer to data availability to enable the ECV products to be produced, or in terms of scientific research required to enable the current ECV products to be evolved to respond to the specified user requirements. In this current version the target requirements and gap analysis, ECV products generated by the Soil Moisture Service, the Glaciers Service, the Lakes Service and the Ice Sheets Service are provided. It is noted that the Glaciers Service was not included in the previous version of the Target Requirements and Gap Analysis Document – published in November 2019 - but it will be part of this and all subsequent versions.

...

Within this document each thematic service is addressed separately, but in a similar manner. Initially an overview of each product is provided, including the required input data and auxiliary products, a definition of the retrieval algorithms and processing algorithms versions; including, where relevant, a comment on the current methodology applied for uncertainty estimation. The target requirements for each product is then specified which generally reflect the GCOS ECV requirements. The result of a gap analysis is provided that identifies the envisaged data availability for the next 10-15 years, the requirement for the further development of the processing algorithms, and the opportunities to take full advantage of current, external, research activities. Finally, where possible, areas of required missing fundamental research are highlighted, and a comment on the impact of future instrument missions is provided.]

Executive Summary

Soil Moisture
The C3S soil moisture product comprises a long-term data record called a CDR which runs from 1978 (PASSIVE and COMBINED) or 1991 (ACTIVE) to December 2018. This CDR is updated on a dekadal basis (approximately every 10 days) in an appended dataset called an ICDR. The CDR and ICDR products are provided as NetCDF 4 CF and each of the three products are generated with three temporal resolutions (daily, dekadal, monthly), meaning that the service provides a total of 18 soil moisture products.

...

Reliance on External Research
Since the C3S programme only supports the implementation, development and operation of the CDR processor, any scientific advances of the C3S products entirely rely on funding provided by external programmes, e.g. CCI+, H-SAF, Horizon2020. Thus, the implementation of new scientific improvements can only be implemented if external funding allows for it. This depends both on the availability of suitable programmes to support the R&D activities and the success of the C3S contractors in winning potential suitable calls.

Soil Moisture ECV Service

The C3S Soil Moisture production system provides the climate community with a stable source of soil moisture data derived from satellite observations through the Climate Data Store of the Copernicus Climate Change Service (C3S). The C3S soil moisture product comprises a long-term data record called a Thematic Climate Data Record. This CDR, product version v201812, is updated on a dekadal basis (approximately every 10 days) in an appended dataset called an Interim Climate Data Record (ICDR). Both the CDR and ICDR consist of three surface soil moisture datasets: ACTIVE, PASSIVE and COMBINED. The ACTIVE and the PASSIVE product are created by using scatterometer and radiometer soil moisture products, respectively; the COMBINED product is a blended product based on the former two datasets. The CDRs run from 1978 (PASSIVE and COMBINED) or 1991 (ACTIVE) to December 2018.

...

In this document an analysis is made that compares the current performance of the C3S Soil Moisture products against its potential in the future. This analysis is performed by (1) evaluating both the risk and opportunities of current and future satellite coverage and data availability, (2) the current fitness-for-purpose compared to the user requirements and how this will evolve in the upcoming years, and (3) ongoing and future research that would be beneficial for integration into the CDR and ICDR processing algorithms.

Introduction

This section provides the product specifications and target requirements for the C3S soil moisture product, which have been derived from community requirements as well as international standards. The purpose of this section is to provide these requirements independent of any assessments such that the requirements can be tracked as the product develops. As part of the cyclical process employed in the generation of the C3S product, the needs of the community and hence the requirements presented here will be updated as required.

Soil Moisture Products

Product descriptions

The C3S soil moisture product comprises a long-term data record called a CDR which runs from 1978 (PASSIVE and COMBINED) or 1991 (ACTIVE) to December 2018.  This CDR is updated on a dekadal basis (approximately every 10 days) in an appended dataset called an ICDR.  The theoretical algorithm and the processing implemented in the CDRs and ICDRs are exactly the same and the data provided is consistent between them.

...

Soil Moisture Service: User Requirements

The target requirements are the same for each of the 18 products (listed in Table 1) produced as part of the C3S soil moisture product. The requirements are listed in Table 2.

...

Soil Moisture Service: Gap Analysis

This section provides a Gap Analysis for the soil moisture product. The purpose of this section is to describe the opportunities, or obstacles, to the improvement in quality and fitness-for-purpose of the Soil Moisture CDR. In this section we address the data availability from existing space-based observing systems; development of processing algorithms; methods for estimating uncertainties; scientific research needs; and opportunities for exploiting the new generation of Sentinels.

...

This section provides a Gap Analysis for the soil moisture product. The purpose of this section is to describe the opportunities, or obstacles, to the improvement in quality and fitness-for-purpose of the Soil Moisture CDR. In this section we address the data availability from existing space-based observing systems; development of processing algorithms; methods for estimating uncertainties; scientific research needs; and opportunities for exploiting the new generation of Sentinels.

Description of past, current and future satellite coverage

Figure 2 shows spatial-temporal coverage that is used for the construction of the CDR and ICDR for all three C3S Soil Moisture products (ACTIVE, PASSIVE, and COMBINED). An extensive description of these instruments and the data specifications can be found in the C3S ATBD [RD.5] (Chapter 1, Instruments). This gives an indication of the continuously changing availability of sensors over time as used in the production of the soil moisture data records. In the C3S ATBD [RD.5] (Chapter 3.3, Merging strategy) how this variability is taken into account and how this affects the quality of the final product is explained. The recent developments in the data availability for both scatterometers and passive radiometers are described in Section 1.4.1.1 and 1.4.1.2, and how this potentially affects the COMBINED product in 1.4.1.3.  

...

Figure 2: Spatial-temporal coverage of input products used to construct the CDR/ICDR (a) ACTIVE, (b) PASSIVE, (c) COMBINED. Blue colours indicate passive, red colours active microwave sensors. The periods of unique sensor combinations are referred to as ‘blending period’. Modified from Dorigo et al. (2017).

Active

Active microwave observations used in the production of C3S soil moisture data products (see Table 3) are based on backscatter measurements from the European Remote Sensing Satellites (ERS) 1 and 2's Active Microwave Instrument (AMI) wind scatterometer, and the Advanced SCATterometer (ASCAT) onboard the Meteorological Operational Satellites (MetOp). The sensors operate at similar frequencies (5.3 GHz C-band) and share a similar design. ERS AMI has three antennas (fore- mid-, and aft-beam) only on one side of the instrument while ASCAT has them on both sides, which more than doubles the area covered per swath. ERS AMI data coverage is variable spatially and temporally because of conflicting operations with the synthetic aperture radar (SAR) mode of the instrument. In addition, due to the failure of the gyroscope of ERS-2, the distribution of scatterometer data was temporarily discontinued from January 2001 whereas in June 2003 its tape drive failed. Complete failure of ERS-1 and ERS-2 occurred in 2000 and 2011, respectively.
Two MetOp satellites (MetOp-A and MetOp-B) are currently flown in the same orbit, while MetOp-C was launched in 2018 to replace MetOp-A from 2022. From that time, MetOp-A will remain in orbit to serve as backup in case of failure of one of the other MetOp satellites. Continuation beyond the current MetOp program will be provided by the approved MetOp Second Generation (MetOp-SG) program, which will start in 2021/22 and has the goal to provide continuation of C-band scatterometer and other systematic observations for another 21 years, i.e., at least until 2042. Thus, no potential gap in data coverage from C-band scatterometer missions is foreseen for the next two decades. MetOp-C is not yet integrated in the MetOp-ASCAT CDR used as input to C3S.

...

Passive

Several passive microwave radiometers are available that can be used for the retrieval of soil moisture (Table 4), however due to differences in sensor specifications and data access not all are of interest for direct use within the soil moisture climate data record. In general, a lower frequency observation is preferred for soil moisture retrievals, e.g. C-band and L-band. For an in-depth overview of the impact of different frequencies on the quality of the soil moisture retrievals in the PASSIVE product, e.g. due to vegetation influences or radio frequency interference (RFI), see the C3S ATBD [RD.5] (Chapter 3.1.3, Known limitations).

...

Combined

Due to the wide range of available satellites (both active and passive) now and in the upcoming decade, and the flexibility of the system as explained by the merging strategy in the C3S ATBD [RD.5] (Chapter 3.3, Merging strategy), there is very little risk concerning the extension of the COMBINED product into the future. The current quality is not expected to reduce in the upcoming years, however a successful integration of SMAP soil moisture datasets could lead to further improvements in the COMBINED product. 

Development of processing algorithms

This section is based on Chapter 1.4 in the PUG [RD.4]. Table 5 provides the C3S Soil Moisture product target requirements adopted from the GCOS 2011 target requirements and shows to what extent these requirements are currently met by the latest C3S Soil Moisture products (v201812). As one can see, the CDR and ICDR products currently provided by the system are compliant with C3S target requirements and in many cases even go beyond. Further details on product accuracy and stability are provided in PQAD [RD.6] (methodology to assess) and PQAR [RD.7] (assessment).

...

Methods for estimating uncertainties

The soil moisture uncertainty estimates are included in all C3S soil moisture products: ACTIVE, PASSIVE and COMBINED. A short overview is provided of how the uncertainties are estimated through the Triple Collocation Analysis (TCA, Gruber et al., 2016). Soil moisture uncertainty is the error standard deviation of the datasets estimated through TCA.

Triple Collocation Analysis

This section is based on CCI ATBD (Chapter 6.3.1), CCI PUG (Chapter 6.4.1) and Dorigo et al. (2017). Triple collocation analysis is a statistical tool that allows the estimate of the individual random error variances of three datasets without assuming that any of them act as a supposedly accurate reference (Gruber et al. 2016a&b). This method requires the errors of the three datasets to be uncorrelated, therefore triplets always comprise of an active dataset, (ii) a passive dataset, and (iii) the GLDAS-Noah land surface model, which are commonly assumed to fulfil this requirement (Dorigo et al., 2010). Error variance estimates are obtained as:

...

Figure 3: Average error variances of ESA CCI SM for ACTIVE (upper left), PASSIVE (upper right), and COMBINED (lower left) estimated through triple collocation and error propagation for the period July 2012-December 2015. Long-term (July 2012-December 2015) VOD climatology (lower right) from AMSR2 6.9 GHz observations (adapted from Dorigo et al., 2017).

Opportunities to improve quality and fitness-for-purpose of the CDRs

This section provides a brief overview of improvements that are being considered for introduction into the CDR and ICDR in a short term. This covers algorithm improvements and satellite datasets that have already been evaluated. Many of these ongoing research activities and developments are being undertaken within the ESA Climate Change Initiative (CCI) and CCI+ programmes, the continuation of which has not yet been officially approved. Given the large algorithmic dependency on the CCI programme, many of the following sections are based on the CCI ATBD (Scanlon et al., 2019).
Since the C3S programme only supports the implementation, development, and operation of the CDR processor, any scientific advances of the C3S products entirely rely on funding provided by external programmes, e.g. CCI+, H-SAF, Horizon2020. Thus, the implementation of new scientific improvements can only be implemented if external funding allows for it. The latter depends both on the availability of suitable programmes to support the R&D activities and the success of the C3S contractors in winning potential suitable calls.

