Contributors: Leonardo Bagaglini (CNR-ISAC), Paolo Sanò (CNR-ISAC), Giulia Panegrossi (CNR-ISAC) Thomas Sikorski (DWD), Hannes Konrad (DWD), Marc Schröder (DWD)
History of modifications
List of datasets covered by this document
Related documents
Acronyms
Scope of the document
This document provides information on how to use the satellite-based estimates of precipitation by the Copernicus Microwave-based Global Precipitation (COBRA) datasets. The data products (daily and monthly) are first described in terms of their input data, and a brief overview of the computation methods used is also provided. Their target requirements and achieved performances are discussed. Geographical grid specifications, the data format, naming conventions, and the acknowledgement policy are provided.
Executive summary
A global precipitation climate data record (CDR) is a highly valuable product for a large variety of application areas, ranging from general research, climate monitoring, climate analysis and model evaluation, to agriculture and food security, disaster risk reduction, health, and water availability.
The need for the generation of an improved global precipitation product triggered a series of expert workshops dedicated to this topic. In these workshop series, it became clear that not all efforts can be covered by a single institute, nor funding scheme and that the (European) community needs to work together in order to reach that goal. Thus, the development, generation, and provision of a global precipitation CDR has to be pursued in a joint European effort. Reaching the goal of releasing a global precipitation CDR is seen as an evolutionary process with successively increasing quality, scientific and technical complexity, and improved technical specifications such as temporal coverage. With this in mind, the current MW precipitation product COBRA has been developed.
The COBRA climate data record (CDR) is a multi-platform composite of precipitation rates derived from microwave radiometers with global coverage. The data are available in a daily and a monthly temporal resolution and cover the entire years from 2000 until 2017, inclusively. These datasets are provided to the Climate Data Store (CDS) by the Copernicus Climate Change Service (C3S).
Compared to another satellite-based precipitation product, namely GPCP monthly v2.3, the spatial means of the COBRA monthly product achieve absolute deviations of less than 0.3 mm/d in 98.5% of all monthly instances. The respective comparison for the daily product with GPCP daily v1.3 shows that almost 87.5% of all daily spatial means are within 0.3 mm/d. The specified decadal stability criterion of 0.034 mm/d/dec in this comparison is achieved by both the monthly (0.034 mm/d/dec) and the daily (0.018 mm/d/dec) products. Against the same GPCP references, considering all grid cells on the globe, the random error of the monthly (daily) COBRA product is 1.561 mm/d (5.757 mm/d) and the systematic error is 0.140 mm/d (1.532 mm/d). Regional evaluation reveals a trend towards higher systematic error and towards lower random errors from low to high latitudes.
COBRA daily and monthly estimates of precipitation are available from the CDS in NetCDF format. The Copernicus licence applies.
Key Facts
Data Type:
- Precipitation Rates [mm/d]
Coverage:
- Global
- 2000–2017
Resolution:
- Daily / Monthly
- 1° × 1°
1 COBRA daily precipitation
1.1 Product description
The COBRA daily precipitation product provides global daily precipitation estimates (mm/d) from 2000 to 2017 on a regular rectangular geographical grid of 1° × 1° resolution. The product integrates satellite precipitation retrievals from two distinct sources: the PNPR-CLIM algorithm and the extended HOAPS v4 dataset (see ATBD [D1]) The PNPR-CLIM algorithm provides level 2 precipitation rate estimates obtained from the FIDUCEO Fundamental Climate Data Record (FCDR) (V4.1) of level 1 brightness temperatures for two cross-track scanning radiometers, MHS and AMSU-B, and model reanalyses. HOAPS provides level 2 precipitation rate estimates computed from level 1 brightness temperatures measured by four conical-scanning radiometers, SSM/I, SSMIS (CM SAF FCDR), AMSR-E and TMI. The data input details are listed table 1 in and the daily data availability over the entire period is shown in figure 1. Daily values are computed by averaging hourly gridded values across the satellite platforms and subsequent accumulation to daily precipitation.
Table 1: COBRA input data.
