...
Issued by: KNMI/Else van den Besselaar
Issued Date: 06/12/2021
| Info | ||||
|---|---|---|---|---|
| ||||
|
...
| Easy Heading Macro | ||
|---|---|---|
|
History of modifications
| Expand | ||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||||||||||||||
|
Related documents
...
| title | Click here to see the related documents |
|---|
...
Data policy
...
...
ATBD
...
ECA&D Project Team (2021), European Climate Assessment & Dataset Algorithm Theoretical Basis Document (ATBD), Royal Netherlands Meteorological Institute KNMI, De Bilt, NL. version 10.9.
Acronyms
...
| title | Click here to the list of acronyms |
|---|
...
ATBD
...
Algorithm Theoretical Basis Document
...
ECA&D
...
European Climate Assessment & Dataset (https://www.ecad.eu
)
...
EEA
...
European Environment Agency
...
E-OBS
...
Gridded dataset based on the station time series from ECA&D
...
EUMETNET
...
grouping of European National Meteorological Services.
...
NMHS
...
National Meteorological and Hydrological Service
...
RA
...
Regional Association
...
WMO
...
World Meteorological Organization
About the dataset
Introduction
E-OBS is a land-only gridded daily observational dataset for precipitation, temperature, sea level pressure,global radiation, wind speed and relative humidity in Europe. This dataset is based on observations from meteorological stations across Europe which are provided by the National Meteorological and Hydrological Services (NMHSs) and other data holding institutes. The station data network is accessible through the webpages of the European Climate Assessment & Dataset (ECA&D, https://www.ecad.eu, Klein Tank et al. 2002).
The E-OBS dataset is provided at regular latitude-longitude grids with spatial resolutions of 0.1° and 0.25°, and has a daily resolution. The coverage of E-OBS spans much of the European continent, from northern Scandinavia to southern Spain and north Africa, and from Iceland into the Russian Federation at 40°E, but the coverage changes through time as the station coverage expands and decreases in time. The earliest maps for temperature, precipitation and sea-level pressure in E-OBSv20.0e start on 1 January 1950, while the maps for radiation start on 1 January 1980. E-OBSv20.0e runs until 31 July 2019 (although provisional monthly updates are provided through http://surfobs.climate.copernicus.eu/dataaccess/access_eobs_months.php). Full new versions of E-OBS are released twice a year. The latest version at the time of writing is v24.0e but in this document, various earlier versions are used as well to provide information. Except where mentioned explicitly, the difference between subsequent versions is the that the younger dataset is based on time series from stations which provide 6 months of additional data. In some cases, complete new stations are added.
ECA&D and E-OBS are the backbone for the Climate Data node of the Regional Climate Centre for WMO RA VI (Europe and the Middle East, https://www.ecad.eu/RCC-CD).E-OBS daily gridded observations for Europe from 1950 to present: Product user guide
Main variables
...
Variable
...
Unit
...
Description
...
Maximum temperature
...
°C
...
Daily maximum air temperature measured near the surface, usually at 2 meter height.
...
Mean temperature
...
°C
...
Daily mean air temperature measured near the surface, usually at 2 meter height.
...
Minimum temperature
...
°C
...
Daily minimum air temperature measured near the surface, usually at 2 meter height.
...
Precipitation amount
...
mm
...
Total daily amount of rain, snow and hail measured as the height of the equivalent liquid water in a square meter. The data sources for the precipitation are rain gauge data which do not have a uniform way of defining the 24-hour period over which precipitation measurements are made. Therefore, there is no uniform time period (for instance, 06 UTC previous day to 06 UTC today) which could be attached to the daily precipitation.
...
Sea level pressure
...
hPa
...
Daily mean air pressure at sea level. In regions where the Earth's surface is above sea level the surface pressure is used to compute the air pressure that would exist at sea level directly below given a constant air temperature from the surface to the sea level point.
...
Surface shortwave downwelling radiation
...
W/m2
...
The flux of shortwave radiation (also known as solar radiation) measured at the Earth's surface.
...
Relative Humidity
...
%
...
Daily mean relative humidity measured near the surface usually at a height of 2 meters. Relative humidity values relate to actual humidity and saturation humidity. Values are in the interval \[0,100\]. 0% means that the air in the grid cell is totally dry whereas 100% indicates that the air in the cell is saturated with water vapour.
...
Wind speed
...
m/s
...
Daily mean wind speed at 10 meter height.
The underlying station dataset
The station data are provided by 79 participants and the ECA&D dataset contains over 20000 meteorological stations (status May 2021). Metadata of the series, including the source and metadata of the meteorological stations are provided through the ECA&D website.
For a considerable number of countries the number of stations used is the complete national network and therefore much more dense than the station network that is routinely shared among NMHSs which is the basis of other gridded datasets. Figure 1 shows the map with stations that are used in the E-OBS dataset. This figure shows the high density of stations in many parts of Europe, but it also shows the inhomogeneity in the coverage of the stations, with generally a less dense coverage in the circum-Mediterranean and eastern Europe. The density of stations gradually increases through collaborations with NMHSs within European research contracts. Figure 1 also shows the steep increase in the number of stations in the 1950s, levelling-off in the early 1960s for temperature and reaching a maximum in the 1980s-1990s for precipitation. The decline in the most recent part of the record mostly relates to slow and infrequent updates of data by the NMHSs that do not provide updates on a regular basis.
For the series that are not updated by the NMHSs an alternative dataset is used. Synoptic data is routinely exchanged between NMHSs through the Global Telecommunication System and these data are used as an alternative for a maximum of 10 years when validated data directly from the NMHS is missing. The impact of using synoptic data on the quality of the dataset and the ability to reliably monitor variations in climate is addressed by van den Besselaar et al. (2012).
Figure 1: Map with the station coverage in ECA&D which is the basis for the E-OBS precipitation dataset v20.0e (left). The number of stations which provide precipitation (red) and temperature (green, blue, purple) vs. time (right).
User community and user feed-back
Initially (Haylock et al. 2008), this gridded dataset was developed to provide a validation dataset for the suite of Europe-wide climate model simulations produced as part of the EU ENSEMBLES project. While E-OBS remains an important data set for model validation, it is also used more generally for monitoring the climate across Europe, particularly with regard to the assessment of the magnitude and frequency of daily extremes.
Figure 2: Cumulative number of registered and confirmed E-OBS users (left, status 28 August 2019) The number of new E-OBS users per month since mid-2009 (right).
...
