Contributors: H. Konrad (DWD), G. Panegrossi (CNR ISAC), E. Cattani (CNR ISAC)
History of modifications
Related documents
Acronyms
General definitions
The meaning of the terms uncertainty, accuracy and error is often difficult to interpret and may be treated differently in various referred documents. In this document we adopt the following interpretation:
The accuracy, uncertainty or error of an estimated ECV is described by three differently contributing components:
- The systematic error
- The random error
- The time-dependent error
The systematic error is commonly the mean error or the Bias. For non-Gaussian distributions of the error the median or the mean absolute error can be a more useful quantity.
The random error is commonly the root-mean-squared deviation RMSD. Sometimes the Bias is subtracted yielding the centered root-mean-squared deviation cRMSD. Notice that if the Bias is zero the two mentioned quantities are equal and may be interpreted as the standard deviation of the error (often denoted standard error).
The time-dependent error is commonly the change in Bias over time (for ECVs over decades). We call this parameter stability.
More details on the estimation of these parameters are given in the Report on Updated KPIs (D4).
Scope of the document
This document provides relevant information on requirements and gaps for the Precipitation GPCP TCDR v1.0 and ICDR v1.x. It is divided into three parts. Part 1 describes the product the present document refers to. Part 2 provides the target requirements for the product. Part 3 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).
Executive summary
The precipitation product by the Global Precipitation Climatology Project (GPCP) is described together with its target requirement. The accuracy requirement is expressed as the absolute difference (mean absolute error) between spatially averaged GPCP fields and reference fields and is set to 0.3 mm/day. The targeted value for stability, i.e. the linear trend of the accuracy time series is 0.034 mm/day/decade.
The product is generated through merging of microwave and infrared precipitation estimates and outgoing longwave radiation from satellites, combined with rain gauge analyses from the Global Precipitation Climatology Centre (GPCC) over land. The temporal and spatial availability of all input data sources are described including a description of the evolution over time of surface measurements used by GPCC and of the satellite-based precipitation estimates. It is important to note that GPCP itself does not distinguish between a temporally fixed TCDR and a continuously updated ICDR. Instead, GPCP updates its long-term daily and monthly products continuously. The TCDR/ICDR distinction is only introduced in the scope of the brokering of the GPCP product to C3S, in order to harmonize its structure with other data products. Thus, the temporal coverage for the TCDR of monthly data is 1979-2017 while the corresponding ICDR covers 2018 onwards.
A gap in global coverage by polar-orbiting satellites with infrared information is expected to occur in 2020 which will thereafter be partially filled by data from the EUMETSAT Polar System Second Generation satellites.
It is emphasized that the dedicated effort within this C3S project for developing an algorithm for precipitation retrieval will not be based on the GPCP product. Instead, it will be a microwave-based TCDR exploiting MW imagers and sounders available at all latitudes in the time period 2000-2017. The use of both imagers and sounders will ensure high spatial coverage and temporal sampling of the precipitation. The new product will also utilize data from satellite-borne precipitation radars and improved methods for data merging and calibration.
1. Product description
1.1 Precipitation GPCP TCDR v1.0 + ICDR v1.x
The GPCP global precipitation products merge rainfall estimates from several microwave- and infrared satellite-borne sensors and the gridded rain gauge-based product by the Global Precipitation Climatology Centre (GPCC). It is created and maintained by the GPCP team at the University of Maryland (UMD), where the intellectual property rights remain with, and brokered to the Climate Data Store (CDS) by the Copernicus Climate Change Service (C3S). The GPCP TCDR v1.0 covers the period 01/1979-12/2017 (monthly data, v2.3), or 10/1996-12/2017 (daily data, v1.3), respectively. The GPCP ICDR v1.x starts in 01/2018 and at the time of writing extends until 09/2020 for both monthly and daily data.
It should be noted that, in the case of GPCP, the distinction between TCDR and ICDR at the cutoff date 2017/12/31 is only made in the scope of the brokering to C3S, as the initial data providers at UMD continuously produce new data as new satellite and other data become available (see PUGS [D3]). There is also an interim GPCP monthly product with a latency of only a few days to weeks compared to two to three months for the final GPCP product, but this interim product is replaced on the servers by the final one and, due to the envisaged latency in updating GPCP in the CDS of three months, is not included in the data deliveries to C3S. Moreover, the GPCP-Interim product is constructed in a similar manner to the final GPCP monthly analysis, except that for the input gauge analysis, the GPCC First Guess product is replaced by the GPCC Monitoring Product. For this reason, the Interim product would not be homogeneous to the provided GPCP TCDR. Finally, UMD interprets the GPCP-Interim as a provisional product and strongly discouraged from using it for months when the full GPCP is available.