ACTIVE products

Higher resolution sampling of ERS-1

An ERS-1 product with an improved spatial sampling (25x25 km) is expected to be provided by ESA in the near future. This would make the ERS-1 product consistent with ERS-2.

Intercalibration of MetOp-B and metOp-A

Intercalibration between MetOp-B and MetOp-A NRT data is only available after June 2015, therefore MetOp-B can only be used after this date. A backward processing of MetOp-B may be performed once intercalibrated data becomes available from H-SAF/EUMETSAT.

Improved vegetation correction for ASCAT

An improved vegetation correction algorithm has been developed for ASCAT (Vreugdenhil et al., 2016) and is currently employed in the offline research product. The correction method has not yet been transferred to the NRT product distributed by HSAF. Once the new implementation is transferred to the operational NRT product, this will also be readily ingested into the CDR and ICDR.

PASSIVE products

Introduction of SMAP soil moisture

As the SMAP observation frequency is similar to SMOS, the current algorithm as developed for SMOS (Van der Schalie et al., 2016 & 2017) can also be applied to the SMAP observations. First results (Van der Schalie & De Jeu, 2016) show good results, however, this dataset first needs to be tested thoroughly within the testing framework of the CCI Soil Moisture before it can be introduced into the C3S soil moisture CDR.

Merging

All products

1.4.4.3.1.1 Separate blending of climatologies and anomalies

...

In the current versions, gaps are only filled if the weight of the available product is above a relatively crudely defined empirical threshold. This threshold will be refined to find a best compromise between data density and product accuracy.

PASSIVE product

1.4.4.3.2.1 Using both night-time and day-time observations

Based on extensive product validation and triple collocation attempts to address the uncertainty of both ascending (daytime) and descending (night-time) modes will be made. Based on these results, this will guide decisions on how both observation modes can be considered in the generation of a single merged passive product, potentially leading to improved observation frequency with respect to the single descending mode used in the current PASSIVE product. An important step towards this approach was made by Parinussa et al. (2016).

ACTIVE product

1.4.4.3.3.1 Data gaps

In the framework of the C3S work, investigations into the potential use of ERS to fill gaps in the ASCAT time series will be undertaken.

Scientific Research needs

In the previous section, research activities that are already in an advanced stage of development and which could potentially be introduced into the CDR and ICDR in the short-term were discussed. However, in the long-term, some fundamental research is needed in order to improve the soil moisture products even further.

ACTIVE products

Inter-Calibration of Backscatter Data Records

To directly compare Level 2 surface soil moisture values retrieved from the ERS-1/2 AMI-WS and MetOp-A/B/C ASCAT, it is a pre-condition that these instruments have more or less exactly the same Level 1 calibration [RD.5]. Unfortunately, this is not yet the case owing to the fact that individual instrument generations underwent a somewhat different calibration procedure. Research is ongoing to improve the calibration between these sensors.

Estimation of Diurnal Variability

ASCAT measurements are performed for descending orbits (equator crossing 09:30, local time) and ascending orbits (equator crossing 21:30, local time). It has been noted that the backscatter measurements and, consequently, the Level 2 (L2) surface soil moisture retrievals from satellite platforms, although not dependent on temperature, show in some regions a difference between morning (i.e., day or sun-lit) and evening (i.e., night or dark) acquisitions (Friesen et al., 2012; Friesen et al., 2007). Currently, it is not clear if these observed diurnal differences are due to changes in the instrument between ascending or descending passes (e.g. due to the strong temperature differences in the sun-lit or dark orbital phases), shortcomings in the retrieval algorithm (e.g. neglecting diurnal differences in vegetation water content), or if these are just a natural expression of diurnal patterns of the surface soil moisture content. The underlying reasons for diurnal differences are to be investigated by comparing satellite ascending and descending orbit soil moisture retrievals.

Improved Modelling of Volume Scattering in Soils

It has long been noted that backscatter measurements over desert areas and semi-arid environments during a long dry spell exhibit an unusual behaviour that may lead to a situation where soil moisture from scatterometers is often less accurate than radiometer retrievals (Wagner et al., 2007, Gruhier et al., 2009). Characteristics of backscatter should be explored in more depth in very dry environments to recognie and potentially correct for spurious soil moisture retrievals.

Dry and Wet Crossover Angles

The crossover angle concept adopted in the retrieval method for scatterometers, states that at the dry and wet crossover angles, vegetation has no effect on backscatter (Wagner, 1998). These crossover angles have been determined empirically based on four study areas (Iberian Peninsula, Ukraine, Mali, and Canadian Prairies). Nevertheless, the empirically determined dry and wet crossover angles are used on a global scale in the surface soil moisture retrieval model. A known limitation of the global use of these crossover angles is that, depending on the vegetation type, or more precisely the evolution of biomass of a specific vegetation type, crossover angles may vary across the globe, which is not yet considered in the model. Furthermore, for some regions on the Earth's surface the crossover angle concept may not be applicable, in particular regions without vegetation cover (i.e., deserts). Recent investigations have shown that improved retrievals can be obtained by a local optimisation of cross-over angles (Pfeil et al., 2018).

Backscatter in Arid Regions

In arid regions, or more specifically in desert environments, it appears that the dry reference shows seasonal variations, which are assumed to reflect vegetation phenology. However, this cannot be true for desert environments, which are characterised by very limited or no vegetation at all. In principle, seasonal variations of the dry reference are desirable to account for backscatter changes induced by vegetation; referred to as vegetation correction. Vegetation correction is based upon changes in the slope parameter, which can be also observed in desert environments. These variations seem to have a big impact particularly in areas with very low backscatter. Hence, it needs to be clarified whether it is a real physical process, noise or something else reflected in the slope parameter.

PASSIVE products

Updated temperature input from Ka-band observations

Land surface temperature plays a unique role in solving the radiative transfer model and therefore directly influences the quality of the soil moisture retrievals. The current linear regression to link Ka-band measurements to the effective soil temperature has been re-evaluated by Parinussa et al. (2016) for daytime observations. An update to the linear regression for land surface temperature showed a significant increase in soil moisture retrieval skill. This research highlighted the importance and impact of correct temperature input into the algorithm. Further scientific work is needed to improve the surface temperature derived from microwave observations in order to significantly improve the skill of the soil moisture retrievals. Also, in order to remove model dependency for the L-band soil moisture retrievals that use modelled surface temperature as an input, investigations into combining the L-band observations with Ka-band observations from other satellites with similar overpass times are needed.

Development of a solely satellite based PASSIVE soil moisture data record

Within the climate community there is a strong preference for climate records that are solely satellite based. Any additional dataset that is used in a soil moisture retrieval algorithm could potentially lead to a dependency between a model and an observation. This is also why research was set up to investigate the possibility of developing an independent ancillary-free soil moisture dataset (Scanlon et al., 2019). Ancillary data could also have a strong impact on the spatial distribution of soil moisture. Artificial patterns of the 1 degree FAO soil property map are still visible in the original LPRM soil moisture product, however, these patterns disappear when only the dielectric constant is used. More research is needed to derive soil moisture from the dielectric constant records without making use of any ancillary datasets; with such an approach an independent dataset would be created that could be used as a benchmark for different modelled soil moisture datasets.

COMBINED products

Improved sensor inter-calibration

Currently, inter-calibration between active and passive datasets is done using CDF-matching against a long-term consistent land surface model. However, in order to achieve a full model independence of the CCI SM products alternative inter-calibration approaches will be investigated, for instance using lagged-variable based approaches or homogeneity tests (Su et al., 2015, 2016). Also the use of an L-band climatology as scaling reference for the COMBINED product is being investigated (Piles et al., 2018) .

Error characterisation

Estimation of random uncertainty per observation

The current C3S soil moisture product is generated with associated uncertainty estimates.  These estimates are based on the propagation of uncertainties, estimated with the triple collocation analysis, through the processing scheme; this process is described within the ATBD [RD.5]. Notice that these uncertainty estimates represent the average random error variance of the entire considered time period, which is commonly assumed to be stationary in the triple collocation. Future research shall focus on the estimation of the uncertainties of each individual measurement, which is driven, e.g., by the vegetation canopy density or the soil wetness conditions at the time of observation.

Stability assessment and correction

To test for inhomogeneities, the MERRA-2 data is compared to the C3S soil moisture; this procedure is described in the PQAD document (Dorigo et al., 2017). The inhomogeneity testing is achieved by first identifying potential locations of breakpoints in the time-series (for example where a change in sensors used occurs). Where the discontinuity values are greater than 1 % it is considered that this indicates a potential discontinuity in the time-series. The stability is then expressed in terms of the longest "stable" time-period within the dataset for each pixel. This gives a qualitative indication of the stability of the dataset, however, in future assessments of the dataset, the stability will be expressed in terms of m³ / m³ / y, thereby allowing demonstration against the KPIs. In addition, it is currently being investigated whether a break, once detected, can be corrected for. In this way, the "stable" time period can be extended.

Opportunities from exploiting the Sentinels and any other relevant satellite

As described in 1.4.1, there are many upcoming satellites relevant for soil moisture retrievals that are expected to be launched in the upcoming years. This section will give a more in-depth description of the instruments that could have a substantial impact on the quality of the soil moisture CDR and ICDR.

Sentinel-1

Soil Moisture retrieved through Sentinel-1 at 1km spatial resolution is currently in evolution at the Copernicus Global Land Service (Bauer-Marschallinger et al., 2019). Integration of a dataset like this could drastically improve the spatial resolution of the CDR, but only for data after 2014. So, for the data to be used, a strategy for handling a CDR with changing spatial resolution over time has to be developed. Sentinel-1 also has the potential to improve the soil moisture record spatial resolution using downscaling approaches together with other sensors. The combination with the ASCAT sensor seems promising (Bauer-Marschallinger et al., 2018) but, for integration into a CDR, the current approaches still need to overcome issues with temporal and spatial consistency.

SMAP

As described in Section 4.2.2, one of the first steps that should be taken is the further development and integration of a SMAP-based dataset. SMAP (Entekhabi et al., 2010) is the latest L-band mission specifically designed for soil moisture retrievals. Although the radar failed shortly after launch, the radiometer is functioning well and produces L-band brightness temperature with a higher radiometric accuracy then previous L-band missions. Secondly, due to the improved RFI mitigation system, RFI has become a relatively small issue concerning L-band retrievals. This leads to higher quality L-band soil moisture retrievals, including in areas like South-East Asia which used to have many RFI issues. Integration of SMAP based soil moisture retrievals could significantly improve the CDR.