Prior to the inclusion of E-OBS in the Copernicus Data Store, users were encouraged to register as an E-OBS user (although not mandatory to be able to download the dataset). By registering as an E-OBS user, they are notified when a new version is available and in case a problem has been found with the dataset. This registration system is used to track the number of registered and confirmed users. Figure 2 shows the cumulative number of users since the start of this system (early 2009) and the number of new users per month. Since this registration is not mandatory, it is expected that there are more users of the E-OBS dataset than visualized here. Figure 3 shows a pie chart of the user groups of ECA&D and E-OBS based on a survey from 2013, which shows that the use of E-OBS has spread beyond the climate community.
The position of E-OBS is unique in Europe because of the relatively high spatial resolution, the daily resolution of the dataset, the provision of precipitation, temperature, sea level pressure, global radiation, wind speed and relative humidity, and the length of the dataset. All variables provided in E-OBS start in 1950, except for wind speed which starts in 1980. Finally, the station data on which E-OBS is based are available through the ECA&D webpages (provided the owner of the data has given permission to share these data). In these respects E-OBS contrasts with other datasets.
E-OBS enjoys widespread use in the scientific community. Figure 4 shows the cumulative citations to the articles which documents the station dataset and the articles which document the earlier versions of E-OBS and the current ensemble version. The status at the end of 2021 was that over 3100 studies are found referring to E-OBS.
Figure 4: The number of cumulative citations to Klein Tank et al. (2002), which documents the ECA&D stations dataset, the citations to Haylock et al. (2008) which documents the first versions of the E-OBS dataset, and to Cornes et al. (2018) which documents the latest ensemble version of the E-OBS dataset (status end of 2021). Source: Google Scholar.
The E-OBS dataset is used in the EEA report "Climate change, impacts and vulnerability in Europe 2016. An indicator-based report" to quantify climate change over Europe. The EEA advices the European Commission on climate change issues.
About 3 to 4 email discussions with users per month on the E-OBS dataset reach the producing staff. These are answered usually within a few working days.
The user interaction has led to improvement in the E-OBS gridded dataset in the past. Three examples of improvement are given below. In E-OBSv4.0 (released March 2011) a user from the Regional Climate Modelling community noticed that the correlation between the E-OBS daily rainfall estimate and his simulations correlated well, except over Poland (Figure 5). This problem turned out to be related to the absence of relevant metadata of the underlying station data. Accumulated 24-hour rainfall amounts, usually measured in the morning, can be coupled to the start of the measuring interval or to the end of the measuring interval. There is no standardization in relating the date to this measurement practice across Europe. ECA&D staff lacked access to the metadata which indicate that rainfall estimates from Poland were related to the end of the measuring interval. A time shift of one day in the underlying station data for Poland alleviated this problem.
Figure 5: Correlation between daily rainfall of E-OBSv4.0 and results from a Regional Climate Model. The left panel shows relatively high correlations across most of Europe, with the exception of Poland. In the right panel the E-OBS dataset is shifted one day back and the correlation shows the opposite pattern. Problem noticed by a colleague from Météo-France.
...
Figure 6: Precipitation on 21 January 1995 in E-OBSv4.0 (left) and 22 January 1995 (middle). After a day-shift of the Luxembourg precipitation series, a more realistic precipitation pattern was obtained (right). Problem noticed by a colleague from KNMI.
An example where the gridding method of E-OBS was improved following feed-back from a user relates to the number of drizzle-days in E-OBSv8.0 (released April 2013). Drizzle-days are days with precipitation totals between 0.1 and 0.5mm/day. Figure 7 shows a distinct raster-like pattern in the number of drizzle days in E-OBSv8.0 on the rotated 0.22° grid. Following this feed-back, a modification was made to the method which removed the raster-pattern from this metric. Note that these rotated grids are no longer available for E-OBSv18.0e and higher.
Figure 7: The number of drizzle-days in E-OBSv8.0 before (left) and after (right) the modification in the gridding method. Problem noticed by a colleague from SMHI.
Methodology and the uncertainty estimation
E‐OBS is calculated by adopting a two‐stage process to produce the daily fields: (1) the daily values are aggregated to monthly values that are initially fitted with a deterministic model, to capture the long‐range spatial trend in the data, and (2) the residuals from this model are then interpolated using a stochastic technique (Gaussian Random Field, GRF) to produce the daily ensemble. Monthly values are used in the first step of the interpolation, since the relationship of altitude to the meteorological fields can be difficult to discern in daily resolution data, particularly for precipitation.
In the latest versions of E-OBS (from v18.0e onward), uncertainty is estimated using stochastic simulation to produce an ensemble of realizations of each daily field. A set of spatially correlated GRFs are generated for each day that are conditional on the residuals from the deterministic spatial trend model described above. The spatial structure of the random fields is defined through the calculation of an empirical variogram. An example of the uncertainty estimate is shown in Figure 8 for a day in a wet spell for central Europe. It shows, as an illustration, the first four ensemble members and a plot of the station locations, the ensemble-mean precipitation and the uncertainty in this estimate in the 24-hour accumulated precipitation. Further details of the method, comparisons against the older version of E-OBS (using the previous gridding method) and comparisons against national datasets are documented by Cornes et al. (2018). Below is a section providing guidance on the use of an ensemble dataset.
...
|
Related documents
| Expand | ||||
|---|---|---|---|---|
| ||||
|
Acronyms
| Expand | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||||||||
|
About the dataset
Introduction
E-OBS is a land-only gridded daily observational dataset for precipitation, temperature, sea level pressure, global radiation, wind speed and relative humidity in Europe. This dataset is based on observations from meteorological stations across Europe which are provided by the National Meteorological and Hydrological Services (NMHSs) and other data holding institutes. The station data network is accessible through the webpages of the European Climate Assessment & Dataset (ECA&D, www.ecad.eu, Klein Tank et al. 2002).
The E-OBS dataset is provided on regular latitude-longitude grids with spatial resolutions of 0.1° and 0.25°, and has a daily resolution. The coverage of E-OBS spans much of the European continent, from northern Scandinavia to southern Spain and north Africa, and from Iceland into the Russian Federation at 40°E, but the coverage changes through time as the station coverage expands and decreases in time. The earliest maps for temperature, precipitation, sea level pressure and radiation in E-OBS start on 1 January 1950, while the maps for wind speed start on 1 January 1980. Full new versions of E-OBS are released twice a year and provisional monthly updates are provided through http://surfobs.climate.copernicus.eu/dataaccess/access_eobs_months.php. The latest version at the time of writing is v29.0e, but in this document various earlier versions are used to provide additional information. Except where mentioned explicitly, the difference between subsequent versions is that each version is based on time series from stations which provide 6 months of additional data compared to the previous version. In some cases, new station time series have been added.
ECA&D and E-OBS are the backbone for the Climate Data node of the Regional Climate Centre for WMO RA VI (Europe and the Middle East).