Name | Data type | Time period | Data source | Notes |
FIDUCEO FCDR v4.1 | Brightness temperature | 2000 – 2017 | FIDUCEO | Data from satellite (sensor) |
ERA5 | Daily: sea-ice cover, snow cover | 2000 – 2017 | ECMWF | |
Monthly: Total precipitable water, freezing level, 2 meter temperature. | ||||
HOAPS v4 | Rain rate | 2000-2017 | CM SAF / DWD | Data used to produce HOAPS v4 for CM SAF were extended: SSMIS platforms for the years 2015-2017 were included, as well as AMSR-E and TMI. Satellite (sensor):
|
Figure 1: Daily data usage per platform (colors) over the time period. Vertical black lines indicate gaps.
1.2 Target requirements
Table 2 summarizes the error components associated with the COBRA daily mean precipitation and daily GPCP (v1.3, Adler et al. 2017; Huffman et al. 2001), taken as reference. Systematic and unsystematic (random) root mean squared errors were computed according to Willmott (1981). Namely, the mean squared error (computed grid-cell by grid-cell) was decomposed into a linear part, obtained as the mean squared error between the linear regression of the reference towards the product and the reference itself, and the complementary non-linear part; the squared roots of these components are the systematic and unsystematic error respectively.
Table 2: Error characteristics of the COBRA daily mean precipitation v1.0.
| Dataset property | Global | |Lat|<30° | 30°≤|Lat|<60° | 60°≤|Lat| |
|---|---|---|---|---|
| Random error | 5.757 mm/d | 6.411 mm/d | 5.209 mm/d | 1.972 mm/d |
| Systematic error | 1.532 mm/d | 1.408 mm/d | 1.908 mm/d | 2.134 mm/d |
Table 3 provides an overview of achieved and required accuracy and stability properties of the COBRA dataset. The target requirements have been formulated in the context of monitoring the quality of precipitation products within C3S via Key Performance Indicators. They refer to the comparison of the dataset to a specific reference dataset on a (almost) global scale. Here, the achieved specifications are computed with respect to GPCP v1.3 (daily) CDR. Since there are known issues related to the NOAA15 platform, which is used before April 1st, 2001, the evaluated period has been limited to thereafter. We note here that the accuracy and stability requirements for daily and monthly data are the same. The higher temporal variability in daily data explains the 10% of daily instances violating the accuracy criterion.
Table 3: Accuracy and stability characteristics of the daily COBRA CDR with respect to the GPCP v1.3 CDR. Accuracy and stability calculations refer to the period April 1st, 2001 until December 31st, 2017, since NOAA15 satellite, which is used before April 1st, 2001, has known issues. For accuracy, the percentage of the daily instances that meet the target requirement is given.
Achieved | Target requirement | ||
Global | |Lat|<60° | ||
|---|---|---|---|
Accuracy | 87.5% achieved | 91.8% achieved | 0.3 mm/d |
Stability | 0.018 mm/d/dec | 0.033 mm/d/dec | 0.034 mm/d/dec |
1.3 Data usage information
The data format is NetCDF-4 and the filename convention is
COBRA_yyyy-mm-dd<SpatialResolution><TemporalResolution>v<Version>.nc .
The date yyyy-mm-dd is codified by a four digit year (yyyy), a two digit (zero-padded) month (mm) and a two-digit (zero-padded) day (dd). The short cut 1D for SpatialResolution denotes the latitude-longitude grid of 1° × 1° resolution. TemporalResolution is D (daily). Finally, <Version> is 1.0.
The content of the daily files is listed in table 4.
Table 4: COBRA daily precipitation v1.0 files content.
