Introduction
The Copernicus Arctic Regional Reanalysis (CARRA) is a very high resolution (2.5 km) reanalysis system for two domains of the European part of the Arctic. The Arctic is an essential area of the Globe having a significantly stronger warming than other parts of the Earth. The CARRA dataset will cover more than 30 years of reanalysis between September, 1990 to present (including the WMO reference climate period of 1991-2020). The dataset is updated at the end of every month adding 1 month new data to the dataset (with 2-3 months delay with respect to real time). The Arctic regional reanalysis system adds value to the ERA5 global reanalysis with the use of more (local) observations, with the better description of surface characteristic (like sea ice or glaciers for instance) and with its higher horizontal resolution. All the details of the CARRA system is described in its Full System Documentation.
Uncertainty information
Some static uncertainty information is provided for the Copernicus Arctic Regional Reanalysis (CARRA) dataset. There are two measures provided for uncertainties: one is called Upper Bound Error Standard Deviation and the other one Refined Error Standard Deviation. For the CARRA-East domain summer and winter values are provided, while only annual values for the CARRA-West domain. Details of the meaning of these values can be found in the CARRA Data User Guide. The table below indicates the available variables and levels. The data can be downloaded clicking on the domain names at each rows.
Variable mame | Levels | Domain |
Mean sea level pressure and surface pressure | Mean sea level and surface | CARRA-East and CARRA-West |
U-component wind velocity | Pressure levels: 50, 100, 150, 200, 250, 300, 400, 500, 600, 700, 800, 850, 900, 925, 950, 1000 hPa | CARRA-East and CARRA-West |
V-component wind velocity | Pressure levels: 50, 100, 150, 200, 250, 300, 400, 500, 600, 700, 800, 850, 900, 925, 950, 1000 hPa | CARRA-East and CARRA-West |
Temperature | Pressure levels: 50, 100, 150, 200, 250, 300, 400, 500, 600, 700, 800, 850, 900, 925, 950, 1000 hPa | CARRA-East and CARRA-West |
Geopotential | Pressure levels: 50, 100, 150, 200, 250, 300, 400, 500, 600, 700, 800, 850, 900, 925, 950, 1000 hPa | CARRA-East |
Relative humidity | Pressure levels: 50, 100, 150, 200, 250, 300, 400, 500, 600, 700, 800, 850, 900, 925, 950, 1000 hPa | CARRA-East |
| Specific humidity | Pressure levels: 50, 100, 150, 200, 250, 300, 400, 500, 600, 700, 800, 850, 900, 925, 950, 1000 hPa | CARRA-West |
Known issues
General data issues
- The physiography data including the divisions of sub-grid tiles into sea, nature, inland water and urban fractions is based on combined ECOCLIMAP-I and ECOCLIMAP-II data. However, ECOCLIMAP-II (Faroux et al. 2013, https://gmd.copernicus.org/articles/6/563/2013/ data is not defined for Iceland, Svalbard and the Russian Arctic islands. This means that there is a minor inconsistency in the static tile fraction fields in these regions with respect to all the other regions in CARRA. As an example this inconsistency might manifest in differences in urban areas outside the ECOCLIMAP-II region.
- Be careful when using the single level, pressure level and model level wind component variables - they are pointing to the local grid directions and not to the geographic east-west or north-south directions. Please note that this is not an issue, but a choice, which might not be trivial for the users. The variables 10m wind direction (in the single level entry) and the wind direction at selected height levels (in the height level entry) do not have this issue.
Small changes in the dataset after 1st July 2022 - after the move of the CARRA production from the HPC in Reading to that in Bologna (1 July 2022 in the produced data set), there are two unintended changes to the dataset:
The projection information in the output files changed. This impacts only the projection metadata in the output files; the physiographic and meteorological data are unaffected. Thus, the users experience a slight shift (typically corresponding to less than a km) in latitude/longitude (latlon) values provided for a grid point. The correct values are the ones used before 1st July 2022, so if you are comparing old and new data on a latlon point and require highest possible accuracy, you need to look at grid point values rather than latlon values to get a correct comparison and you should take the old latlon values as the correct ones. More precisely the change concerns the value of a parameter called “shapeOfTheEarth” which was set to 6 before, and has changed to 0 after. This corresponds to assuming a spherical Earth with Earth radius of 6 371 229.0 m before, and 6 367 470.0 m after. The shift of location for the grid points varies, and is typically less than a km.
The “typeOfLevel” entry for the parameter 228141 (Snow depth water equivalent) changed from “surface” to “heightAboveGround”.
Due to some technical problems some very small amount of data (less than 0.4% of the single level dataset) were missing in the original publication. These gaps were later covered with re-running the CARRA system and now they are available in the CDS. Please note that these re-runs were done on the new ECMWF ATOS computer platform and not in the originally used CRAY platform. Since the consistency between the CARRA system on CRAY and ATOS was demonstrated this does not cause any inconsistency in the dataset (keeping in mind the previous issue above).
- The surface latent heat flux computation is incorrectly related only to liquid water even if the surface temperature is below zero. This has an effect on the CARRA surface latent heat flux variable. The 2m relative humidity variable is related to saturation humidity over water as its standard definition.