Main variables
Variable | Unit | Description |
Maximum temperature | °C | Daily maximum air temperature measured near the surface, usually at 2 metres above the surface. |
Mean temperature | °C | Daily mean air temperature measured near the surface, usually at 2 metres above the surface. |
Minimum temperature | °C | Daily minimum air temperature measured near the surface, usually at 2 metres above the surface. |
Precipitation amount | mm | Total daily amount of rain, snow and hail measured as the height of the equivalent liquid water in a square metre. The data sources for the precipitation are rain gauge data which do not have a uniform way of defining the 24-hour period over which precipitation measurements are made. Therefore, there is no uniform time period (for instance, 06 UTC previous day to 06 UTC today) which could be attached to the daily precipitation. |
Sea level pressure | hPa | Daily mean air pressure at sea level. In regions where the Earth's surface is above sea level, the surface pressure is used to compute the air pressure that would exist at sea level directly below, given a constant air temperature from the surface to the sea level point. |
Surface shortwave downwelling radiation | W/m2 | Daily mean flux of shortwave radiation (also known as solar radiation) measured at the Earth's surface. |
Relative Humidity | % | Daily mean relative humidity measured near the surface usually at a height of 2 metres. Relative humidity values relate to actual humidity and saturation humidity. Values are in the interval [0,100]. 0% means that the air in the grid cell is totally dry, whereas 100% indicates that the air in the cell is saturated with water vapour, defined with respect to saturation over water. |
Wind speed | m/s | Daily mean wind speed at 10 metres above the surface. |
Data access information
| Description | Link |
|---|---|
| E-OBS is available to users via the Climate Data Store | https://cds.climate.copernicus.eu/cdsapp#!/dataset/insitu-gridded-observations-europe%20?tab=overview |
| E-OBS is also available through the C3S2_311_Lot3 portal | https://surfobs.climate.copernicus.eu/dataaccess/access_eobs.php |
The underlying station dataset
The station data are provided by 85 participating institutions and the ECA&D dataset contains over 23600 meteorological stations (status March 2024). Metadata of the time series, including the source and information about the meteorological stations are provided through the ECA&D website.
For a considerable number of countries, the number of stations used is the complete national network and therefore much more dense than the station network that is routinely shared among NMHSs, which is the basis of other gridded datasets. Figure 1a shows a map with all available stations that are used for precipitation in the E-OBSv20.0e dataset, regardless of their start or stopdates. This figure shows the high density of stations in many parts of Europe, but it also shows the inhomogeneity of the station coverage, with generally a lower density of coverage in the circum-Mediterranean and eastern Europe. The density of stations gradually increases with time (Figure 1b), through collaborations with NMHSs within European research contracts. Figure 1b also shows a steep increase in the number of stations in the 1950s, which levels-off in the early 1960s for temperature and reaches a maximum in the 1980s-1990s for precipitation. The decline in the most recent part of the record mostly relates to slow and infrequent updates of data by the NMHSs that do not provide updates on a regular basis.
For the time series that are not updated by the NMHSs, an alternative dataset is used. Synoptic data is routinely exchanged between NMHSs through the Global Telecommunication System (GTS) and these data are used as an alternative to extend these time series to the present for a maximum of 10 years when validated data from the NMHS is missing. The impact of using synoptic data on the quality of the dataset and the ability to reliably monitor variations in climate is addressed by van den Besselaar et al. (2012).
Figure 1: Map with the station coverage in ECA&D which is the basis for the E-OBS precipitation dataset v20.0e (a). The number of stations which provide precipitation (red) and temperature (green, blue, purple) as a function of time (b).
User community and user feed-back
Initially (Haylock et al. 2008), this gridded dataset was developed to provide a validation dataset for the suite of Europe-wide climate model simulations produced as part of the EU ENSEMBLES project. While E-OBS remains an important data set for model validation, it is also used more generally for monitoring the climate across Europe, particularly with regard to the assessment of the magnitude and frequency of daily extremes.
Figure 2: Cumulative number of registered and confirmed E-OBS users (a, status 28 August 2019). The number of new E-OBS users per month from mid-2009 (b).
Figure 3: Pie chart of the user groups of ECA&D and E-OBS based on a survey from 2013.
Prior to the inclusion of E-OBS in the C3S Climate Data Store (CDS), a (voluntary) user registration system was in place which allowed the number of registered users to be tracked. Figure 2 shows the cumulative number of registered users from the start of this system (early 2009) and the number of new registered users per month. Figure 3 shows a pie chart of the user groups of ECA&D and E-OBS based on a survey from 2013, which shows that the use of E-OBS has spread beyond the climate community.
The position of E-OBS is unique in Europe because of the relatively high spatial resolution, the daily resolution of the dataset, the provision of precipitation, temperature, sea level pressure, global radiation, wind speed and relative humidity, and the length of the dataset. All variables provided in E-OBS start in 1950, except for wind speed which starts in 1980. Finally, the station data on which E-OBS is based are available through the ECA&D webpages (provided the owner of the data has given permission to share these data). In these respects E-OBS contrasts with other datasets.
E-OBS enjoys widespread use in the scientific community. Figure 4 shows the cumulative citations to the articles which document the station dataset and the articles which document the earlier versions of E-OBS and the current ensemble version. The status at the end of 2021 was that over 3100 studies had been found to refer to E-OBS.
Figure 4: The number of cumulative citations to Klein Tank et al. (2002), which documents the ECA&D stations dataset, the citations to Haylock et al. (2008), which documents the first versions of the E-OBS dataset, and to Cornes et al. (2018), which documents the latest ensemble version of the E-OBS dataset (status end of 2021). Source: Google Scholar.
The E-OBS dataset is used in the EEA report "Climate change, impacts and vulnerability in Europe 2016. An indicator-based report" to quantify climate change over Europe. The EEA advises the European Commission on climate change issues.
Each month, about 3 to 4 email discussions on the E-OBS dataset take place between users and the provider. Usually, these queries are answered within a few working days.
The user interaction has led to improvement in the E-OBS gridded dataset in the past. Three examples of improvement are given below. In E-OBSv4.0 (released March 2011), a user from the Regional Climate Modelling community noticed that the correlation between the E-OBS daily rainfall estimate and his simulations correlated well, except over Poland (Figure 5). This problem turned out to be related to the absence of relevant metadata of the underlying station data. Accumulated 24-hour rainfall amounts, usually measured in the morning, can be tagged with the start of the measuring interval or to the end of the measuring interval. There is no standardization in relating the date to this measurement practice across Europe. ECA&D staff lacked access to the metadata which indicate that rainfall estimates from Poland were related to the end of the measuring interval. A time shift of one day in the underlying station data for Poland alleviated this problem.