Variable Name | Dimension(s) | Unit | Description |
|---|---|---|---|
Coordinates | |||
lat | 1 | °N (degrees North) | Latitude of grid cell centre |
lat_bnds | 2 | °N (degrees North) | Boundaries of top (northern) and bottom (southern) grid cell edge |
lon | 1 | °E (degrees East) | Longitude of grid cell centre |
lon_bnds | 2 | °E (degrees East) | Boundaries of left (western) and right (eastern) grid cell edge |
time | 1 | Seconds since 1970-01-01 | Time stamp of the current day |
time_bnds | 2 | Seconds since 1970-01-01 | Boundaries of the time interval covered by time variable |
platform_id | 1 | N/A | An integer used for internal platform assignment |
instrument_id | 1 | N/A | An integer used for internal instrument assignment |
Data Variables | |||
precip | 3 (time, lat, lon) | mm/d | Accumulated daily precipitation rates that are represented by a single multi-platform composite |
precip_stdv | 3 (time, lat, lon) | mm/d | Daily mean of intra-platform standard deviation derived from hourly values |
Quality Variables | |||
quality_flag | 3 (time, lat, lon) | N/A | Mean of PNPR-CLIM quality flags, whose assigned data was used in composite creation. The quality flag accounts for several environmental features affecting the retrieval quality |
num_obs | 4 (time, lat, lon, instrument_id) | N/A | Total number of observations separated by instrument type |
num_covered_hours | 3 (time, lat, lon) | N/A | Accumulated number of hours, for which data of at least one platform is available on the respective day |
Ancillary Variables | |||
platform_name | 2 (time, platform_id) | N/A | Names of all platforms that are used for composite creation for the respective day. The names are allocated to the platform identifier. Platform names are saved as char array. Thus there is an additional dimension in the netCDF file describing the length of the longest string. |
instrument_name | 2 (time, instrument_id) | N/A | Assigned names of the specific instrument identifier. Instrument names are saved as char array. Thus there is an additional dimension in the netCDF file describing the length of the longest string. |
2 COBRA monthly precipitation
2.1 Product description
The COBRA monthly precipitation product provides global monthly mean precipitation estimates (mm/d) from 2000 to 2017 on a regular rectangular geographical grid of 1° × 1° resolution. The underlying precipitation rate estimates are the same as for the daily product (see section 1.1 and table 1). The monthly data are obtained by averaging PNPR-CLIM and HOAPS v4 instantaneous precipitation rate estimates in one month, first per platform, and then across platforms. In the latter step, the monthly precipitation seen by single platforms is weighted by the availability of the platform in the month.
2.2 Target requirements
Table 5 summarizes the error components associated with the COBRA monthly mean precipitation and monthly mean precipitation derived from daily GPCP data (v1.3, Adler et al. 2017; Huffman et al. 2001), taken as reference. Systematic and unsystematic (random) root mean squared errors were computed according to Willmott (1981), analogously to what done in section 1.2.
Table 5: Error characteristics of the COBRA monthly mean precipitation v1.0.
| Dataset property | Global | |Lat|<30° | 30°≤|Lat|<60° | 60°≤|Lat| |
|---|---|---|---|---|
| Random error | 1.561 mm/d | 1.721 mm/d | 1.321 mm/d | 0.841 mm/d |
| Systematic error | 0.140 mm/d | 0.266 mm/d | 0.461 mm/d | 0.946 mm/d |
Table 6 provides an overview over achieved and targeted accuracy- and stability-related properties of the dataset. The target requirements have been formulated in the context of monitoring the quality of precipitation products within C3S via Key Performance Indicators. They refer to the comparison of the dataset to a specific reference dataset on a (almost) global scale. Here, the achieved specifications are computed in comparison with GPCP v2.3 (monthly) CDR. Since there are known issues related to the NOAA15 platform, which is used before April 1st, 2001, the evaluated period has been limited to thereafter.
Table 6: Accuracy and Stability characteristics of the monthly COBRA CDR with respect to the GPCP v2.3 CDR. Accuracy and stability calculations refer to the period April 1st, 2001 until December 31st, 2017, since NOAA15 satellite, which is used before April 1st, 2001, has got known issues. For the accuracy the percentage of the monthly instances that meet the target requirement is given.
Achieved | Target requirement | ||
Global | |Lat|<60° | ||
|---|---|---|---|
Accuracy | 98.5% achieved | 100% achieved | 0.3 mm/d |
Stability | 0.034 mm/d/dec | 0.051 mm/d/dec | 0.034 mm/d/dec |
2.3 Data usage information
The data format is NetCDF-4 and the filename convention is
COBRA_yyyy-mm-dd<SpatialResolution><TemporalResolution>v<Version>.nc .
The date yyyy-mm-dd is codified by a four digit year (yyyy), a two digit (zero-padded) month (mm) and a two-digit (zero-padded) day (dd). The short cut 1D for SpatialResolution denotes the latitude-longitude grid of 1° × 1° resolution. TemporalResolution is M (monthly). Finally, <Version> is 1.0.