Water Cycle Observation Mission (WCOM)

Although there are many uncertainties and concerns around the WCOM (Shi et al., 2016) mission, e.g. potential data accessibility issues, it would be a very interesting mission for the further development of the passive soil moisture retrieval algorithm. As described in Table 4, the payload of the WCOM satellite includes an L-S-C (1.4, 2.4 and 6.8 GHz) tri-frequency Full-polarized Interferometric synthetic aperture microwave radiometer (FPIR) and a Polarized Microwave radiometric Imager (PMI, 6 frequencies between 7.2 to 150 GHz). This wide range of simultaneous observations provide a unique tool for further research on soil moisture retrieval algorithms. Firstly, this allows for simultaneous retrieval of temperature from the Ka-band, which can be used in the soil moisture retrieval from the L-band observation, opposed to using modelled temperature. Secondly, this provides an opportunity for the first time to study S-band based soil moisture retrievals. Thirdly and most importantly, it provides a perfect tool for the development of a multi-frequency soil moisture retrieval approach based on L-, S-, C-, and X-bands, potentially leading to improved soil moisture retrievals.

Copernicus candidate missions under consideration

Two ESA missions that are currently under consideration as Copernicus candidate missions (http://missionadvice.esa.int/), a Microwave Radiometer Mission and an L-Band SAR Mission, would be an important step forward in safeguarding the future of the soil moisture climate records. With the upcoming MetOp-SG and Sentinel-1s, the active soil moisture retrievals have an expected satellite support up to 2040. However, for the passive soil moisture retrievals, and especially the development of long-term L-band based climate data records, the future is uncertain after SMOS and SMAP. For C-band frequencies and above, there is also some uncertainty after GCOM-W2. Therefore, these missions could form an important step in safeguarding the continuation of soil moisture climate data records with at least the same level of quality in the upcoming decades.

Glaciers ECV Service

Introduction

The C3S Glacier Distribution and Glacier Change Service provide 3 Climate Data Records (CDRs) to the Climate Data Store (CDS). The three CDRs are provided via the latest  versions of the Randolph Glacier Inventory (RGI6.0) and the Fluctuations of Glaciers (FoG) database (https://doi.org/10.5904/wgms-fog-2019-12). The Glacier Change Service provides the two CDR’s of elevation change and mass change from the FoG database. The Glacier Distribution Service provides the CDR of glacier area, and alongside this it also provides glacier outlines, both of which are provided via RGI6.0 which in turn is based on data from the Global Land Ice Measurements from Space (GLIMS) data repository.

The C3S Glacier Distribution and Change Service provides basically two types of data sets in a vector format, glacier outlines with attribute information (shape files with polygon topology) and glacier elevation changes (shape files with point topology). Their basic structure and contents are adopted from the Randolph Glacier Inventory (RGI, see Pfeffer et al. 2014) and the Fluctuations of Glaciers (FoG) database from WGMS. Both datasets are provided in a first version and are later on updated, extended and improved through the work progress in the project. For glacier area, the updates will be related to selected parts of the full datasets and intially provided to GLIMS rather than the CDS. For elevation changes, the updates will be integrated in the full dataset (i.e. new FoG version) before its latest update (also including other contributions) is provided to the CDS. In this document, the details applied to RGI5.0 [RD.8] and RGI6.0 are presented, which serve as v1 and v2 for the Glacier Distribution Service and the FoG database [RD.9] for the Glacier Change Service.

Glacier Products: The Glacier Distribution Service

The following descriptions have been adopted from the technical note describing the RGI (Arendt et al. 2015) that is available online at [RD.8]. They apply in the same way to RGI6.0 and the ICDRs produced by the Glacier Distribution Service.

Technical Specification of the dataset

The RGI is provided as shape files containing the outlines of glaciers in geographic coordinates (longitude and latitude, in degrees), which are referenced to the WGS84 datum. Data are organised by first-order region. For each region there is one shape file (.SHP with accompanying .DBF, .PRJ and .SHX files) containing all glaciers and one ancillary .CSV file containing all hypsometric data. The attribute (.DBF) and hypsometric files contain one record per glacier. Each object in the RGI conforms to the data-model conventions of ESRI ArcGIS shape files. That is, each object consists of an outline encompassing the glacier, followed immediately by outlines representing all of its nunataks (ice-free areas enclosed by the glacier). In each object, successive vertices are ordered such that glacier ice is on the right.

Data fields and hypsometry file

The following attributes are provided with the dataset: GLIMS-ID, RGI-ID, first and second order RGI region, glacier name, area (size in km²), begin and end date, minimum, median and maximum elevation, length, slope, aspect, latitude, longitude, reference, principal investigator (PI), sponsoring agency, and publication. As an additional file (csv format) the area-elevation distribution (hypsometry) is provided for each glacier in 50 or 100 m bins. They are described in the following in more detail. Table 6 provides an overview of their key characteristics. Please note that this is a selection of attributes. For the full information please download the individual files at glims.org/RGI.

...

Data sources

The glacier outlines in the RGI are derived from a variety of sources by a global community (see Pfeffer et al. 2014). Whereas the main input data source is satellite data (Landsat, ASTER, SPOT), analysts have also digitised outlines from topographic maps, aerial photography and other sources. A major input dataset for the RGI stems from the GLIMS database that has been compiled by a large number of participants since 2000 (e.g. Kargel et al. 2005). Whereas GLIMS is a multi-temporal database containing all outlines being made available, the RGI is a snapshot in time referring only to one dataset. However, even within a small region glacier, outlines can refer to different points in time as clouds may cover parts of a satellite scene and require mosaicking. In part, also the outline of an individual glacier can be composed from several scenes spanning several years. For this reason, a Begin and End date is provided for each glacier in the attribute table. In a few cases, however, the entries in this field are guesswork as no information was provided by the analyst.

...

Glacier outlines that were separated from their neighbours when received were accepted without change, subject only to initial quality control. However, many glacier outlines were originally obtained or contributed as glacier complexes, that is, as collections of contiguous glaciers that meet at glacier divides but not being separated. Semi-automated algorithms were used (Bolch et al. 2010, Kienholz et al. 2013) to separate these complexes into glaciers using a watershed algorithm applied to the most appropriate DEM available. The quality of raw output from the algorithms primarily depends on the quality of the DEM used to calculate the divides. Even when a high-quality DEM is available, the algorithm output requires intense manual checking and corrections. These checks were carried out in detail only in a few regions. Elsewhere, in many cases further work is necessary to inspect the quality of drainage divides. We will also use scenes from Sentinel-2A/B to create glacier outlines and the new ArcticDEM, TanDEM-X DEM, and ALOS AW3D30 DEM (as available and appropriate) to derive drainage divides and topographic parameters for each glacier.

Data Citation requirement

The following reference must be cited when using the RGI version 5.0: Arendt et al. (2015) [RD.8].  For RGI6.0, please cite RGI consortium (2017) [RD.10].

The Glacier Change Service

The following descriptions have been adopted from the attribute description of the FoG database (version doi: 10.5904/wgms-fog-2016-08) hosted by the WGMS [RD.9]. 

Technical specification of the dataset

The time series of glacier-wide changes in elevation and mass are both provided as shape files containing the location of the glacier label point in geographic coordinates (longitude and latitude in degrees), which are referenced to the WGS84 datum and some general statistic information about the glacier. For both observation methods, the shape files with the glacier locations (.shp with accompanying .dbf, .prj and .shx files) come with one ancillary .csv file containing the time series of glacier change observations based on the glaciological and on the geodetic method, respectively. Table 8,Table 9,Table 10 andTable 11 summarise the detailed descriptions provided in the following.

Data fields

Data fields of tables GLACIER_MASS_BALANCE_SERIES and GLACIER_ELEVATION_CHANGE_SERIES

...

Data sources

Internationally coordinated glacier monitoring began in 1894 and the periodic publication of compiled information on glacier fluctuations started one year later (Forel, 1895). For data compilation, the WGMS and its predecessor organisations have been organising periodical calls-for-data through an international scientific collaboration network with National Correspondents for, currently, 36 countries and thousands of contributing observers around the world. The data is published in the bi-annual 'Global Glacier Change Bulletin' (GGCB) series which serves as an authoritative source of illustrated and commentated information on global glacier changes based on the latest observations from the scientific collaboration network of the WGMS.
The glaciological method (cf. Cogley et al. 2011), based primarily on ablation stake and snow pit measurements, provides mass-budget estimates that are integrated within glacier-wide averages of mass changes in metres of water equivalent (m w.e.). The glaciological method provides quantitative results at high temporal resolution, which are essential for understanding climate–glacier processes, and for allowing the spatial and temporal variability of the glacier mass balance to be captured, even with only a small sample of observation points. It is recommended to periodically validate and calibrate annual glaciological mass-balance series with dekadal geodetic balances in order to detect and remove systematic biases.
The geodetic method (cf. Cogley et al. 2011) provides overall glacier volume changes over a longer time period by repeat mapping from ground, air- or spaceborne surveys and subsequent differencing of glacier surface elevations. The geodetic method includes all components of the surface, internal and basal balances and can be used for a comparison with the glaciological (surface-only) mass budgets of the same glacier (Zemp et al. 2013) and for extending the glaciological sample in space and time (Cogley 2009). For the conversion of geodetic results to glaciological mass-balance units (m w.e.), a glacier-wide average density of 850±60 kg m^–3 is commonly applied (cf. Huss 2013). The results of the glaciological and the geodetic methods provide conventional balances, which incorporate climatic forcing and changes in glacier hypsometry and represent the glacier contribution to runoff (cf. Cogley et al. 2011).

Data citation requirement

When using the glacier elevation change data, the original data source (as indicated in the dataset) and/or the WGMS FoG database must be cited: WGMS (2017).

Glacier Service: User Requirements

International documents (GCOS, IGOS)

Several requirements for the ECV Glaciers and Ice Caps have been described in documents from international organisations related to the two main products (a) glacier outlines/inventories and (b) elevation/mass changes provided by the Copernicus Glacier Distribution and Change Service. Foremost, GCOS (2011) identifies these Products under T.3.1 and T.3.2:
Product T.3.1: 2D vector outlines of glaciers and ice caps (delineating glacier area), supplemented by digital elevation models for drainage divides and topographic parameters
Product T.3.2: Elevation change of glaciers and ice caps, from geodetic methods, in regions where outlines are available.
The listed benefits of both products are:

...

Table 14: Target requirements for glaciers according to GCOS (2016).

Community needs

Information about glacier extents and elevation changes are required for a wide range of applications and calculations serving societal needs. These range from hydro-power production at a local scale, to river-runoff at a regional scale and determination of their contribution to sea level at a global scale. It includes scientific applications such as modelling of past and future glacier extents, climate change impact assessment and improved process understanding from observed changes, but also early warnings related to hazards and potential flooding (due to combined snow and ice melt in late spring) from governmental agencies and national hydrological services. They all need reliable input data on glaciers and their changes through time for accurate modelling of impacts.