Figure 5: Correlation between daily rainfall of E-OBSv4.0 and results from a Regional Climate Model. Panel (a) shows relatively high correlations across most of Europe, with the exception of Poland. In panel (b), the E-OBS dataset is shifted one day back and the correlation shows the opposite pattern. Problem noticed by a colleague from Météo-France.
A similar problem was reported for precipitation over Luxembourg. Figure 6 shows a rain shower moving from France into Germany, which seems to arrive in Luxembourg one day earlier than in reality. Following these problems, the ECA&D staff queried all NMHSs in Europe which provided data and completed metadata for those stations where sufficiently detailed metadata was missing. For some cases it was found that 24-hour precipitation totals were assigned to the end of the measuring interval, which is usually in the early morning. This meant that the largest part of the 24-hour accumulation period coincided with the day before that in the date stamp of the measurement. The contrast in the date stamp of the measurement and the date on which (most of) the rainfall occurred, motivated ECA&D staff to shift the rainfall time series by one day (prior to the gridding in E-OBS).
Figure 6: Precipitation [mm/day] on 21 January 1995 in E-OBSv4.0 (a) and 22 January 1995 (b). After a one day-shift of the Luxembourg precipitation series, a more realistic precipitation pattern was obtained (c). Problem noticed by a colleague from KNMI.
An example where the gridding method of E-OBS was improved following feed-back from a user relates to the number of drizzle-days in E-OBSv8.0 (released April 2013). Drizzle-days are days with precipitation totals between 0.1 and 0.5mm/day. Figure 7 shows a distinct raster-like pattern in the number of drizzle days in E-OBSv8.0 on the rotated 0.22° grid. Following this feed-back, a modification was made to the method, which removed the raster-pattern from this metric. Note that these rotated grids are no longer available for E-OBSv18.0e and higher.
Figure 7: The number of drizzle-days in E-OBSv8.0 before (a) and after (b) the modification of the gridding method. Gray scales relate to the number of drizzle days, with darker colours indicating higher number of drizzle days. Problem noticed by a colleague from SMHI.
Methodology and the uncertainty estimation
E‐OBS is calculated by adopting a two‐stage process to produce the daily fields: (1) the daily values are aggregated to monthly values that are initially fitted with a deterministic model, to capture the long‐range spatial trend in the data, and (2) the residuals from this model are then interpolated using a stochastic technique (Gaussian Random Field, GRF) to produce the daily ensemble. Monthly values are used in the first step of the interpolation, since the relationship of altitude to the meteorological fields can be difficult to discern in daily data, particularly for precipitation.
In the latest versions of E-OBS (from v18.0e onward), uncertainty is estimated using stochastic simulation to produce an ensemble of realizations of each daily field. A set of spatially correlated GRFs are generated for each day, that are conditional on the residuals from the deterministic spatial trend model described above. The spatial structure of the random fields is defined through the calculation of an empirical variogram. An example of the uncertainty estimate is shown in Figure 8 for a day in a wet spell for central Europe. It shows, as an illustration, the first four ensemble members and a plot of the station locations, the ensemble-mean precipitation and the uncertainty in this estimate of the 24-hour accumulated precipitation. Further details of the method, comparisons against an older version of E-OBS (using the previous gridding method) and comparisons against national datasets are documented by Cornes et al. (2018). Below is a section providing guidance on the use of an ensemble dataset.
The mean across the ensemble members is calculated and is provided as the "best-guess" ensemble mean field. The spread is calculated as the difference between the 5th and 95th percentiles over the ensemble to provide a measure of the 90% uncertainty range. Up to version 23.1e, a 100-member ensemble is used for the E-OBS datasets, except for the global radiation dataset which initially started with a 10-member ensemble. From version 24.0e onwards, all E-OBS datasets have their uncertainty captured in a 20-member ensemble. The motivaton to reduce the ensemble size to 20 is based on an analysis of the propagation of the uncertainty into a standard set of climate indices. The analysis showed that the propagetd uncertainty in E-OBS saturated at 20 members already. This analysis is reported on in the Product User Guide for the E-OBS-based indices.
Figure 8: Example of the estimate of 24-hour precipitation [mm/day] for 1 June 2013. This is one day of a wet spell that caused severe flooding in central Europe. Shown are the first four ensemble members of E-OBSv16.0e for precipitation (a) and the station locations (black dots), the ensemble-mean precipitation estimate (coloured shading) and isolines of the ensemble spread, which relate to the uncertainty in the precipitation estimate (b). This example shows that the largest uncertainty coincides with the area with a relatively low station density and a high precipitation amount.
The E-OBS precipitation dataset is daily, meaning the observations cover 24 hours per time step. The exact 24-hour period can be different per region and element (variable). The reason for this is that some data providers measure from midnight to midnight while others might measure from morning to morning. Since E-OBS is an observational dataset, no attempts have been made to adjust the time series for these offsets. It is made sure, where known, that the largest part of the measured 24-hour period corresponds to the day attached to the time step in E-OBS (and ECA&D).
The global radiation dataset not only uses ground-based in-situ observations, but, in order to properly estimate spatial variability, it also incorporates the CERES global radiation satellite derived product and includes elevation as a spatial predictor. On daily time-scales the global radiation is mainly dependent on cloud patterns. This makes daily global radiation spatially and temporally highly variable. The in-situ observations are interpolated using Multiple Adaptive Regression Splines (MARS). The dimension reduced satellite data is used to interpolate between the in-situ observations, where the in-situ observations are the 'anchor points' (van der Schrier et al. 2021).
Relative humidity (hu) is interpolated in the same way as the temperature fields. The input station data is transformed to √(100-hu), with hu in percent, which exhibits a more normal distribution. These transformed values are interpolated and afterwards changed back to relative humidity itself.
The dataset for wind strength uses an approach inspired by Cornes et al. (2018). For this dataset, we apply greedy forward selection linear regression (FSLR) to include the high-resolution effects of the auxiliary variables on monthly-mean data. These data provide a 'background' for the daily estimates. We apply cross-validation to avoid FSLR over-fitting and use full-cycle bootstrapping to create FSLR ensemble members. Then, we apply Gaussian process regression (GPR) to regress the daily anomalies. We consider the effect of the spatial distribution of station locations on the GPR gridding uncertainty. As this dataset contains several decades of daily gridded wind fields, computational efficiency is of the utmost importance. We alleviate the computational cost of the FSLR and GPR analyses by incorporating greedy algorithms and sparse matrix algebra in the analyses.
The gridded wind fields are calculated as a statistical ensemble of realizations. In the present analysis, the ensemble spread is based on uncertainties arising from the auxiliary variables as well as from the spatial distribution of stations.