Likewise, the version numbering for GPCP products in C3S (v1.0 / v1.x) is only introduced in the scope of the brokering. The actual dataset versions (v2.3 for monthly, v1.3 for daily data) are the relevant ones for outreach to users.
The spatial coverage is global, at a resolution of 2.5 degree (monthly data), or 1.0 degree (daily data), respectively. A list of input data is given in Table 1 in the PUGS [D3], based on Tables 1 and 2 in the C-ATBD [D1]. An overview is also given in this document, Section 3.1.1. The methods used for generating the dataset are described in the C-ATBD [D1] and Adler et al. (2003). Recent changes related to the upgrade to latest version (v2.3) are discussed by Adler et al. (2018).
2. User Requirements
2.1 Precipitation GPCP TCDR v1.0 + ICDR v1.x
2.1.1 Summary of target requirements (KPIs)
For the GPCP TCDR v1.0, the targeted value for accuracy, i.e. absolute difference (or mean absolute error according to General definitions) between spatially averaged GPCP fields and reference fields at each available time, is 0.3 mm/d or less. The targeted value for stability, i.e. the linear trend of the accuracy time series is 0.034 mm/d/dec or less.
For the GPCP ICDR v1.x, we adopt the strategy outlined in the Report on Updated KPIs [D4], i.e., the ICDR targets for accuracy are based on percentiles in the respective TCDR’s spatially averaged accuracy. With the monthly TMPA 3B43 product (TRMM, 2011) as reference dataset, the targeted upper and lower bounds for differences of spatial averages over low- to mid-latitudes (±50°) of the GPCP ICDR precipitation fields are 0.189 mm/d and -0.175 mm/d, respectively. For the daily GPCP ICDR (reference dataset: daily TMPA 3B42 product; Goddard Earth Sciences Data and Information Services Center, 2016), the bounds are 0.343 mm/d and -0.332 mm/d. The respective PQAR [D5] contains the details of the analysis of the percentiles of the TCDR performance against the TMPA products.
The TMPA products used as references for evaluating GPCP KPIs have been decommissioned and are only available until 12/2019. Consequently, for GPCP deliveries covering times after that date, we compare full global means (instead of latitudes between ±50°) with corresponding values in ERA5 (C3S, 2017). The targeted upper and lower bounds for the differences between GPCP monthly data and ERA5 are -0.338 mm/d and -0.063 mm/d. For the evaluation of daily data, these bounds are -0.535 mm/d and 0.017 mm/d. The PQAR [D5] contains the details of the analysis of the percentiles of the TCDR performance against ERA5.
Note that according to the Report on Updated KPIs [D4], the GPCP ICDR needs to be inside the above upper and lower bounds in 95% of all available time slices, which is verified through a binomial test.
2.1.2 Discussion of requirements with respect to GCOS and other requirements
GCOS formulates requirements for precipitation products only for monthly resolution (GCOS, 2016, table 23). The given accuracy target of 0.5 mm/h [sic] is less strict than the one applied here; the stability target is stricter than the one applied here (0.02 mm/decade [sic]).
2.1.3 Data format and content issues
The data are provided in one NetCDF file per month (monthly product) or per day (daily product). The files are compliant with the Climate and Forecast (CF) 1.6 convention and the Attribute Convention for Dataset Discovery (ACDD) 1.3. The products are both on a global equidistant latitude/longitude grid (Level 3), with a 2.5° (monthly), or a 1.0° (daily) spacing. The C-ATBDs [D1, D2] list all input and ancillary data at the respective epochs when they were used. Errors are only propagated and provided for the monthly product.
There are no updates for the previously delivered TCDR. However, the dataset itself and thus later deliveries of the related ICDR, are updated monthly with a latency of two to three months.
The GPCP monthly product is complete within its temporal and spatial scope. At some dates, there are missing values in a small number of grid cells in the case of the GPCP daily product. Details on missing values are provided in the Product Quality Assessment Report [D5].
3. Gap Analysis
3.1 Description of past, current and future satellite coverage
3.1.1 Precipitation GPCP TCDR v1.0 + ICDR v1.x
The GPCP TCDR v1.0 / ICDR v1.x monthly product is generated through merging of microwave and infrared precipitation estimates, and outgoing longwave radiation retrievals from satellites, combined with rain gauge analyses from the Global Precipitation Climatology Centre (GPCC) over
land.