Glacier Distribution Service

Today, users within the glaciological community can convert, or even directly assimilate, shape files in their models and applications and therefore we provide our products also in this most precise vector datasets. Downstream applications might aggregate or grid the information to a sampling distance of choice, but this is very specific to the respective application / model and thus not useful to be pre-scribed by the data providers. In an ideal case, glacier outlines are compiled globally each year to combine them with other data sets and get the most precise modelling results. However, as mentioned above, glaciers do not change that fast - in most regions of the world, i.e. the outlines obtained in a specific year might be used for a time period of +/-5 to 10 years without introducing too large errors in the derived products. The uncertainty in their extents is often higher than their changes over a 5 to 10 year period, in particular when they are debris covered or frequent seasonal snow is hiding the true glacier perimeter. This might be different when working on a more regional or local scale, but for such applications higher quality datasets are often available or can be created.
Similarly, the community need for a consistent interpretation of glacier extents has not yet been achieved as glacier outlines are created by a globally distributed community (GLIMS/RG) and the interpretation of what a glacier is and which parts belong to it can depend on the application and experience of the analyst. A key point is that glacier outlines are often created by science projects and these might have different needs compared with other applications. Moreover, there is some room for interpretation and related differences might not be classified as "this is right" or "this is wrong", i.e. both can be correct despite being different. Due to this variability in interpretation of glacier extents by the respective analyst, it is not advisable to use datasets other than RGI for change assessment without a prior careful checking of their interpretation rules. A key user need would be that change assessment can be done with any dataset in the GLIMS glacier database, but we are not yet there as the differences in interpretation can be as large as real glacier changes over a longer period. A further problem is that sometimes satellite scenes with adverse snow conditions were used and the mapped glacier extent is way too large. Here is some room for improvement of existing datasets.
From a more technical point of view one can summarise community needs for glacier outlines as:

...

All of the above needs are not yet fully met and one goal of the work in C3S will be to improve on this situation. Related target regions described in the C3S document GAD (2017).

Glacier Change Service

The situation is similar for information on glacier elevation changes. Whereas a standardised dataset on glacier fluctuations (length and mass) with global coverage is available from the WGMS database (the FoG dataset also provided to Copernicus), the dataset has shortcomings in spatio-temporal coverage. Only a few glaciers have direct and at least annual measurements of mass balance over more than 30 years (about 40 out of 200,000). However, these are used as reference glaciers for regional to global scale applications. Users either use the data provided by WGMS as it is, or they convert / aggregate them on a diversity of spatial (e.g. per RGI region) and temporal (e.g. pentads) scales to obtain averages for comparison with other datasets (e.g. Vaughan et al. 2013). Apart from calculating the sea-level contribution of glaciers (in its simplest form this can be obtained by multiplying the averaged mass balance for a region with the glacier area in this region) or detection of climatic trends and variability (Huss et al. 2010), a key application is run-off contribution and water resources assessment (Bliss et al. 2014). For many, in particular drier climatic regions, the meltwater from glaciers released during the hot summer months is key for agriculture and livelihood (Kaser et al. 2010). Also, direct economic impacts of dwindling glaciers are relevant (Vergara 2007) and the transformation of a landscape hosting glaciers to a desert of rocks and unconsolidated debris is both heart-breaking and dangerous. Whereas it is unclear on how documenting these changes can help in slowing down its progress, the awareness and sensibility it has created for a larger problem (global temperature rise) across the population is amazing. Still, when climate change or its impacts should be illustrated, a comparison of glacier pictures from today and in the past is shown (www.gletscherarchiv.de, nsidc.org/data/glacier_photo).

...

Whereas glacier-wide results are available from WGMS in a standardised format (WGMS 2017) allowing easy implementation in models or spread-sheets, the best way of providing distributed elevation change fields derived from geodetic methods to the community has yet to be explored (this is currently done in the framework of the Glaciers_cci project). Common possibilities are to provide one (mean) value per glacier (linked to an ID and coordinates), to provide changes averaged per elevation bin for each glacier (resulting in a somewhat longer dataset), or to provide results in a grid format (GeoTIFF) with a pre-scribed spatial resolution. The latter might have restrictions in dissemination, as sometimes (national) DEMs cannot be shared, but also requires much larger files. With the increased spatial resolution of freely available DEMs (e.g. 2 m for the Arctic DEM derived from high-resolution Worldview images) the latter point might become an issue in the future. For science applications it is clear that users always want to have raw data to have full control over the further processing (e.g. methods and error propagation), but this generally results in incomparable results. For C3S, all datasets are provided with a standardised processing in the formats described in preceding sections (section 2.3.1, section 2.2.1).

Glacier Service: Gap Analysis

Description of past, current and future satellite coverage

Glacier Distribution

Historic Development

For glacier outlines derived from optical satellite data (Landsat type) the possibilities for data retrieval have constantly improved over the past decades. However, there has been one major break point that changed everything: the opening of the Landsat archive in 2008 (Woodcook et al. 2008, Wulder et al. 2012). Without this step it would have never been possible to utilise the vast archive of images (>3 million scenes) for global scale applications, in our case the global glacier inventory (Pfeffer et al. 2014). However, it has to be mentioned that the glaciological community found a way to have free access to multi-spectral (15 m) ASTER data (a sensor on-board the Terra spacecraft launched in 1999) by registering to the GLIMS (Global Land Ice Measurements from Space) initiative and establishing GLIMS as a major science application of the ASTER data acquisition strategy (Raup et al. 2000). With the GLIMS database in place and algorithms for automated glacier mapping being developed, population of the database with glacier outlines slowly started. There were three main bottlenecks in the earlier days of glacier mapping:

...

In 2008, the archive was opened, and all scenes were provided as already orthorectified GeoTIFFs. This had the enormous advantage that analysts did not have to do this important but very time-consuming step by themselves. The quality of the orthorectification was overall sufficient (about /- 1 pixel RMSE) but regionally variable and mostly dependent on the source data used for the GLS2000 DEM that served as a baseline dataset for this purpose. That DEM was largely based on the SRTM DEM and other national (NED) or military sources outside the SRTM coverage. A first crisis occurred for global glacier mapping when the scan line corrector of Landsat 7 ETM sensor failed in May 2003. Although scenes were still usable in the middle third, the so-called SLC-off scenes now had strong limitations in global coverage. In consequence, the 20-year old (but still working) Landsat 5 with its Thematic Mapper (TM) sensor was reactivated and helped to complete coverage. When TM failed in 2011, it had more or less continuously acquired calibrated images of the Earth's surface for 27 years (since 1984), the longest time-series for a civilian EO satellite on record (Belward and Skoien 2015).
The regional data gaps in 2012 caused only a small problem in time series continuity as the successor of the Landsat 7 ETM+ sensor, Landsat 8 OLI, was on its way and started acquisitions in 2013 with unprecedented quality (e.g. new bands, revised spectral ranges, and 16-bit quantisation). As the free data access policy was continued and the geometric quality of the orthorectification further improved, a second promising phase of global glacier mapping started. This second phase reached new dimensions with the launch of Sentinel-2A in June 2015 (and Sentinel-2B in March 2017), as its much higher spatial resolution (10 m) and larger swath width (290 instead of 180 km) allows glacier mapping (and the still required manual corrections of debris cover) with unprecedented quality (Paul et al. 2016). Therefore, compared to the situation 5 years ago (where only disturbed ETM+ scenes were available) or 10 years ago (where only orthorectified GLCF scenes were freely available), we are now in a glaciologist's paradise. The main remaining issue to be solved by space agencies in the very near future is the replacement of the outdated (year 2000) and coarse resolution (90 m) DEMs used for orthorectification of high-resolution satellite imagery. Over rapidly changing glaciers in steep topography the resulting poor geocoding accuracy hinders several glaciological applications (Kääb et al. 2016).

Spectral and spatial properties

Automated glacier mapping (clean ice) is largely based on calculating a simple band ratio (e.g. red/SWIR) and applying a threshold to create a binary glacier mask that can be converted to glacier outlines using a raster to vector conversion (e.g. Hall et al. 1988, Bayr et al. 1994, Paul et al. 2002). This works well, as spectral properties of ice and snow are very different in the SWIR (where both have a very low reflectance) compared to the red or NIR. Owing to the windows of atmospheric transmission and physical principles, the spectral ranges of the required spectral bands are very similar on all optical sensors that can be used for glacier mapping (Table 15). Accordingly, the methods developed for automated glacier mapping with Landsat TM can also be used for Landsat ETM+ and OLI, Terra ASTER, SPOT HRV, Sentinel-2 MSI and several others. The only requirement is a spectral band in the SWIR, otherwise only manual delineation of outlines can be applied (this is for example required for all the very-high resolution sensors such as Quickbird, GeoEye, Kartosat or Worldview as well as for aerial images).
A change in the spectral range of the panchromatic band on Landsat 8 OLI (now covering only green and red instead of green to NIR, see Table 15) now allows using the 15 m band also for glacier mapping with a pan/SWIR ratio (Paul et al. 2016). Accordingly, the resulting outlines are two times sharper. By pan-sharpening also the other visible bands with the 15 m band (which is now possible) also the quality of the manual editing (debris cover) can be improved. However, resolving crevasses, and thus giving a realistic glacier representation, seems to require at least 10 m spatial resolution. On the downside, the ASTER sensor lost its SWIR band in 2008, so automated glacier mapping only works with scenes acquired before that date.

...

All Landsat TM, ETM+, and OLI sensors have the same spatial resolution in the red, NIR and SWIR bands (30 m) resulting in the spatial consistency of products. For ASTER, SPOT and Sentinel-2 the resolution of the SWIR band is half as good as for the visible and near infrared (VNIR) bands (15 / 30 m for ASTER and 10 / 20 m for SPOT and Sentinel-2). If the higher resolution product should be generated from these sensors, it is required to first resample the SWIR band two times (at best using a simple bilinear interpolation). All sensors have sun-synchronous orbits with acquisition times around 10:30 am local, a compromise between solar elevation (casting stronger shadows when low) and cloud development (often starting before noon in mountain regions). Apart from the global acquisition strategy, seasonal snow at the end of the ablation (or dry) period and clouds are main obstacles to produce glacier outlines regularly. In some regions it can take more than 10 or 15 years before the next useful acquisition is made (Paul et al. 2011). For this reason, it is necessary to analyse new acquisitions each year and process them as required.