Cross-validation is used to tune the GPR hyper parameters. Where conventional GPR hyperparameter tuning aims at an optimal prediction of the gridded mean, instead, we tune the GPR hyperparameters for optimal prediction of the gridded ensemble spread.
The Global 30 Arc-Second Elevation Data Set (GTOPO30), a global raster Digital Elevation Model (DEM) with a horizontal grid spacing of 30 arc seconds (approximately 1 kilometre) developed by the United States Geological Survey (USGS), is used for the elevation file.
Earlier versions of E-OBS (up to v17.0) used the method described by Haylock et al. (2008) and Van den Besselaar et al. (2011) for the interpolation of temperature, precipitation and sea-level pressure.
Additional post-processing for the relative humidity and global radiation datasets
An additional quality control has been performed on the relative humidity fields where unphysically low values of relative humidity have been observed as a consequence of low station density in combination with the gridding method. As of E-OBSv24.0e, the grid boxes with values below 5% are set to missing in each of the 20 ensemble members. This additional check mirrors the quality check applied to the input stations. The ensemble mean and ensemble spread are derived from these adjusted ensemble members, making sure that all humidity values are equal to or above 5%. Note, that the approach to grid transformed values of humidity (rather than the humidity observations directly) already makes sure that no humidity values higher than 100% are possible.
Similarly, the global radiation dataset has been post-processed in the areas and in the periods where it suffers from unrealistic values. This problem occurs in areas and periods where the station network is of low density. Following a similar procedure to that for relative humidity, the same quality-control rules are applied to the grids as are applied to the stations. This means that any daily global radiation sum in each of the E-OBS ensemble members is set to missing where it falls below the 3% TOA (Top Of Atmosphere) level or exceeds the maximum expected global radiation at the earth surface on a clear sky day. The quality control procedure is detailed in ECA&D (2021). This additional post-processing is in place as of E-OBSv24.0e.
Known issues
There are a couple of known issues with the E-OBS dataset or in certain versions and/or for certain elements. These issues are given on the following page which is updated when needed: https://surfobs.climate.copernicus.eu/userguidance/known_issues_eobs.php
Comparison with other datasets
E-OBS is used routinely in monitoring the European climate and contributes to the monthly and annual State-of-the-Climate bulletins issued through C3S, and to the State-of-the-Climate bulletins published annually in The Bulletin of the American Meteorological Society (Europe section).
The calculation of the European averaged temperature is documented by Van der Schrier et al. (2013). In this study, a quantitative analysis of the sources of uncertainty of the European average temperature indicates that the uncertainties due to urbanization, statistical interpolation, and the potential inhomogeneities in the input records to E‐OBS dominate the total uncertainty estimate for the European-averaged temperature.
The uncertainty estimate in the European averaged temperature is used to make a comparison between E-OBSv6.0 and global observational datasets and global reanalyses. Although E-OBSv6.0 differs considerably with the latest versions of E-OBS (station density has increased, the method to calculate the gridded temperature estimate and the uncertainty has changed, and the spatial resolution has increased), the temperature averaged over the European continent and averaged over the year will not be dramatically affected as the averaging in time and space makes the European averaged temperature independent of the improvements in the fine-scale structure. The increase in station density means that the uncertainty estimate of the statistical interpolation has decreased, making the uncertainty estimates for the early versions of E-OBS somewhat pessimistic for the latest versions. The comparisons between E-OBS and global observational and reanalyses datasets are given in Figure 9. A comparison between E-OBSv7.0 and ERA-Interim, focused specifically on indices of the numbers of days above the 90th percentiles of daily maximum temperature, the number of days below the 10th percentiles of daily minimum temperature and trends in daily surface maximum/minimum temperature extremes across Europe, is documented by Cornes and Jones (2013).
Figure 9: Time series of annual, European averaged temperature anomalies [°C] from 1950 to 2011, with respect to the 1961-1990 climatology, as calculated for E-OBSv6.0 compared to global observational datasets (a) and global reanalysis (b). The lower plots give the E-OBS uncertainty margins in grey boxes and the difference between E-OBS and the global datasets (Van der Schrier et al. 2013).
A comparison with the observational CRUTS3.1 dataset, which has a spatial resolution more comparable to that of E-OBS, is given in Figure 10.
Figure 10: Time series of European averaged annual temperatures [°C] from 1950 to 2011 as calculated from E-OBSv6.0 (in black), including the uncertainty margins in grey boxes, and the European averaged annual temperature as calculated using the CRUTS3.1 temperature dataset (in red).
Guidance on the use of an ensemble dataset
Ensemble datasets are climate datasets that consist of a number of equally probable realizations, and relate to data in gridded format. The ensemble aims to give a measure of uncertainty in the data field; such datasets are widely used in a number of areas of climate change science, like perturbed physics simulations of climate models or numerical weather prediction models. The term "ensemble" is also used when referring to multiple realizations from gridded observational datasets. Such datasets, like E-OBS, are formed from the interpolation of station values. Although the aim is the same as ensembles calculated for model simulations – to quantify uncertainty in the data – the generation and hence interpretation of the realizations is quite different. When constructing gridded observational datasets there are a number of key parameters that depend on well-informed decisions and do not flow from e.g. first principles. These key parameters will affect the final gridded field as they directly affect the output from the gridding algorithm. Four of the more important of these key parameters include:
- Search radius for inclusion of stations influencing a grid box
- Estimates of the impacts of homogeneity issues on the quality of the input station data
- Impact of a number of possible co-variates used in the gridding (e.g. latitude, longitude, elevation and distance from the coast or an inland water body)
- Including station data where the base period (e.g. 1961-1990 or 1981-2010) is less precisely known
The usual approach in the production of a gridded dataset is to determine the best or most likely values for the key parameters used to interpolate station values to values on a regular grid. With an ensemble version of the dataset, a number of these key parameters can be varied, producing a range of possible gridded datasets (referred to as the ensemble). The advantage of producing a dataset this way is that with an ensemble it can be easier to quantify the uncertainty in the grid box averages. Using standard statistical approaches this can be difficult as a number of the error components have spatial and temporal structures which are difficult to model. Where users require a single measure of the interpolated daily fields, then the "best guess" (ensemble mean) values should be used. However, the ensemble spread should always be consulted as the uncertainty of the gridded field varies across the domain, and is ultimately determined by the variations in station coverage. The individual ensemble members are mainly intended for users who require the uncertainty in the gridded fields to propagate through to various other applications. If a user requires rainfall data for hydrological modelling, then each of the ensemble members could be fed into the hydrological model. In this way, the uncertainty in the rainfall interpolation would propagate through to the hydrological model output. Subsequent versions of E-OBS are updated using recent data from the European national meteorological services and by the inclusion of new time series. Usually, the amount of input station data explains the differences between subsequent versions of E-OBS.