The microwave estimates are mainly based on:
- Data from the Special Sensor Microwave/Imager (SSM/I) and its successor Special Sensor Microwave Imager/Sounder (SSMIS) on board the Defense Meteorological Satellite Program (DMSP) satellites. The SSMI and SSMIS instruments provide the key input data that are used for low latitudes (40°N-40°S) within GPCP over both ocean and land for the period 1987–present, with the beginning of the operational activities of SSMIS sensor in 2009 (Adler et al., 2018).
- The precipitation products from the TIROS Operational Vertical Sounder (TOVS, temporal coverage 1987-2003) and Atmospheric Infrared Sounder (AIRS)/Advanced Microwave Sounding Unit (AMSU) flying on NASA’s Aqua satellite (operational since 2003) are exploited for filling in SSM/I-SSMIS data voids at higher latitudes (polar and cold land regions) caused by the shortcomings in precipitation retrieval over frozen surfaces (Adler et al., 2003; 2018).
- For the whole duration of the GPCP dataset the geosynchronous IR-based estimates are obtained through the Geostationary Operational Environmental Satellite (GOES) precipitation Index (GPI; Arkin and Meisner, 1987) technique applied to GOES, METEOSAT, and Geostationary Meteorological Satellite (GMS)-Multifunction Transport Satellite (MTSAT)-Himawary-8 geostationary satellite series data.
- In case of unavailable geostationary IR data, the NOAA Advanced Very High Resolution Radiometer (AVHRR) observations are used.
- The Outgoing Longwave Radiation (OLR) Precipitation Index (OPI) monthly precipitation estimates based on IR data from all NOAA-series satellites are employed prior the SSM/I era to extend GPCP data back to 1979 (Adler et al., 2003; Xie and Arkin, 1998).
- GPCC data: a new set of rain gauge data was introduced in the GPCP monthly v2.3, i.e. the GPCC Full analysis v7 for the period 1979-2013 and the GPCC Monitoring analysis v5 for 2014 and beyond (Adler et al., 2018)
The GPCP CDR (including TCDR and ICDR) is characterized by changes in the satellite instrumentations (e.g. transitions from SSM/I to SSMIS, from TOVS to AIRS/AMSU, and the sequence of satellites in the various geostationary programs), that imply variations in the sensor spectral channels exploited in the retrieval algorithms and their calibrations, and could potentially introduce inhomogeneities into the data set. This issue is considered in GPCP by considering appropriate overlap periods between the sensors in order to adjust the estimates from the newer sensors to those of the older sensors with a longer period of use, and by continuously monitoring the CDR for the presence of anomalous behaviors. In the GPCP monthly v2.3 CDR, brokered for the CDS, a small negative anomaly in the mean global ocean precipitation leading to an incipient downward trend has been solved (Adler et al., 2018). The issue was connected to a discontinuity generated by the transition from the SSM/I to SSMIS and a TOVS sensor degradation, which likely compromised the precipitation estimates at the end of the TOVS period.
The EO data relevant to the generation of the GPCP One-Degree-Daily (1DD) product are the same already described for the monthly product although the retrieval algorithm is different from the one used for the GPCP monthly analysis (Huffman et al., 2001). The GPCP 1DD makes use of the IR
brightness temperatures from the geostationary constellation and AVHRR to extract the precipitation estimates in the range 40°N-40°S through the Threshold-Matched Precipitation Index (TMPI) technique. This approach is an adaptation of the GPI, where the SSM/I-SSMIS–based precipitation frequency (from GPROF algorithm) and the GPCC monthly precipitation estimate are exploited to locally constrain the TMPI IR brightness temperature threshold and conditional rain rate, respectively. Finally, TOVS-AIRS-AMSU data are used to produce the precipitation estimates outside the latitudinal band 40°N-40°S.
Due to the orbital drift of existing satellites, the approaching expected end of satellite lifetimes and the termination of the US Defense Meteorological Satellite Program, a gap in global coverage by polar-orbiting satellites is expected to occur in 2020. As of November 2020, all DMSP platforms are beyond life time and might experience channel losses or total failure any time. Planned satellite mission to partially fill this gap are the EUMETSAT Polar System Second Generation satellites.
3.2 Development of processing algorithms
3.2.1 Precipitation GPCP TCDR v1.0 + ICDR v1.x
The processing algorithm and its further development lie in the hands of the GPCP authors. In the scope of brokering the dataset to the Climate Data Store, we accept the algorithm in its current form. The change in instrumentation mentioned in Section 3.1.1 is a certainly a challenge when developing the algorithm, but issues are constantly identified and the algorithm improved if feasible (Section 3.1.1).