Global coverage

Apart from clouds and seasonal snow, global coverage of glaciers is also limited by the acquisition strategy. Whereas this has changed today, as Landsat 8 and Sentinel-2 acquire images more or less continuously, the limited on-board storage capacity of Landsat TM/ETM+ resulted in a pattern of acquisitions around ground receiving stations (Goward et al. 2006). As the network of these stations was successively extended, more and more regions where covered. During the commercial phase of Landsat acquisitions in the 1990s, in several regions images were only acquired upon request, so that large regions are not covered. In consequence, nearly all glaciers in High Mountain Asia are not covered before 1988 (Figure 4).
A second consequence of the distributed acquisition is that the orthorectified product (L1T) revealed freely to the community in 2008 was constrained to the holdings in the USGS archives at LPDAAC. Scenes outside the US were strongly underrepresented, and it was with some luck to have useful acquisitions over a particular region already covered. The still ongoing Landsat Global Archive Consolidation (LGAC) is aiming at transferring all Landsat scenes from around the world into the USGS archive and processing them to the L1T standard (Wulder et al. 2016). For this reason, global coverage is constantly increasing and new possibilities of product generation emerge.
Spatial coverage with Sentinel-2 will also increase in the coming years and the 5 day (or shorter) repeat cycle with both Sentinels will help to increase the chance for cloud-free acquisitions over time. The problem with seasonal snow cover, however, will remain for the time being and it should not be expected that every region with glaciers has a useful acquisition within 5 years.

Figure 4: Landsat TM acquisitions from 1982 to 2005 (from Goward et al. 2006). Overall, considering the foreseen continuation of the Landsat series and the already planned Sentinel-2C/D satellites, the outlook for satellite-based glacier mapping and monitoring are very promising. The possibilities we are facing in the coming years might revolutionise our understanding of glaciers as the now available sensors can be applied to several other glaciological investigations (e.g. flow velocities and snow lines) with unprecedented spatial and temporal resolution.

Glacier Change

Historic Development

For more than a century, the World Glacier Monitoring Service (WGMS) and its predecessor organisations have been compiling and disseminating standardised data on glacier fluctuations. The historical development of this service as well as of the related datasets and science are summarised in Haeberli (2008), Zemp (2012) and Zemp et al. (2014).

The main variable currently observed in standardised formats are changes in glacier mass, elevation and volume, area, and length. Glacier changes are observed using in-situ and remote sensing methods. The glaciological mass balance is obtained from ablation stake and snow pit measurements and provides seasonal to annual information on glacier contribution to runoff. Geodetic methods from in-situ, airborne and space borne platforms provide multi-annual to dekadal information on glacier elevation changes. Based on assumptions on the density of snow, ice and firn, the observed geodetic elevation changes can be converted to mass balance and runoff contribution (e.g. Huss 2013). Glacier elevation change and mass balance are a relatively direct reaction to the atmospheric conditions. They are, thus, relatively easy to interpret but comparably difficult to measure. Glacier front variations on the other hand, are an indirect and delayed reaction to climate change that are, thus, more difficult to interpret but easy to measure (from both in-situ and remotely sensed observations). Their much longer time series allows extending the observational series back into the Little Ice Age period.

Global Coverage

Zemp et al. (2015) provide a detailed overview of the available datasets and discuss the potential and the shortcomings for scientific assessments. The Global Glacier Change Bulletin (WGMS 2015) and the GTN-G Global Glacier Browser (http://www.gtn-g.org) provide a periodically updated overview on and access to all data products, respectively. Figure 5 provides a graphical representation of global glacier distribution and the fraction covered by available glacier change observations.

...

More recently, DEMs (e.g. SPOT SPIRIT, ASTER, ALOS AW3D30, TanDEM-X) from satellite missions after SRTM are used to determine elevation changes over a more recent period (e.g. Gardelle et al. 2013, Berthier et al. 2016, Rankl and Braun 2016, Brun et al. 2017, Dussaillant et al. 2019, Braun et al. 2019). Integrating the results of these and other forthcoming studies into the WGMS database is one key goal of the C3S glacier service.

Development of processing algorithms

We follow the further development of processing algorithms closely. They will be performed as part of the CCI+ project but also by the science community. In case improvements appear in the literature, we will test if the method is sufficiently robust for the products we wish to provide for C3S. For the brokered datasets we will also analyse the methods used to generate them before forwarding them to the CDS. For example, the currently applied automated processing lines for optical stereo images have created much more robust elevation change trends compared to earlier calculations (e.g. Brun et al. 2017, Dussaillant et al. 2019).

Methods for estimating uncertainties

General remarks

The differences in interpretation described in 2.4.2 also limit the possibilities for a rigorous quality assessment of the generated products. In general, the interpretation differences are much larger than those introduced by the method applied and often also larger than changes over a 5-10 year period. Another problem is that the possibility for quality assessment and validation depends on the availability of appropriate higher resolution (e.g. aerial photography) or higher quality (e.g. LIDAR DEMs) datasets. However, these are often not appropriate for a comparison, as they are obtained at a different date (outlines) or over another period (DEMs). For example, a comparison is impossible when snow conditions are worse in the high-resolution data. Another problem with such datasets is that they are in general very expensive or even unavailable.

Real validation is, thus, only performed occasionally and other measures for quality assessment have been developed, partly considering the variability in interpretation as a measure of uncertainty. In addition, uncertainty assessment for elevation changes derived from DEM differencing has issues that are difficult to take into account. Whereas several statistical measures can be applied, interpolation of data voids, artefacts, and differences in spatial resolution are difficult to quantify (McNabb et al. 2018). In the following two sections we briefly summarise the main methods for both products. They are sorted according to the level, which is related to the workload required to perform the related calculations. A full description can be found in the Uncertainty Characterisation Report (UCR) of the Glaciers_cci project (Glaciers_cci, 2016).

Accuracy assessment for glacier outlines

In principle, all glacier outlines are quality checked by the analyst before submission. This glacier-by-glacier quality control is required, as the automated methods do not map glacier ice under debris cover or partly also in shadow (omission errors) and additional map turbid lakes or ice bergs as glaciers (commission errors). These had to be corrected/removed before an outline has acceptable quality (Figure 6) and is, thus, the first step in quality assessment (Level 0a). A critical point for change assessment is to determine if an observed change is significant, i.e. larger than the uncertainty. This requires determination of a quantitative accuracy measure. Common practice is to either:
(a) adopt a value from more detailed studies (e.g. Paul et al. 2013) that are specific to the dataset under consideration,
(b) calculate a minimum and maximum extent for all glaciers by adding a buffer related to typical uncertainties (+/-1/2 or 1 pixel) and report the standard deviation as an estimate of accuracy (L1),
(c) digitising or correcting several (at least 5 better 10) glaciers with different characteristics (e.g. large/small, debris/clean) at least three (better 5) times and use the normalised standard deviation of the derived areas as a measure for accuracy (L2).
In case of appropriate high-resolution or field data being available to create a reference dataset, it is possible to determine both, uncertainty and accuracy. The direct comparison of glacier extents can be assigned to Level 3a, whereas the more sophisticated comparison of outline position would be Level 3b. In this latter case, the result would also include uncertainties in geolocation so that the related product accuracy is in general worse than the pure comparison of glacier areas. Table 16 provides an overview for all measures.

...

Accuracy assessment for glacier elevation changes (DEM differencing)

Until recently, comprehensive uncertainty assessments have rarely been carried out and mass balance and elevation change data have often been applied using rough error estimation or even without consideration of errors. Based on an expert workshop, Zemp et al. (2013) propose a framework for reanalysing glacier mass balance series that includes conceptual and statistical toolsets for assessment of random and systematic errors, as well as for validation and calibration (if necessary) of the glaciological with the geodetic balance results. These are widely accepted and applied by the mass balance observers (e.g. Huss et al. 2015, Andreassen et al. 2016, Thomson et al. 2017). The geodetic method provides a great opportunity to tackle this challenge. However, a basic requirement is a sound uncertainty estimate for such geodetic change assessments.
Table 17 provides an overview of the measures for uncertainty assessment that can be applied to the elevation change product (DEM differencing). The table does not consider uncertainties introduced during post-processing, e.g. the method selected to fill data voids and reduce artefacts. The mandatory step to be performed in any case is image co-registration (e.g. following Nuth and Kääb 2011) horizontally as well as vertically if this is sensible and both datasets have the same geodetic datum. As a further minimum requirement and first quantitative descriptor of product accuracy, the elevation differences over stable ground should be given. Stable means outside of glaciers, hydro-power lakes and forests, as these have an impact on the differences as well as conditions over time and within a year might change.

...

In C3S, either the results of the uncertainty or accuracy assessment as documented for the related dataset by the authors (in case external datasets are integrated) or measures Nr. 1 and 2 (see Table 17) for the provided datasets are used. In case a comparison to ICESat is sensible (depends on the region), also measures 3 and 4 will be applied to determine both uncertainty and accuracy. Where high quality DEMs are available (e.g. national or Arctic DEM), we will also use this L2 measure to determine product accuracy. An example for such a comparison is shown in Figure 7 for the tongue of Findelen Glacier in the Swiss Alps.

Opportunities to improve quality and fitness-for-purpose of the CDRs

Considering the above possibilities and constraints, we have identified a couple of regions that will be subject to data production in C3S. The most demanding issue is likely to be the improvement of the quality of the outlines in the RGI for South America, namely in the Andes of Argentina and Chile (RGI Region 17, southern Andes). In this region scenes with adverse snow conditions have been used for the RGI, resulting in glacier extents that are too large. Moreover, some lakes are not removed, debris cover is sometimes missing, and bare rock has been classified as glacier (Figure 8). Both countries have produced new glacier inventories, but it is unclear when they will be available in GLIMS.

...

Figure 9: The new estimate of specific (colour-coded) and cumulative glacier mass balances (numbers in Gt) over the 1961-2016 period by Zemp et al. (2019). Values are obtained from statistically combining the annual mass balance measurements in the field with the dekadal observations of the geodetic mass balance as derived from satellite data (e.g. from DEM differencing).

Scientific Research needs

To keep the work in C3S distinct from CCI, we will address scientific research needs (e.g. related to algorithm development, uncertainty characterisation, or key science questions) in CCI. However, change assessment in key regions will also be performed and published by C3S.

Opportunities from exploiting the Sentinels and any other relevant satellite

The most important innovation we expect in the near future is the application of a precise and freely available DEM for improved orthorectification of Sentinel-2 images in mountain topography by space agencies. Once this is achieved, we can use Sentinel-2A/B jointly, merge overlapping regions from adjacent paths without a shift in geolocation, combine Sentinel-2 with Landsat 8 data, generate drainage divides that are consistent with the image geometry, and calculate topographic information from elevation data that refers to current glacier extents. Until then, workload will be high and the dense time series of Sentinel-2 images cannot be used optimally.

...

Figure 10: Regional glacier hypsometry and observational coverage. For each of the 19 first-order regions, glacier hypsometry from RGI 6.0 (blue) is overlaid with glacier hypsometry of both the geodetic (grey) and the glaciological (black) samples. Values for the total number (N) and total area (S) of glaciers are given for each region, together with the relative coverage of both the glaciological and the geodetic samples. Elevations are given in m a.s.l. (metres above sea level).

Lake ECV Service

Introduction

Section 3.2 briefly presents the Lake ECV products provided in the service - lake surface water temperature (LSWT) and lake water level (LWL) as background to the remainder of the report.
Section 3.3 presents known statements of requirements directly relevant to the products in the context of the C3S, in terms of definitional, coverage, resolution, uncertainty, format and timeliness requirements. The C3S team's view and interpretation of these statements of requirement and their relevance to the C3S service is stated.
Section 3.4 presents an analysis of gaps and opportunities:

...

lake ECV components not presently delivered by the Hydrology service within the C3S 312b LHC service.