E-OBSv29.0e vs E-OBSv28.0e
E-OBSv29.0e was released in March 2024 and spans the period 1950-01-01 to 2023-12-31, while E-OBSv28.0e was released in October 2023 spanning 1950-01-01 to 2023-06-30. The most important changes between these two versions is the amount of data that is used. Existing networks of synoptic, climatological and (manual) rain-gauge station data are updated with the latest measurements that are received directly from the national or regional meteorological services. The group of meteorological services that provide such frequent updates of these networks are those from Germany, Czech Republic, Bosnia and Herzegovina, Norway, Slovenia, Finland, Ireland, Estonia, Sweden, Luxembourg, Netherlands, Portugal, Spain, Switzerland, Italy (Emilia-Romagna), Montenegro, Belgium, France, Denmark, UK, Hungary and Catalonia (Spain). For the remaining countries, data from the network of synoptic stations is used to update the time series (as documented in section 2.3). The other main changes for E-OBSv29.0e are:
- Included new stations and updates for Ukraine, Portugal and Belgium
- Included data from Global Summary of the Day for southeast Europe
- Updated Polish precipitation series that were wrongly included
- Included radiation series for Trentino in Italy
Minimum temperature
Figure 11 shows the difference in the annual mean climatology of daily mean temperature for the period 1991-2020 between E-OBS versions 29.0e and 28.0e. Figure 12 shows the seasonal mean differences between these two versions for the period Dec 1999 – Nov 2000.
Figure 11: Difference in annual mean climatology of daily minimum temperature [°C] for the period 1991-2020 between versions 29.0e and 28.0e.
Figure 12: Differences in seasonal means of daily minimum temperature [°C] for Dec 1999-Nov 2000 between versions 29.0e and 28.0e.
Mean temperature
Figure 13 shows the difference in the annual mean climatology of daily mean temperature for the period 1991-2020 between E-OBS versions 29.0e and 28.0e. Figure 14 shows the seasonal mean differences between these two versions for the period Dec 1999 – Nov 2000.
Figure 13: Difference in annual mean climatology of daily mean temperature [°C] for the period 1991-2020 between versions 29.0e and 28.0e.
Figure 14: Differences in seasonal means of daily mean temperature [°C] for Dec 1999-Nov 2000 between versions 29.0e and 28.0e.
Maximum temperature
Figure 15 shows the difference in the annual mean climatology of daily mean temperature for the period 1991-2020 between E-OBS versions 29.0e and 28.0e. Figure 16 shows the seasonal mean differences between these two versions for the period Dec 1999 – Nov 2000.
Figure 15: Difference in annual mean climatology of daily mean temperature [°C] for the period 1991-2020 between versions 29.0e and 28.0e.
Figure 16: Differences in seasonal means of daily mean temperature [°C] for Dec 1999-Nov 2000 between versions 29.0e and 28.0e.
Precipitation
Figure 17 shows the difference in the annual mean climatology of daily precipitation for the period 1991-2020 between E-OBS versions 29.0e and 28.0e. Figure 18 shows the seasonal mean differences between these two versions for the period Dec 1999 – Nov 2000.
Figure 17: Difference in annual mean climatology of daily precipitation [mm/day] for the period 1991-2020 between versions 29.0e and 28.0e.
Figure 18: Differences in seasonal means of daily precipitation [mm/day] for Dec 1999-Nov 2000 between versions 29.0e and 28.0e.
Sea level pressure
Figure 19 shows the difference in the annual mean climatology of daily mean sea level pressure for the period 1991-2020 between E-OBS versions 29.0e and 28.0e. Figure 20 shows the seasonal mean differences between these two versions for the period Dec 1999 – Nov 2000.
Figure 19: Difference in annual mean climatology of daily mean sea level pressure [hPa] for the period 1991-2020 between versions 29.0e and 28.0e.
Figure 20: Differences in seasonal means of daily mean sea level pressure [hPa] for Dec 1999-Nov 2000 between versions 29.0e and 28.0e.
Radiation
Figure 21 shows the difference in the annual mean climatology of daily mean global radiation for the period 1991-2020 between E-OBS versions 29.0e and 28.0e. Figure 22 shows the seasonal mean differences between these two versions for the period Dec 1999 – Nov 2000.
Figure 21: Difference in annual mean climatology of daily mean global radiation [W/m2] for the period 1991-2020 between versions 29.0e and 28.0e.
Figure 22: Differences in seasonal means of daily mean global radiation [W/m2] for Dec 1999-Nov 2000 between versions 29.0e and 28.0e.
Relative humidity
Figure 23 shows the difference in the annual mean climatology of daily mean relative humidity for the period 1991-2020 between E-OBS versions 29.0e and 28.0e. Figure 24 shows the seasonal mean differences between these two versions for the period Dec 1999 – Nov 2000. 
Figure 23: Difference in annual mean climatology of daily mean relative humidity [%] for the period 1991-2020 between versions 29.0e and 28.0e.
Figure 24: Differences in seasonal means of daily mean relative humidity [%] for Dec 1999-Nov 2000 between versions 29.0e and 28.0e.
Wind speed
As E-OBSv29.0e does not have radiation ready at the time of writing (8 March 2024), the figures below show still the differences between v28.0e and v27.0e. Figure 25 shows the difference in the annual mean climatology of daily mean wind speed for the period 1991-2020 between E-OBS versions 28.0e and 27.0e. Figure 26 shows the seasonal mean differences between these two versions for the period Dec 1999 – Nov 2000.
Figure 25: Difference in annual mean climatology of daily mean wind speed [m/s] for the period 1991-2020 between versions 28.0e and 27.0e.
Figure 26: Differences in seasonal means of daily mean wind speed [m/s] for Dec 1999-Nov 2000 between versions 28.0e and 27
The E-OBS precipitation dataset is daily, meaning the observations cover 24 hours per time step. The exact 24-hour period can be different per region and element. The reason for this is that some data providers measure between midnight to midnight while others might measure from morning to morning. Since E-OBS is an observational dataset, no attempts have been made to adjust time series for this 24-hour offset. It is made sure, where known, that the largest part of the measured 24-hour period corresponds to the day attached to the time step in E-OBS (and ECA&D).
The global radiation dataset not only uses ground-based in-situ observations, but, in order to properly estimate spatial variability, it also incorporates the CERES global radiation satellite derived product and includes elevation as spatial predictors. On daily time-scales the global radiation is mainly dependent on cloud patterns. This makes daily global radiation spatially and temporally highly variable. The in-situ observations are interpolated using Multiple Adaptive Regression Splines (MARS). The dimension reduced satellite data is used to interpolate between the in-situ observations, where the in-situ observations are the 'anchor points' (van der Schrier et al. 2021).