The content of the daily files is listed in table 7.
Table 7: COBRA monthly precipitation version v1.0 files content.
Variable Name | Dimension(s) | Unit | Description |
|---|---|---|---|
Coordinates | |||
lat | 1 | °N (degrees North) | Latitude of grid cell centre |
lat_bnds | 2 | °N (degrees North) | Boundaries of top (northern) and bottom (southern) grid cell edge |
lon | 1 | °E (degrees East) | Longitude of grid cell centre |
lon_bnds | 2 | °E (degrees East) | Boundaries of left (western) and right (eastern) grid cell edge |
time | 1 | Seconds since 1970-01-01 | Time stamp of the current month |
time_bnds | 2 | Seconds since 1970-01-01 | Boundaries of the time interval covered by time variable |
platform_id | 1 | N/A | An integer used for internal platform assignment |
instrument_id | 1 | N/A | An integer used for internal instrument assignment |
Data Variables | |||
precip | 3 (time, lat, lon) | mm/d | Monthly mean precipitation that is represented by a single multi-platform composite |
precip_stdv | 3 (time, lat, lon) | mm/d | Monthly mean of intra-platform standard deviation derived from hourly values |
Quality Variables | |||
quality_flag | 3 (time, lat, lon) | N/A | Mean of PNPR-CLIM quality flags, whose assigned data was used in composite creation. The quality flag accounts for several environmental features affecting the retrieval quality |
num_obs | 4 (time, lat, lon, instrument_id) | N/A | Total number of observations separated by instrument type |
num_covered_hours | 3 (time, lat, lon) | N/A | Accumulated number of hours, for which data of at least one platform is available in the respective month |
Ancillary Variables | |||
platform_name | 2 (time, platform_id) | N/A | Names of all platforms that are used for composite creation for the respective month. The names are allocated to the platform identifier. Platform names are saved as char array. Thus there is an additional dimension in the netCDF file describing the length of the longest string. |
instrument_name | 2 (time, instrument_id) | N/A | Assigned names of the specific instrument identifier. Instrument names are saved as char array. Thus there is an additional dimension in the netCDF file describing the length of the longest string. |
3 Data access information
The daily and monthly COBRA Climate data records (CDRs) are provided by the Copernicus Climate Change Service (C3S, https://climate.copernicus.eu/). Both CDRs can be downloaded from the Climate Data Store (CDS). The data can be found in the following catalogue entry of the CDS:
https://cds.climate.copernicus.eu/cdsapp#!/dataset/satellite-precipitation?tab=overview
The CDS provides detailed documentation and additional information on the dataset, too. These documents are available under the following link:
https://cds.climate.copernicus.eu/cdsapp#!/dataset/satellite-precipitation?tab=doc
3.1 Contact
Dataset related requests can be addressed to the Copernicus support team via the ECMWF Support Portal:
3.2 Licence
The COBRA precipitation products are published under the Copernicus Products Licence, which is available here:
https://apps.ecmwf.int/datasets/licences/copernicus
References
Adler, Robert, et al. (2017): Global Precipitation Climatology Project (GPCP) Climate Data Record (CDR), Version 1.3 (Daily). NOAA National Centers for Environmental Information. doi:10.7289/V5RX998Z.
Willmott, J.C. (1981): On the validation of models. Physical Geography, 2:2, 184-194, DOI: 10.1080/02723646.1981.10642213.
Huffman, G.J., et al. (2001): Global Precipitation at One-Degree Daily Resolution from Multi- Satellite Observations. J. Hydrometeor., 2(1), 36-50.
This document has been produced in the context of the Copernicus Climate Change Service (C3S).
The activities leading to these results have been contracted by the European Centre for Medium-Range Weather Forecasts, operator of C3S on behalf of the European Union (Delegation agreement signed on 11/11/2014). All information in this document is provided "as is" and no guarantee or warranty is given that the information is fit for any particular purpose.
The users thereof use the information at their sole risk and liability. For the avoidance of all doubt , the European Commission and the European Centre for Medium - Range Weather Forecasts have no liability in respect of this document, which is merely representing the author's view.