The Lake ECV products

Brokered and Generated CDR v1.0

The LSWT climate data record (CDR) brokered to the C3S is a daily gridded product derived from observations of one or more satellites (L3S, level-3 super-collated). The reported LSWT is an estimate of the daily mean surface temperature of the lake, wherever at least one valid observation has been made within the spatial grid cell on a given day. The grid is a regular latitude-longitude one at 0.05 degree intervals.
In addition to the cell-mean LSWT data, the product contains:

...

The CDR v1.0 contains scientifically consistent time series since the same physics-based algorithm has been employed for all the sensors so that the brokered dataset can be used seamlessly with the extended one.

Generated CDR v2.0

The generated LSWT v4.0 CDR v2.0 extends the CDR v1.0 time series to August 2019. The generated CDR v2.0 is identical in format and scientific methodology to the CDR v1.0 dataset. The CDR v2.0 starts from the day following the last in the CDR v1.0, is scientifically the same as the CDR, and is thus intended to be used seamlessly with it. The CDR v2.0 includes satellite data from AVHRR on MetOp-A and MetOp-B.

LWL V3.0: Brokered and Generated CDR

The LWL climate data record (CDR) brokered to the C3S is a timeseries product derived from observations of one or more satellites. The reported LWL is an estimate of the mean surface height of the lake, wherever at least three valid observations have been made within the intersect between the satellite ground track and a given lake.
In addition to the lake-mean LWL data, the timeseries contains:

...

The v3.0 CDR covers the period 1993 to 2018 under identical reprocessing, so there is no brokered/extended distinction in this case. The satellites contributing to the time series are: TOPEX/Poseidon, Jason-1/2/3 and Sentinel-3A.

Lakes Service: User requirements

There not having been a precursor ESA Climate Change Initiative project addressing the Lake ECVs, the is no substantive survey of user requirements for satellite-derived lake products. Presently, this section relies on statements for the Lake ECV from GCOS, published literature, experience from other CDR projects, and requirements emerging from the definition of the service. The requirements will be updated in future versions using requirements that emerge from users of the service and their feedback, and from any user requirements survey that is undertaken in a future CCI+ project.

LSWT (v4.0)

Definitional requirements

Coverage

Spatial and temporal resolution

Uncertainty requirements

Communication of uncertainty

Data uncertainties

Format requirements

Timeliness requirements

LWL (V3.0)

Definitional requirements

Coverage

Spatial and temporal resolution

Data uncertainties

Format requirements

Timeliness requirements

Lakes Service: Analysis of gaps and opportunities

Satellite observational constraints and opportunities

Lake surface water temperature

The LSWT observing system from space consists of ~1 km resolution infra-red imaging radiometers. In particular, the following sensors can be exploited for LSWT retrieval:

...

Lake water level

Table 19: LWL Gap Analysis Summary

Improvement of retrieval algorithms

Lake surface water temperature

LSWT estimation has three steps:

...

All R&D progress in the ESA Lake CCI will ultimately enter the C3S service via the CCI-generated brokered dataset, and validated transition of the updated research code to generate future annual C3S time series extensions.

Lake water level

The current state-of-the-art R&D that lead to the V3.0 CDR relies partly on a manual approach to estimate the geographic extraction zone of altimetry measurements. An automatic version of this R&D is currently being implemented in the frame of the present project to ramp-up the products and be able to provide water level for a wider network of lakes. New lakes should thus be proposed in the Test CDR V2.0. The method relies on a database of lake delineations and a land/water mask (from Global Surface Water Explorer, Pekel et al. 2016), intersected with the theoretical ground-track of the satellites.
Then, the extracted data must be corrected for various propagations (ionosphere, wet troposphere, dry troposphere…etc) and geophysical corrections (geoid, pole tide, solid earth tide…etc) based on models and with limitations. The geoid model, in particular, does not include small wavelengths of the geoid and this must be estimated based on altimetry data and a posteriori corrected. The algorithm is currently being improved to cover both simple (cf Figure 11, Error! Reference source not found.left panel) and complex (cf Figure 11, right panel) cases.

...

These two implementations are performed to improve the number of lakes monitored in the LWL product (see Section 3.4.1.2). Additionally, other R&D algorithms should be developed within the CCI-Lakes project and then be implemented in the C3S-312b-Lot4 products for operational use to improve the quality of the product.

Improvement of uncertainty estimation

Lake surface water temperature

L3C uncertainty: A comprehensive approach to estimate the LSWT uncertainty in L3 has been developed within the CCI SST work and it comprises the following components:

...

The different uncertainties are aggregated; in the products the total uncertainty is provided. The uncertainty can be validated, and the various components can be further refined (parameters better estimated and better validated) over time and understanding of the spatial and temporal scales of the error correlations over lakes can be improved. Alternative methods of representing the uncertainties (i.e. ensembles) can potentially also be considered.
L3S uncertainty: The per-lake inter-satellite bias correction generates an uncertainty which is included in the estimation of the L3S LSWT uncertainty.
The uncertainty estimate for LSWT is mature, and the ongoing evolution should focus on determining appropriate parameters to use for additional sensor data streams and updating such parameters for all sensors if reason to do so emerges.

Lake water level

The uncertainty variable distributed in the LWL product along the Water Level variable is currently estimated based on the Median Absolute Deviation of the consecutive along-track water level measurements before it is averaged. It estimates the precision of the measurements but not the accuracy part. The improvement of this uncertainty variable depends on the success of the CCI lakes project, but no strategy is currently foreseen to improve this variable.
The ongoing offline validation exercise will provide global statistics on the LWL product and a characterisation of the global uncertainty based on:

• Comparison to other altimetry products (e.g. G-REALM)
• Comparison to in situ data (e.g. HYDROLARE)

Lake ECV components not presently in the service

The GCOS definition (RD.1) of the Lake ECV includes, in addition to the LSWT and LWL, the elements of lake surface reflectance, lake area and lake ice cover and thickness. A review of the opportunity to broker datasets addressing these gap areas is scheduled for 2020.

Ice Sheets and Ice Shelves ECV Service

Introduction

This section aims at providing users with the relevant information on requirements, and gaps, for the four products provided by the Ice Sheets and Ice Shelves Service. It is divided into three sections. Section 4.2 describes the products currently provided by the Service. Section 4.3 provides the target requirements for the product. Section 4.4 provides a past, present, and future gap analysis for the product and covers both gaps in the data availability and scientific gaps that could be addressed by further research activities (outside C3S).

Ice Sheets Product description

The ice sheets and shelves service provides four products.

Ice velocity

The velocity grid represents the average annual ice surface velocity (IV) of Greenland in true metres per day. The geographic extent is the Greenland Ice Sheet, including peripheral glaciers. The ice sheet boundaries are based on the latest version of the Randolph Glacier Inventory (RGI 6.0, RGI Consortium, 2017) with updated glacier fronts for marine terminating glaciers. The basic IV product contains the horizontal (V_x, V_y) and vertical (V_z) components of the velocity vector.

...

Figure 12: Example IV product covering the Greenland Ice Sheet, depicted are from left to right the easting component, the northing component and the magnitude of velocity.

Instruments

The IV product is primarily derived by applying feature tracking on repeat pass Copernicus Sentinel-1 SLC data. Sentinel-1 is a C-band synthetic aperture radar (SAR) mission and the constellation currently comprises two identical satellites (Sentinel-1A and -1B) with a repeat cycle of 6/12-days. The Interferometric Wide (IW) swath mode is the standard operation mode over land surfaces including land ice. It applies the Terrain Observation by Progressive Scans (TOPS) acquisition technology, providing a spatial resolution of about 3 m and 22 m in slant range and azimuth, respectively, with a swath width of 250 km. Sentinel­1 is the main source for regular and comprehensive monitoring of land ice motion.

Algorithm name and version

The ENVEO software package (ESP v2.1) is a state-of-the-art IV retrieval algorithm designed for various SAR sensors (e.g. Sentinel-1, TerraSAR-X, ALOS PALSAR, Cosmo-SkyMed). The processor has been tested rigorously through intercomparisons with other packages and extensive validation efforts. The ESP-IV processing system runs on common Linux operating systems and has successfully been connected to cluster systems utilising several hundreds of cores. This is especially of interest for campaign processing of big data sets as for Greenland. The existing system for annual IV production for Greenland is fully operational. Further improvements of the software are planned and discussed in section 4.4.

Auxiliary data

Auxiliary data needed for input in the IV processor are a digital elevation model (DEM) and polygon shapefiles of the ice sheet boundary.

DEM

A DEM is needed for geometric co-registration of repeat pass SAR data and geocoding of the final products. This requires an accurate DEM without artefacts, as spurious jumps in the derived velocity fields can occur otherwise. For the IV maps produced in the Greenland Ice Sheet CCI, the Greenland Ice sheet Mapping Project (GIMP) DEM (Howat et al., 2014) was used. For C3S a new DEM was compiled and implemented based on the recently released 90 m TanDEM-X Global DEM (Rizzoli et al., 2017). Known issues relating to processing artefacts, outliers and gaps, are filled in using a gap interpolation method. The extent of the DEM is equal to the IV product.

Ice sheet boundary

The ice sheet and glacier boundaries are based on the latest version of the Randolph Glacier Inventory (RGI 6.0, RGI Consortium, 2017) with updated glacier fronts. The inventory has been compiled from more than 70 Landsat scenes (mostly acquired between 1999 and 2002) using semi-automated glacier mapping techniques (Rastner et al., 2012).

Antarctic surface elevation change

The product provides estimates of surface elevation change over the Antarctic ice sheets and ice shelves, over a long period, using level 2 radar altimeter data from five satellite missions: ERS-1, ERS-2, EnviSat, CryoSat-2 and Sentinel-3A. Its algorithms and processing scheme are based on previous work for the ESA Antarctic Ice Sheet Climate Change Initiative and are guided by the GCOS (Global Climate Observing System) targets for the Ice Sheets Land ECV (Essential Climate Variable).

...

Figure 13: Example Antarctic SEC product showing the rates of change derived for the period from 01-07-2007 to 01-07-2012. This merges data from EnviSat and CryoSat-2. In this case the data extends only as far south as the EnviSat southern orbit limit

Instruments

The instruments used are the ERS-1 RA, ERS-2 RA, EnviSat RA2, CryoSat-2 SIRAL and Sentinel-3A SRAL. The data products used are the ERS-1 and ERS-2 Reaper L2, the EnviSat L2 GDR_v2.1, the CryoSat-2 L2i LRM (Low Rate Mode) and SIN (Synthetic aperture radar INterferometer) and the Sentinel-3A L2 which is currently optimised for ocean studies.