- Albedo data: Albedo data is provided both at analysis and forecast times, but only the forecast time data are recommended for use. Due to rounding errors in the albedo calculation routine, unphysical albedo values (>100%) will sometimes occur. See Section 5.5 of the Data User Guide for further details.
Data availability issues
| Description |
|---|
Sensible and Latent heat fluxes are available only on time step +3 for the following periods: CARRA-East:
CARRA-West:
The parameter Sea-ice thickness is only being archived after July 2022. We are working on back-archiving this parameter for the rest of the dataset. The parameter Snow on ice total depth is missing for some isolated dates. These are: CARRA-East:
CARRA-West:
The following parameters are missing for various short, intermittent periods for single levels throughout the dataset. These will be back-archived as soon as possible:
|
Minor data assimilation issues (issues when some data was not properly assimilated)
Observational data is the backbone information source of a reanalysis system. However, during the course of reanalysis, for various technical reasons, some of the data streams got left out. In many cases this is only discovered after the reanalysis has been run. It can also occur that instances in which some radiance data, that should have been blacklisted, are used. Whenever it is possible and the impact is judged to be of significance, the affected periods have been rerun. However, reruns have not been performed for periods with only minor issues in the observational data streams. This has been decided largely based on experiences from verification and real time monitoring. If the impact has been judged to be relatively insignificant, and not to affect the overall consistence and integrity of CARRA reanalysis, reruns have not been performed.
For completeness, a list of the occurrences of missing observational data and affected periods are listed below.
| Affected period of the issue | Domain | Description | Cause |
|---|---|---|---|
July - October 2024 | CARRA-East CARRA-West
| A small issue was revealed in the ERA5T production for the period July - October 2024. The problem is linked to the assimilation of incorrect snow observations on the Alps. Details are available at ERA5: data documentation#Knownissues. The CARRA Timely Update system is using ERA5T data for lateral boundary conditions, however since the ERA5T impact is localised over the Alps and the (near surface) surface variables of ERA5T are not coupled directly to the CARRA system therefore there is no impact expected on the CARRA Timely Update production originating from this ERA5T issue. | Snow observation issue in ERA5T. More details are available at ERA5T issue in snow depth - July-November 2024 |
March - August 2024 Second half of August 2024 | CARRA-East CARRA-West | There is a dramatic reduction in the usage of aircraft data. A fix has been introduced taking effect from the production of September 2024 onwards. | This is due to a failure in Oulan (part of the observation handling in the model system) to decode the aircraft BUFR messages. There were two problems: one was due to a missing BUFR table file which affected AIREP data, the other was due to the Oulan source code being old and outdated and it started to pollute the station identifier string with nonsensical characters which caused Oulan to reject the AMDAR/EAMDAR reports. |
| Mid-October - December 2023 | CARRA-East | In the period mid-October to December 2023 Swedish surface observations had issues with slight deviations from the assumed observation time. This caused a decrease of the accuracy of the surface observations assimilated in the Swedish sub-area of the domain. Largest deviations are seen when a diurnal cycle in the observations are present, e.g. largest in October and reduced for November and December. The additional observation errors were checked to be quite small, so the resulting impact on the quality of the reanalysis in the area is very small. | Change in the Swedish surface observations archiving procedure |
| July 2021 - August 2023 | CARRA-East | Additional non-GTS snow observations from Norway (local obs) have not entered the CARRA system from the beginning of the timely update period July 2021 until Aug 2023. Snow observations over Norway from GTS observations have been used though.
| Issue with decoding |
| 25th April 2023 | CARRA-East CARRA-West | ASCAT data is missing for the 25th of April 2023. Affected cycles:
| Dataflow related issue |
| April 2023 | CARRA-East CARRA-West | Glacier albedo data for April 2023 is not available, so we have used data from October 2022 as an estimate. This is proven to have very little impact (based on some experimental evidence), since the sun is still relatively low on the horizon in April in the affected regions. | Delay of input data delivery |
| March 2023 | CARRA-East CARRA-West | RO data is missing for the entire month of March 2023. The impact is assumed to be small, thus no rerun is considered necessary.