Relative humidity (hu) is interpolated in the same way as the temperature fields. The input station data is transformed using √(100-hu), with hu in percent, to have a more normal distribution. These transformed values are interpolated and afterwards changed back to relative humidity itself.
The dataset for wind strength an approach inspired by the approach of Cornes et al. (2018) is used. For this dataset, we apply greedy forward selection linear regression (FSLR) to include the high-resolution effects of the auxiliary variables on monthly-mean data. These data provide a 'background' for the daily estimates. We apply cross-validation to avoid FSLR over-fitting and use full-cycle bootstrapping to create FSLR ensemble members. Then, we apply Gaussian process regression (GPR) to regress the daily anomalies. We consider the effect of the spatial distribution of station locations on the GPR gridding uncertainty. As this dataset contains several decades of daily gridded wind fields, computational efficiency is of utmost importance. We alleviate the computational cost of the FSLR and GPR analyses by incorporating greedy algorithms and sparse matrix algebra in the analyses.
The gridded wind fields are calculated as a statistical ensemble of realizations. In the present analysis, the ensemble spread is based on uncertainties arising from the auxiliary variables as well as from the spatial distribution of stations.
Cross-validation is used to tune the GPR hyper parameters. Where conventional GPR hyperparameter tuning aims at an optimal prediction of the gridded mean, instead, we tune the GPR hyperparameters for optimal prediction of the gridded ensemble spread.
The Global 30 Arc-Second Elevation Data Set (GTOPO30), a global raster Digital Elevation Model (DEM) with a horizontal grid spacing of 30 arc seconds (approximately 1 kilometre) developed by the United States Geological Survey (USGS), is used for the elevation file.
Earlier versions of E-OBS (up to v17.0) used for the interpolation of temperature , precipitation and sea-level pressure the method described by Haylock et al. (2008) and Van den Besselaar et al. (2011).
Additional post-processing for the relative humidity and global radiation datasets
An additional quality control has been performed on the relative humidity fields where unphysically low values for relative humidity have been observed as a consequence of scarce station density in combination with the gridding method. As of E-OBSv24.0e, the grid boxes with values below 5% are set to missing in each of the 20 ensemble members. This additional check mirrors the quality check applied to the input stations. The ensemble mean and ensemble spread are derived from these adjusted ensemble members, making sure that all humidity values are above 5%. Note that the approach to grid transformed values of humidity (rather than humidity directly) already makes sure that no humidity values higher than 100% are possible.
Similarly, the global radiation dataset has been post-processed in the areas and in the periods where it suffers from unrealistic values. This problem occurs in areas and periods where the station network is of low density. Following a similar procedure as with relative humidity, the same quality-control rules are applied to the grids as are applied to the stations. This means that any daily global radiation sum in each of the E-OBS ensemble members is set to missing where it falls below the 3% TOA (Top Of Atmosphere) level or exceeds the maximum expected global radiation of the earth surface on a clear sky day. The quality control procedure is detailed in the ATBD(2021). This additional post-processing is in place as of E-OBSv24.0e.
Comparison with other datasets
E-OBS is used routinely in monitoring the European climate and contributes to the monthly and annual State-of-the-Climate bulletins issued through C3S, and to the State-of-the-Climate bulletins published annually in The Bulletin of the American Meteorological Society (Europe section).
The calculation of the European averaged temperature is documented by Van der Schrier et al. (2013). In this study, a quantitative analysis of the uncertainty sources to the European average temperature indicates that the uncertainties due to urbanization, statistical interpolation, and the potential inhomogeneities in the input records to E‐OBS dominate the total uncertainty estimate for the European-averaged temperature.
The uncertainty estimate in this metric is used to make a comparison between E-OBSv6.0 and global observational datasets and global reanalyses. Although E-OBSv6.0 differs considerably with the latest versions of E-OBS (station density has increased, the method to calculate the gridded temperature estimate and the uncertainty has changed, and the spatial resolution has increased), the temperature averaged over the European continent and averaged over the year will not be dramatically affected as the averaging in time and space makes this metric independent of the improvements in the fine-scale structure. The increase in station density makes that the uncertainty estimate of the statistical interpolation has decreased, making the uncertainty estimates for the early versions of E-OBS somewhat pessimistic for the latest versions.
Figure 9: European averaged temperature with respect to the 1961-1990 climatology as calculated for E-OBSv6.0 compared to global observational datasets (left) and global reanalysis (right). The lower plot gives the E-OBS uncertainty margins in grey boxes and the difference between E-OBS and the global datasets (Van der Schrier et al. 2013)
The comparisons between E-OBS and global observational and reanalyses datasets are given in Figure 9. A comparison between E-OBSv7.0 and ERA-Interim, focused specifically on indices of the numbers of days above the 90th percentiles of daily maximum temperature, the number of days below the 10th percentiles of daily minimum temperature and trends in daily surface maximum/minimum temperature extremes across Europe, is documented by Cornes and Jones (2013).
A comparison with the observational CRUTS3.1 dataset, which has a spatial resolution more comparable to that of E-OBS, is given in Figure 10.
Figure 10: European averaged annual temperatures as calculated from E-OBS version 6.0 (in black), including the error margins in grey boxes, and the European averaged annual temperature as calculated using the CRUTS3.1 temperature dataset (in red).
Guidance on the use of an ensemble dataset
Ensemble datasets are climate datasets that consist of a number of equally probable realizations, and relate to data in gridded format. The ensemble aims to give a measure of uncertainty in the data field; such datasets are widely used in a number of areas of climate change science, like perturbed physics simulations of climate models or numerical weather prediction.
The term "ensemble" is also used when referring to multiple realizations from gridded observed datasets. Such datasets, like E-OBS, are formed from the interpolation of station values. Although the aim is the same as ensembles calculated for model simulations – to quantify uncertainty in the data – the generation and hence interpretation of the realizations is quite different. With gridded observational datasets, decisions have to be made for a number of features that affect the final gridded field of the key parameters involved in the gridding algorithm. These could include:
- Search radius for inclusion of stations influencing a grid box
- Estimates of the impacts of homogeneity issues on the quality of the input station data
- Impact of a number of possible co-variates used in the gridding (e.g. latitude, longitude, elevation and distance from coast or inland water body)
- Including station data where the base period (e.g. 1961-90 or 1981-2010) is less precisely known
The usual approach in the production of a gridded dataset is to determine the best or most likely values for the key parameters to interpolate station values to values on a regular grid. With an ensemble version of the dataset, a number of these key parameters can be varied producing a range of possible gridded datasets (referred to as the ensemble). For most variables in the E-OBS dataset,100 ensemble members are developed. The advantage of producing a dataset this way is that with the ensemble it can be easier to determine the errors of estimate of grid box averages. Using standard statistical approaches this can be difficult as a number of the error components have spatial and temporal structures which are difficult to model.