Algorithm name and version

The software package has been assembled and tailored to the C3S requirements from previous work on IMBIE (the Ice sheet Mass Balance Intercomparison Exercise) and various ESA Climate Change Initiative projects. The second-year version is called C3S_Ant_Sec_ops_v2.0. The results have been tested against datasets from the previous projects mentioned and validated against the multi-year IceBridge airborne laser altimetry campaigns. The underlying processing system runs on common Linux operating systems.

Auxiliary data

Four auxiliary datasets are needed.

DEM

Radar altimetry over regions of very high slope is generally of poor quality due to confusion over echo provenance. The digital elevation model is used to remove data from areas extremely high slope, i.e. greater than 10°, from the input measurements. It is also used to provide a grid of flags ranking the slope angle in each cell. The model in use is the Slater et al. model based on CryoSat-2 data.

Ice extent

The processing area consists of all of the Antarctic ice sheets, ice shelves and associated ice rises and island. Its boundaries are based on the IceSAT MODIS (Moderate Resolution Imaging Spectroradiometer) 1km resolution mask, produced for the IMBIE2 project by Zwally et al.

Glacial isostatic adjustment

Movements of the surface related to glacial isostasy are corrected for using the Ivins et al. model IJ05.

Tides

Due to the poor resolution of the satellites' land masks in processing Antarctic coastal regions, it is necessary to remove the tides supplied in the L2 products and replace them with a consistent set. The replacements are generated using the Padman et al. CATS 2008a tide model.

Greenland surface elevation change

The Greenland surface elevation change closely follows the Antarctic SEC (see section 4.2.2). The main algorithms are based on previous work for the Greenland Ice Sheet CCI and are guided by the GCOS (Global Climate Observing System) targets for the Ice Sheets Land ECV. A full description of the processing approaches and algorithms are found in Sørensen et al. (2018) and Simonsen and Sørensen (2017, LSM5).

...

The C3S-SEC product is distributed in NetCDF4 format according to the C3S Common Data Model conventions, at 25 by 25 km grid resolution. The grid is an equal area grid as defined by the NSIDC North Polar Stereographic projection with latitude of true scale at 70°N and central meridian at 45°W (EPSG: 3413). This projection is the same as used for the Ice velocity product. In addition to the gridded solution of SEC, the following fields are also available: cartesian x-coordinate (x), cartesian y-coordinate (y), geographical longitude and latitude (lon, lat), grid area (accounting for projection errors), relative elevation change since 1992 (dh), start and end times for the altimeter data used (start_time, stop_time), distance from grid cell centre to observation location, and a number of different accuracy fields for the different parameters.

Instruments

The instruments used are the ERS-1 RA, ERS-2 RA, EnviSat RA2, CryoSat-2 SIRAL and Sentinel-3A. The data products used are the ERS-1 and ERS-2 Reaper L2, the EnviSat L2 GDR_v2.1, CryoSat-2 L2i LRM (Low Rate Mode) and SIN (Synthetic aperture radar INterferometer) and the Sentinel-3A L2 which is currently optimised for ocean studies.

Algorithm name and version

The software package has been assembled and tailored to the C3S requirements from previous work in ESA Climate Change Initiative projects and is evolved with an annual iteration, with the second year version being provided as Vers2.. The results have been validated against the multi-year NASA Operation IceBridge airborne laser altimetry campaigns, see section 4.4.3.3 and Simonsen et al (2017).

The underlying processing system runs on a common Linux operating system, and are divided in two, one for the older mission (ERS-1, ERS-2 and EnviSat) and one for Cryosat-2. For the older missions, the processing is brokered from the Greenland CCI and follow the proposed combination of cross-over and repeat-track algorithms for SEC as documented in Sørensen et al. (2018). This method has been independently validated and inter-compared with stat-of-the-art methods in Levinsen et al. (2015). A 5-year running mean window are used to derive an annual SEC solution. The final monthly solution provided for the C3S-product is derived by a temporal-weighted mean of all solutions covering a given month. For CryoSat-2, the least-square-model solution 5 of Simonsen and Sørensen (2017) has been tailored to the requirements of the C3S. The monthly solution is derived based on 1.5-years of CryoSat-2 data on either side of the month in question. This 3-year running-mean window is chosen for stability of the plan-fit solutions, and to limit the imprint of interannual weather variability in the SEC product and predict climatic signals.

Auxiliary data

The processing approach for the Greenland SEC are in less degree in need of auxiliary data. However, to provide consistent documentation, a full description of the same auxiliary data as in the Antarctic SEC is provided here. If not used, the reason for not considering them is provided.

DEM

The Greenland SEC applies the official level-2 data solutions provided by ESA. When this level-2 product is generated by ESA, a DEM is applied in the geolocation of LRM data. For more information refer to the mission specific documentation for the specific DEM used in the geolocation of the echo. No DEM are used for the combined cross-over and repeat-track solutions, however a DEM is used as an initial parameter for the LSM5-method applied for CryoSat-2. The resulting solution from LSM5 is an update to the DEM. In the CryoSat-2 processing the Greenland Ice sheet Mapping Project (GIMP) DEM version 1 is used (Howat, Negrete, and Smith 2017).  

Ice extent

In the original version the processing was done for all Greenlandic grid-cells with an ice-cover of more than 95%, as given by the PROMICE ice-cover product (Citterio and Ahlstrøm 2013). With the update to version 2, the processing is now done for all ice-covered grid-cells in accordance to the ESA glaciers CCI ice-cover product for the Greenland ice sheet and strongly connected peripheral glaciers and ice caps (Rastner et al. 2012, file version: glaciers_cci_gi_rgi05_TMETM_19942009_v170525.zip).

GIA

No glacial isostatic adjustment is applied to the dataset, due to the large discrepancy in the model GIA signal in Greenland, and the limited bias in the resulting SEC.

Tides

As the extent of floating ice shelves is limited in Greenland, no tidal adjustment is added to the product.

Gravimetric mass balance

The Gravimetric mass balance (GMB) relies solely on data from the Gravity Recovery and Climate Experiment (GRACE) mission. The mission consists of two twin satellites, which measure satellite-to-satellite distance. The gravity field of the Earth can then be derived from the change in the distance between the satellites. This precise evaluation of the gravity-field enables monthly solutions of Earth's gravity field anomalies from the launch in March 2002 to the end of its science mission in October 2017. The GRACE mass-con solution from both the Greenland and the Antarctic ice sheet CCI projects are brokered for the C3S-product and provided for the major ice sheet basins. See Barletta, Sørensen and Forsberg (2013) and Groth and Horwath (2016) for the description of the derivation of GMB from the initial level-2, c20, 1-degree GRACE-data. A GIA model and land ice mask are used as auxiliary data, along with the drainage basin definitions.

Ice Sheets User requirements

The overall requirements for all ice sheet and ice shelf service products are given in Table 20 below.

...

*The GCOS resolution target cannot be met with current satellite data, so the C3S project has set a 25km resolution target.
**It should be noted that there is a difference between volume and mass change of the ice sheet, which seems to be undefined in the GCOS implementation plan.

Ice velocity

The primary GCOS requirements for ice velocity are listed in Table 20. In addition, as part of Ice Sheets CCI Phase 1, user requirements were identified through an extensive user survey within the community. The User Requirements Document (URD) from the Ice Sheets CCI Phase 1 project contains a full description of the results from this survey (Hvidberg et al., 2012). The user requirements for ice velocity are summarised in Table 21.

...

Surface elevation change

The primary GCOS requirements for surface elevation change are listed in Table 20. In addition, as part of the Ice Sheets Greenland/Antarctica CCI Phase 1, user requirements were identified through an extensive user survey within the community. The User Requirements Document (URD) generated contains a full description of the results from this survey (Hvidberg et al., 2012, Shepherd et al., 2018) and its first requirement² matches the GCOS table (Table 20).

Gravimetric mass balance

The GCOS requirements regarding ice mass change do not adequately follow glaciological considerations, as there is a difference between ice sheet volume change (units: km³/year) and mass change (units: Gt/year). It has been assumed that the requirements should be given in water equivalent volumes, hence the conversion of 1-to-1 from volume to mass in Table 20.

Ice Sheets Gap analysis

Ice velocity

Description of past, current and future satellite coverage

The primary source dataset for the Greenland Ice Sheet (GIS) ice velocity product comprises Sentinel-1 (S1) single look complex (SLC) SAR data acquired in Interferometric Wide (IW) swath mode. One of the unique aspects of the S1 mission is the systematic acquisition planning of polar regions, designed to cover the entire GIS margin and large sections of the Antarctic coast continuously. The ongoing acquisition of ice sheet margins is augmented by dedicated ice sheet-wide campaigns for Greenland (annually) and Antarctica.

...

The constellation is currently the primary source for year-round monitoring of IV. In 2019, further expansion of the continuous coverage in Greenland commenced including the interior ice sheet. This provides the opportunity to produce Greenland wide velocity maps at sub-annual, and even monthly, intervals. The Sentinel-1 constellation will continue to operate well into the next decade with two more satellites (Sentinel-1C and -1D) already in development. This, in combination with other new and planned SAR missions (e.g. SAOCOM, NASA-ISRO NISAR), ensures the long-term sustainability of the CDR.

Development of processing algorithms

The existing system (ESP v2.1) at ENVEO for annual IV production for Greenland is fully operational. ESP is a state-of-the-art IV retrieval algorithm suited to accommodate the ongoing evolution of the Copernicus Sentinel-1 mission data. The primary processor will continue to be developed and updated to accommodate new sensors and requirements. Further technical development activities, ongoing and planned, are described in sections 4.4.1.4 and 4.4.1.5.

Methods for estimating uncertainties

The error prediction framework described in Mohr and Merryman-Boncori (2008) is applied to derive estimates of the error standard deviation of slant-range and azimuth velocity measurements. The input to the framework consists of the location of the GCPs used for velocity calibration, and in models for the covariance function (or equivalently the structure function) of all error sources, including noise and atmospheric propagation. For a mathematical formulation, the reader is referred to Mohr and Merryman-Boncori (2008).
In speckle tracking, where coherence is required, the noise component can be estimated from the correlation coefficient. For coherent offset tracking, the maximised coherence becomes equal to the interferometric coherence, and the following expression for the standard deviation, σC, of the shift estimate (in units of resolution elements) holds (DeZan, 2014):

...

Table 22: Summary of QA tests and the metrics that it provides.

QA-IV-1 Comparison of satellite derived velocity products with in situ measured velocity data (GPS). The quality metrics of this test provides: Mean and RMSD of the difference in velocity of IV products and in situ data.

...

QA-IV-5 Another internal QA method widely applied for quality assessment of velocity products is the analysis of stable ground where no velocity is expected. This gives a good overall indication for the bias introduced by the end-to-end velocity retrieval including co-registration of images, velocity retrieval, etc. After performing the matching for the entire region covered by the image pair, the results for the ice covered (moving) area will be separated from ice-free (stable) ground. The masking is done using a polygon of the glacier/rock outline. The quality metrics of this test provides Mean and RMSD of the velocity over stable terrain; mean values should be close to 0.