Precipitation is a notoriously difficult quantity to retrieve from satellite data, due, for example, to its intermittent nature in space and time and to the many phases in which it occurs. Many research groups are dedicated to improving existing and designing new algorithms, and upcoming CDRs will profit from this development, for example also in the case of the MW precipitation product developed directly for C3S, due to be completed in 2021 (see below for more details).
3.3 Methods for estimating uncertainties
3.3.1 Precipitation GPCP TCDR v1.0 + ICDR v1.x
There is no uncertainty estimate provided with the GPCP daily precipitation product. The GPCP monthly precipitation estimate includes a measure for pixel-wise precision, propagated from the raw observations and intermediate satellite products. An assessment of the provided uncertainty is included in the respective Product Quality Assessment Report [D5]. The uncertainties represent at least a one-sigma range when assessed against differences between GPCP and ERA-5 or the monthly TMPA product (TRMM, 2011; Goddard Earth Sciences Data and Information Services Center, 2016), see the PQAR [D5]).
3.4 Opportunities to improve quality and fitness-for-purpose of the CDRs
3.4.1 Precipitation GPCP TCDR v1.0 + ICDR v1.x
The processing algorithm and its further development lie in the hands of the GPCP authors. In the scope of brokering the dataset to the Climate Data Store, we accept the product in its current form. It should, however, be noted that the formulated GCOS requirement for spatial resolution (25 km,
corresponding to ~1/4° at the equator) is not met by the GPCP products (2.5° and 1.0°). Also, the daily product is not fulfilling the requirement on accuracy [D5].
However, there is a dedicated effort within C3S for developing an algorithm for precipitation retrieval, although not based on the GPCP product. The microwave-based (MW-based) TCDR will be based on the exploitation of MW imagers and sounders Level 1 FCDRs available at all latitudes between 2000 and 2017. As opposed to GPCP, where SSMI/SSMIS microwave frequencies < 90 GHz are used, the algorithm for the envisaged instruments, the Microwave Humidity Sounder (MHS) and its predecessor, the Advanced Microwave Sounding Unit B (AMSU-B), will be based on the exploitation of high-frequency channels (> 90 GHz), sensitive also to light precipitation at mid-high latitudes. The use of both imagers and sounders will ensure high spatial coverage and temporal sampling of the precipitation. MW imagers and sounders precipitation rate estimates will be merged and calibrated to provide global MW-based daily and monthly precipitation TCDRs on a 1°x1° grid.
DMSP satellites carrying SSM/I and SSMIS mostly have local overpassing times in the early morning. It is likely that the respective systematic sampling introduces a bias into the retrieved average daily (or even monthly) precipitation rates in areas where precipitation follows a diurnal pattern. The MW-product to be developed within C3S will circumvent these sampling preferences at least partially by including precipitation rates retrieved by MHS and AMSU-B. These are onboard NOAA and METOP satellites which, due to their orbit distribution, will introduce a more complete coverage of the day and thus improve the diurnal sampling. The originally envisaged temporal coverage for the newly developed product was 1999-2016. However, due to just one MW sounder platform available in 1999, with many observations compromised, this time span has been shifted to 2000-2017 for which the availability of high-quality MW sounder data is much better. This change has been communicated to and accepted by C3S.
3.5 Scientific Research needs
3.5.1 Precipitation GPCP TCDR v1.0 + ICDR v1.x
The processing algorithm and its further development lie in the hands of the GPCP authors. In the scope of brokering the dataset to the Climate Data Store, we accept the product in its current form. Research related to the improvement of the precipitation CDR will be carried out within the development of the MW-based precipitation TCDR within C3S. The quality of satellite daily and monthly MW-based precipitation is critically dependent on the number and on the quality of MW radiometers used to ensure maximum spatial and temporal coverage, see section 3.4.1. The MW-based precipitation TCDR developed within C3S, and complementing the GPCP product, will be based on the exploitation of Level 1 FCDRs available for SSM/I / SSMIS MW imagers and MHS/AMSU-B MW sounders available between 2000 and 2017. There are two areas of research that will be explored
during the development of MW-based TCDRs:
- MW-based precipitation development and quality assessment will benefit from the availability of unique cloud and precipitation observations by the two spaceborne radars currently available: the Dual-frequency Precipitation Radar (DPR) on board the NASA/JAXA Global Precipitation Measurement (GPM) Core Observatory, available since March 2014, and the NASA CloudSat Cloud Profiling Radar (CPR) available since 2006. The multi-year, quasi-global, and complementary DPR and CPR measurements offer a unique and extensive resource to analyze spaceborne MW radiometer precipitation observational capabilities. This is particularly useful in remote areas and/or where ground-based observations are sparse or not available (e.g., high latitudes), and in conditions and regimes where MW precipitation retrieval is more challenging (e.g., light precipitation and snowfall).