Where users require a single measure of the interpolated daily fields, then the "best guess" values should be used. However, the standard error should always be consulted as the uncertainty of the gridded field varies across the domain, and is ultimately determined by the variations in station coverage.
The individual ensemble members are mainly intended for users who require the uncertainty in the gridded fields to propagate through to various other applications. If a user requires this rainfall data for hydrological modelling then each of the ensemble members could be fed into the hydrological model. In this way the uncertainty in the rainfall interpolation would propagate through to the hydrological model output.
Specifics for E-OBSv20.0e and E-OBSv19.0e
Subsequent versions of E-OBS are updated using recent data from the European national meteorological services and by inclusion of new series. It is the amount of input station data that explain the differences between subsequent versions of E-OBS.
E-OBSv20.0e is released in October 2019 and spans the period 1950-01-01 to 2019-07-31. New series and updates have been included for Italy, Croatia, Norway and Russia. Monthly, half-yearly and yearly updates are continued for Germany, Czech Republic, Bosnia and Herzegovina, Norway, Slovenia, Finland, Ireland, Sweden, Luxembourg, Netherlands, Portugal, Spain, Switzerland, France, Denmark, UK and the regional meteorological service of Catalonia (Spain).
Figure 11 and 12 give an indication of the differences between v20.0e and v19.0e for precipitation and temperature for the year 2000. The areas where new station data has been added are immediately recognizable in these figures. This holds perhaps even more so for the standard deviation of the daily differences between the datasets.
...
Figure 11: Differences between E-OBSv20.0e and E-OBSv19.0e for winter (DJF) precipitation (upper left) and summer (JJA) precipitation (upper right) for the year 2000 in mm/3 months. The bottom figure shows the standard deviation in daily differences for precipitation (mm/day) between E-OBSv20.0e and E-OBSv19.0e.
Anchor
Figure 12: Differences between E-OBSv20.0e and E-OBSv19.0e for winter (DJF) daily average temperature (upper left) and summer (JJA) daily average temperature (upper right) for the year 2000 in °C. The bottom figure shows the standard deviation in daily differences for daily average temperature (°C) between E-OBSv20.0e and E-OBSv19.0e.
Data licence
The data licence for E-OBS is more restrictive than for other CDS products. In this paragraph, the The background for this difference is explained, below.
There is an agreement between EUMETNET, as the representative of a group of European NMHSs, and the EEA, acting under the delegated task of the European Commission as Copernicus in-situ coordinator, that derived data products, like E-OBS can be provided under an open data licence. However, the group of NMHS data providers to ECA&D (and therefore E-OBS) is larger than the 31 European NMHSs which are member members of EUMETNET. In addition, other data holding institutes (other than meteorological servicesand hydrological service institutes) provide data for E-OBS which do not fall under this agreement.
The situation fact that only part some of the data supplying institutes have agreed to an open data licence of for derived data, makes means that the data licence of for E-OBS needs to resort to be the rather strict licence currently in use, meaning such that the dataset can only be used for non-commercial research and education.
Technical summary
E-OBS has been in operational production since 2009 with bi-annual updates. The Technology Readiness Level for E-OBS is TRL 9 – Actual system proven in an operational environment. The E-OBS dataset is provided in the NetCDF-4 format.
References
Cornes, R. C., & Jones, and P.D. Jones (2013). How well does the ERA‐Interim reanalysis replicate trends in extremes of surface temperature across Europe?. J. Geophys. Res. (Atmospheres), 118(18), 10-262. van der Schrier, G., Knap, W., Dirksen, M., doi:10.1002/jgrd.50799
Cornes, R., G. van der Schrier, E.J.M. van den Besselaar, and P.D. Jones. 2018: An Ensemble Version of the E-OBS Temperature and Precipitation Datasets, J. Geophys. Res. Atmos.(Atmospheres), 123. doi:10.1029/2017JD028200
De Baar, J., et al. G. van der Schrier, and E.J.M. van den Besselaar (2021) Wind speed extension to E-OBS (in preparation)
ECA&D Project Team (2012) European Climate Assessment & Dataset Algorithm Theoretical Basis Document (ATBD). Royal Netherlands Meteorological Institute KNMI, De Bilt, NL, version 10.5
Haylock et al. (2008) A European daily high-resolution gridded data set of surface temperature and precipitation for 1950-2006. J. Geophys. Res. (Atmospheres), doi:10.1029/2008JD010201
Klein Tank et al. (2002) Daily dataset of 20th-century surface air temperature and precipitation series for the European Climate. Intern. J. Climatol. 22:1441-1453, doi:10.1002/joc.773
Morice, C. P., Kennedy, J. J. , RaynerKennedy, N.A. , & Rayner and Jones, P. D. (2012). Quantifying uncertainties in global and regional temperature change using an ensemble of observational estimates: The HadCRUT4 data set. J. Geophys. Res. (Atmospheres), 117(D8).. Res. (Atmospheres), 117(D8).
Van den Besselaar, E. J. M., M. R. Haylock, G. van der Schrier and A.M.G. Klein Tank Vvan den Besselaar et al. (2011) A European daily high-resolution observational gridded data set of sea level pressure. J. Geophys. Res. (Atmospheres), 116(D11110), doi:10.1029/2010JD015468
Van den Besselaar, E. J. M., Klein Tank, A. M. G., Van der Schrier, G., & and Jones, P. D. (2012). Synoptic messages to extend climate data records. Journal of Geophysical Research: AtmospheresGeophys. Res. (Atmospheres), 117(D7), doi:10.1029/2011JD016687
Van der Schrier, G., E. J.M. van den Besselaar, A.M.G. Klein Tank, and G. Verver (2013). Monitoring European average temperature based on the E-OBS gridded data set, J. Geophys. Res. (Atmospheres), 118, 5120-5135, doi:10.1002/jgrd.50444.
Van der Schrier, G., Knap, W., Dirksen, M., Van den Besselaar, E.J.M. and Klein Tank, A.M.G. (2021) Daily global radiation maps based in in-situ observations for Europe. (in preparation)
Van den Besselaar, E.J.M., van der Schrier , and G., de Baar, J. (2021) Daily relative humidity maps for Europe based on in-situ observations (in preparation)
| Info | ||
|---|---|---|
| ||
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 agreementAgreement signed on 11/11/2014 and Contribution Agreement signed on 22/07/2021). 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. |
Related articles
| Content by Label | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|