Opportunities to improve quality and fitness-for-purpose of the CDRs

Regarding ice velocity (IV), the current CDR constitutes an annually averaged Greenland Ice Sheet velocity map based on continuous processing of all acquired Sentinel-1 data (6- and 12-day repeats). These existing measurements can be further exploited to assemble and merge IV maps at higher temporal frequency and compile sub-annual (e.g. seasonal, monthly, weekly) velocity mosaics. This option becomes particularly interesting as the current acquisition plan for Sentinel-1 is extended to also cover the interior ice sheet continuously, permitting comprehensive monitoring of the full Greenland Ice Sheet.

...

Figure 14: Ice velocity map of Nioghalvfjerdsbrae/79Fjord-Glacier and Zachariae Isbræ from Sentinel-1 only (left) and merged product based on Sentinel-1 and Sentinel-2 (right).

Scientific research needs

As mentioned in section 4.4.1.5 a key research need is the development of Sentinel-1 TOPS mode InSAR to derive ice sheet velocity. InSAR is capable of providing high precision and high-resolution velocity over large areal extents and can significantly improve the accuracy of the ice velocity in slower moving areas. The retrieval of ice velocity from TOPS InSAR is, however, challenging and requires additional investigation, particularly for the removal of phase discontinuities and burst boundaries. These are caused by azimuth motion and different line of sight direction at the transitions of adjacent bursts. The phase jumps get more significant with increasing azimuth motion. Additional developments are needed that include taking the variation of the line of sight within bursts into account and requiring separation of azimuth and slant range components of velocity. Additionally, a strategy for performing burst wise phase unwrapping needs to be implemented.

Another research need required for improving the processing algorithm is reduction of the effects of differential ionospheric path delay and removal of ionospheric stripes. These stripes are clearly evident as streaks in the retrieved velocity (particularly over northern Greenland) that are aligned slightly oblique to the LOS direction. Ionospheric disturbances are one of the main sources of error in the IV maps and hinder applications. As the repeat cycle for S1 is short, the potential impact of ionosphere-induced noise on the velocity is high. A way to compensate the ionospheric effects is the implementation of the split-spectrum method in the processor, which permits separating the ionospheric and the non-dispersive phase terms.

Opportunities from exploiting the Sentinels and any other relevant satellite

As mentioned in section 4.1, further expansion of continuous acquisition coverage in Greenland of Sentinel-1 provides an opportunity to produce Greenland-wide velocity maps at high temporal resolution. Additionally, the increased temporal coverage in the interior could reduce the error in the annual maps and facilitate the removal of ionospheric stripes.

Antarctic surface elevation change

Description of past, current and future satellite coverage

The Antarctic SEC data initially came from four satellite missions, and one more has been added in the evolution to the v2 system.

...

Launched after CryoSat-2, the Sentinel-3 mission provides coverage to an EnviSat-like configuration.
In the data product there are no temporal data gaps. At each timestamp a varying pattern of grid cells contain no data. Estimation of the missing data values may be undertaken with care, considering the underlying geophysics of the Antarctic.

Development of processing algorithms

The original system, C3S_Ant_Sec_ops_v1.0, was used to make the initial data product. Its modular layout has allowed it to be upgraded to the recent v2 system C3S_Ant_Sec_ops_v2.0, with minimal alteration. Sentinel-3A data is now routinely ingested. The multi-mission cross-calibration algorithm was updated in 2019. Incremental improvements in the v1 algorithm were tested against a new method entirely. The new method was selected as it was found to retrieve ~18% more data than the improved v1 method, with very similar accuracy where both methods produced results. The system can be configured to process new releases of data (EnviSat v3 and CryoSat-2 Baseline D) when they become available.

Methods for estimating uncertainties

The uncertainty in each surface elevation change rate is calculated from three components summed in quadrature. These components are independent of each other and independent in all grid cells. They are:

...

Figure 18: Validation against Operation IceBridge of the v1 and v2 datasets. V1 results are given in black, and v2 where it corresponds with v1 in red.

Opportunities to improve quality and fitness-for-purpose of the CDRs

The input data used in the product comes from data streams that are constantly being upgraded and refined. The EnviSat GDRv3 dataset will replace the current v2.1 version in the product. When available, the Sentinel-3A land ice processor data products will fill the current gaps left where the orbital track transition from ocean to land is not handled properly. Incorporation of CryoSat-2 baseline D data will add coverage to the region around the south pole, and extra data density elsewhere.

Scientific research needs

In order to identify ice dynamic trends, the main emphasis for scientific research is in a long period of continuous acquisition. Progressive improvements in instrumentation allow for greater accuracy and areal coverage and thus a better focus on interesting regions at the sub-drainage-basin scale.

Opportunities from exploiting the Sentinels and any other relevant satellites

The Sentinel-3 mission will continue the data acquisition timestream. It will allow comparison with CryoSat-2 data that is geographically and temporally close, giving an opportunity to research cross-calibration algorithms in greater detail than usual. Its tandem phase will allow the exploration of the effect of small variations in instrument and orbit on the measurement data.

 Greenland Surface elevation Change

Description of past, current and future satellite coverage

The original release of the Greenland ice sheet surface elevation change data utilised four radar-altimeter satellite missions, and with the evolution to version 2 an additional satellite mission has been included (Sentinel-3A). The satellite coverage for the GrIS is the same as for Antarctica and is listed in section 4.4.2.1 and Table 23. However, as the north-pole is covered by ocean and not ice sheet, the coverage of the GrIS is more complete than for the Antarctic ice sheet, as only the northernmost part of the ice sheet is not covered by the orbit inclination of ERS-1, ERS-2, EnviSat and Sentinel-3A. Satellite radar altimetry is more challenging for the GrIS than the Antarctic ice sheet, as a larger proportion of the ice sheet is located in areas with complex topography. The traditional radar altimeters, with the large footprint size, are especially challenged. Here, the principle of observations only at the point-of-closest-approach results in biasing the observations to points at higher elevation. Hence, to retrieve surface elevation change an optimal combination of the crossover, a-long-track and plan-fitting methods are used for 5-year or 3-year data-windows advancing in steps of one month. To insure good spatial coverage, the individual methods are averaged at a larger grid (25km by 25km polar stereographic) than their native grid resolution by ordinary kriging. The resulting data product is gap free.

Development of processing algorithms

The original system, C3SMontly, had a modular layout in terms of missions. This allows for alterations throughout the processing chain. The foreseen addition of Sentinel-3A into the recent v2 system, C3SMontlyVers2, were done without any changes to the main structure of the operational code of the original version. However, a major update to the system was the addition of the ordinary-kriging module, which allows for surface elevation change predictions at all ice sheet grid-cells, and not only at low slope as in the original version. The system can be configured to process new releases of data (EnviSat v3 and CryoSat-2 Baseline D) when they become available.

Methods for estimating uncertainties

The uncertainty is given by the combination of the epoch uncertainty (derived from the supplied input data) and the model uncertainty. Figure 19 shows the distributions of the fitting stability and accuracy evaluated for all surface elevation estimates. The increased number of pixels just above the GCOS requirements in the version 2 fitting stability 2 is introduced by the shortening of the data-record used for CryoSat-2 and Sentinel-3, alongside the increased number of observations at coastal locations, where the uncertainty is larger due to the complex topography. This is mainly due to more weather variability introduced by the shortening of the averaging window, but is removed in the accuracy estimate by averaging data on sub-grid-cell level.

...

Figure 20: Difference in the rate of elevation change from OIB vs. the C3S Greenland Surface elevation Changeproduct version 1 and 2. As the OIB level 4 data consist of data from all repeats of older flight paths, the years in the figure refer to the first year of observations, e.g. 1993 includes data for all repeats of the 1993 flightpath until 2017. The upper-left panel shows the point-to-point agreement, alongside the one-to-one line. The lower-left panel shows the complete distribution for all years, which is averaged in the right panel to show the spatial distribution.

Opportunities to improve quality and fitness-for-purpose of the CDRs

The input data used in the SEC product comes from data-streams that are constantly being upgraded and refined. The EnviSat GDRv3 dataset will replace the current v2.1 version in the product. When available, the Sentinel-3A land ice processor data products will fill the current gaps left where the orbital track transition from ocean to land is not handled properly. When available, the inclusion of the CryoSat-2 baseline D promises to improve the data quality at the coastal regions with the updated slope model being applied in the product.

Scientific research needs

The scientific research needs for the SEC product over Greenland are the same as for the Antarctic surface elevation change product, section 4.4.2.5.

Opportunities for exploiting the Sentinels and any other relevant satellites

These are the same as for the Antarctic surface elevation change product, section 4.4.2.6.

Gravimetric mass balance

Description of past, current and future satellite coverage

The GRACE mission ended in October 2017, resulting in a data gap until data are released from the GRACE-Follow-On mission. GRACE-FO was launched on May 22, 2018 and is promising to continue the data record left by GRACE and once available and data has been validated GRACE FO data will be used to provide the next version of the Gravimetric Mass Balance (GMB) product to CDS..

Development of processing algorithms and methods for estimating uncertainties

The GRACE solution provided for the major drainage basins are brokered from the Greenland and the Antarctic ice sheet CCI projects. For both processing algorithms and uncertainty estimates refer to Barletta, Sørensen and Forsberg (2013), and Groth and Horwath (2016). The primary GCOS requirements for Gravimetric mass balance are met in terms of horizontal resolution (Table 20). If typical ice densities are assumed, the measurement uncertainties are at present about twice the requirement. This emphasises the outstanding scientific question of how to deal with the signal leakages between changing bodies of mass, such as individual drainage basins and peripheral glaciers and ice caps.

Opportunities to improve quality, fitness-for-purpose of the CDRs

In addition to understanding the signal leakage, a major opportunity lies with the R&D activity anticipated in the community in relation to the GRACE-FO mission. The new mission will be able to provide GMB for the ice sheets and continue the long time-series of GRACE. However, the merging/bridging of GRACE and GRACE-FO provide an outstanding scientific need.

References

Soil Moisture

BAUER-MARSCHALLINGER, B., FREEMAN, V., CAO, S., PAULIK, C., SCHAUFLER, S., T. STACHL, MODANESI, S., MASSARI, C., CIABATTA, L., BROCCA, L. & WAGNER, W. 2019. Toward Global Soil Moisture Monitoring With Sentinel-1: Harnessing Assets and Overcoming Obstacles. IEEE Transactions on Geoscience and Remote Sensing, 57, 520-539.

...

Zwally, H. Jay, Mario B. Giovinetto, Matthew A. Beckley, and Jack L. Saba, 2012, Antarctic and Greenland Drainage Systems, GSFC Cryospheric Sciences Laboratory, at httphttps://icesat4earth.gsfc.nasa.gov/cryo_/data/ant_grn_drainage_systems.phppolar-altimetry/antarctic-and-greenland-drainage-systems.

Related articles