- Advanced merging and calibration procedure will be developed to account for inhomogeneity between MW imagers and sounders precipitation rate estimates related to differences in the processing algorithms, differences in sensor spectral channels, viewing geometry, spatial resolution. Error characterization will account for the outcome of the merging and calibration procedure and for the presence of anomalous behaviors in the TCDRs.
3.6 Opportunities from exploiting the Sentinels and any other relevant satellite
3.6.1 Precipitation GPCP TCDR v1.0 + ICDR v1.x
Authority over which satellite data are used for the GPCP products lies in the hands of the GPCP development team. They have adapted new sensors and missions when feasible or necessary in the past and will likely continue to do so.
There is no plan of using Sentinel data in the developments of the MW-based precipitation TCDRs. However, both development and quality assessment of MW-based TCDRs will benefit from the availability of unique cloud and precipitation observations by the two spaceborne radars currently available: the Dual-frequency Precipitation Radar (DPR) on board the NASA/JAXA Global Precipitation Measurement (GPM) Core Observatory, available since March 2014, and the NASA CloudSat Cloud Profiling Radar (CPR), available since 2006 (see Section 3.5.1).
References
Adler, R. F., M. R. P. Sapiano, G. J. Huffman, J.-J. Wang, G. Gu, D. T. Bolvin, L. Chiu, U. Schneider, A. Becker, E. Nelkin, P. Xie, R. Ferraro, and D.-B. Shin, 2018: The Global Precipitation Climatology Project (GPCP) Monthly analysis (new version 2.3) and a review of 2017 global precipitation. Atmos., 9, 138, doi:10.3390/atmos9040138.
Adler, R. F., G. J. Huffman, A. Chang, R. Ferrraro, J. Janoviak, B. Rudolf, U. Schneider, S. Curtis, D. T. Bolvin, A. Gruber, J. Susskind, P. A. Arkin, and E. Nelkin, 2003: The version 2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979-present). J. Hydrometeor., 4, 1147-1167.
Arkin, P. A., and B. N. Meisner, 1987: The relationship between large-scale convective rainfall and cold cloud over the Western Hemisphere during 1982-1984. Mon. Wea. Rev., 115, 51-74.
Copernicus Climate Change Service (C3S), 2017: ERA5: Fifth generation of ECMWF atmospheric reanalyses of the global climate. Copernicus Climate Change Service Climate Data Store (CDS), 21/12/2018. https://cds.climate.copernicus.eu/cdsapp#!/home.
GCOS (2016), THE GLOBAL OBSERVING SYSTEM FOR CLIMATE: IMPLEMENTATION NEEDS, GCOS GCOS-200, GOOS-214, https://library.wmo.int/doc_num.php?explnum_id=3417
Goddard Earth Sciences Data and Information Services Center (2016), TRMM (TMPA) Precipitation L3 1 day 0.25 degree x 0.25 degree V7, Edited by Andrey Savtchenko, , Goddard Earth Sciences Data and Information Services Center (GES DISC), doi: 10.5067/TRMM/TMPA/DAY/7
Huffman, G. J., R. F. Adler, P. A. Arkin, A. Chang, R. Ferraro, A. Gruber, J. Janoviak, A. McNab, B. Rudolf, and U. Schneider, 1997: The Global Precipitation Climatology Project (GPCP) combined precipitation dataset. Bull. Amer. Meteor. Soc., 78, 5-20.
Huffman, G. J., R. F. Adler, M. M. Morrissey, D. T. Bolvin, S. Curtis, R. Joyce, B. McGavock, and J. Susskind, 2001: Global precipitation at one-degree daily resolution from multisatellite observations. J. Hydrometeor., 2, 36-50.
Tropical Rainfall Measuring Mission (TRMM) (2011), TRMM (TMPA/3B43) Rainfall Estimate L3 1 month 0.25 degree x 0.25 degree V7, Greenbelt, MD, Goddard Earth Sciences Data and Information Services Center (GES DISC), doi: 10.5067/TRMM/TMPA/MONTH/7
Xie, P., and P. A. Arkin, 1998: Global monthly precipitation estimates from satellite-observed outgoing longwave radiation. J. Climate, 11, 137-164.