| Dataflow related issue |
| February 2023 | CARRA-East CARRA-West | ASCAT data is missing for the 13th and 14th February 2023. Affected cycles:
| Dataflow related issue |
| January 2023 | CARRA-West | GEUS stations are missing for the first days of 2023, leading to somewhat fewer observation stations from Greenland. Affected dates:
| |
| December 2022 | CARRA-East CARRA-West | GNSS-RO data has been missing for the first few days of December 2022. Affected dates:
| Dataflow related issue |
| February 2021 - August 2021 | CARRA-East CARRA-West | There is a reduction of AMV observations assimilated of about 50%. Affected days:
| |
| January 2020 | CARRA-East | 20-50 local observations are missing for two days in 2020 on the CARRA-East domain. Affected dates:
| |
January 2020 (2020-01-05 to 2020-01-17) | CARRA-West | IASI data is not assimilated. | |
March - April 2019 (2019-03-01 to 2019-04-07) | CARRA-East CARRA-West | CryoClim data (satellite snow information) is not available. | |
January 2018 - 2021 (2018-01-01 to present) | CARRA-West | IMO stopped delivering observation data for road-side temperature with starting date Jan 2018 for lack of confidence on the data quality (around 100 stations in total). The impact of these data is estimated to be local. | |
| From 2018-01-14 onwards | CARRA-East CARRA-West | New buoy type assimilated leading to ~3 times more assimilated observations. | |
| December 2017 - January 2018 | CARRA-Ea st CARRA-Wast | Lack of microwave radiance data from METOP-A and B. There is a limited impact on upper air parameters, and we have data from NOAA satellites. Affected cycles:
| |
June - July 2017 (2017-06-15 to 2017-07-18) | CARRA-East CARRA-West | CryoClim data (satellite snow information) is not available. | |
| August 2016 | CARRA- East CARRA- West | Some cycles run without IASI assimilation. Affected cycles:
| Temporary issue fetching IASI data. |
| April 2015 | CARRA-E ast | Assimilation of AMSU-B is disabled for the following cycle: CARRA-East 2015-04-25 21Z | Corrupt file |
| 2015 - 2019 | CARRA-East CARRA-West | In the runs either TAC based data or WMO bufr based data have been used for BUOY and SHIP observation data. In a period there have - at least for BUOY data - been only partly overlap and some observations only available in one of the formats. Accordingly, both types should have been used in a period some time after 2014. A big jump in the number of used BUOY data were seen when changing from TAC based BUOY data, only, to WMO bufr based data, only. | |
| November 2014 - October 2022 | CARRA-East | The Andøya (WMO ID 01010) radiosonde has not been assimilated before October 2022. Affected period:
| Misconfigured blacklist file |
| September 2014 - January 2015 | CARRA-East CARRA-West | RO data are missing. The impact is limited. This affects the following periods:
| RO data are missing for some periods due to a misconfigured executable |
| February - March 2014 | CARRA- East | IASI data are missing for a period in 2014 on the CARRA-East domain. Affected dates: 2014-02-12 to 2014-03-12 | |
March 2010 | CARRA- West | Assimilation of AMV data is disabled for the following date: 2010-03-05 | Issues with MARS tape |
March 2009 | CARRA-East | CARRA-East cycle run without AMV assimilation for the following date: 2009-03-12 09Z | Corrupt file |
May 2007 | CARRA- East | Assimilation of AMV data is disabled for the following date: 2007-05-11 | Issues with MARS tape |
March 2002 | CARRA- West | Assimilation of AMV data is disabled for the following date: 2002-03-16 | Issues with MARS tape |
December 2001 - April 2005 (2001-12-03 00Z to 2005-04-26 21Z) | CARRA-East | Minor issue with channel 9 of NOAA-16 AMSU-A. This is not expected to have significant impact on the reanalysis result since the channel peaks relatively close to the model top. | Minor issue with channel 9 of NOAA-16 AMSU-A related to larger than expected analysis increment. |
February - December 2001 (2001-02-01 00Z to 2001-12-03 00Z) | CARRA-East | NOAA-16 AMSU-data was not assimilated. The impact of assimilation of this data type is limited, and we already have coverage from the NOAA-15 for this period. | NOAA-16 AMSU-data was not assimilated due to a misconfiguration. |
| 2000 - 2021 | CARRA-West
| In the MOD10A1 glacier albedo product there are some spatial gaps without data. One of these gaps is over the Storstrømmen glacier in Eastern Greenland, another is over the Petermann Glacier in Northwestern Greenland. In these gaps average albedo data as a function of altitude are used based on the available data. This has only a small impact on the overall atmospheric reanalysis but entails that the reanalysis dataset should be corrected with better glacier albedo data, if available, before being used for detailed analyses of the surface fluxes or boundary layers above these particular glaciers. | |
| 1998-05-03 to 1998-05-14 | CARRA- West | Assimilation of AMV data is disabled for the following period | Issues with MARS tape |
| 1997-2005 and 2006-2020 | CARRA-East | For the years 1997-2005 there could have been observations from 10-15 more Norwegian stations than usual. From 2006-2020 only a few stations were excluded. Based on our verification experiences for CARRA data, this is estimated to have only a relatively minor effect on the quality of the surface data over the local area in Norway, and only close to the stations that were left out. | Some local non-GTS surface observations on the Norwegian mainland were left out in the early years of the reanalysis in the CARRA-East domain. By mistake a filter that excludes observations that are no longer observing (as of 2020) was applied to local observations over Norway. Surface GTS observations have been used as normal and are not affected. |
| 1997-2000 | CARRA-East CARRA-West | Channel 4 in the MSU satellite data is not being used in this period. This channel is peaking at around 100 hPa. | Blacklist problem. |
| 1997 - 1999 | CARRA-East CARRA-West | For these three first years climatological average glacier albedo data are used based on data from the period 2000-2006. | The MOD10A1 glacier albedo product is only available since 2000. |
| 1994 - | CARRA-East CARRA-West | Volcanic aerosols are not included in production after 01. September 1994 |
