Analysis of future climate conditions is based on simulations of the most recent generation of global climate models (GCM). These underwent a rapid evolution over the past decades, primarily with regard to the number of interconnected processes and feedbacks in the climate system, which the current models are capable to take into account. Also the spatial resolution of GCMs improved, although for detailed overview of local climate it is still necessary to use methods of precising the output, a so-called downscaling. This means either incorporating more detailed regional climate models (RCM) or by using statistical methods applied directly to the GCM output (for example the incremental method etc.).
Given the outputs of climate models are associated with systematic errors (due to the necessary simplification of the complex real-world processes), they need to be corrected in order to obtain meaningful results about the simulated properties of the climate system. In general, when working with expected values of meteorological elements (such as seasonal and annual averages), the changes given by the models can be treated as they are, without any modification. The problem arises when analyzing daily data and extreme values such as temperature maxima and minima, precipitation values above given thresholds etc., because of the low spatial resolution of GCMs.
The experience of the authors of the presented methodology so far leads to a preference for refining the GCM outputs for use at the local level using statistical methods instead of involving dynamic downscaling using RCMs. This is mainly due to the systematic errors of the RCMs (more humid and cold climate during the control historical RCM run) and also due to the fact that while for the GCMs we have simulations from the latest generation of climate models, CMIP6 (used in the latest IPCC report, AR6), for the RCMs, Euro-CORDEX simulations driven by the older generation of global climate models, CMIP5 (i.e. models from the previous IPCC report, AR5), are available for Europe. This means that the available RCM simulations do not take into account the latest scientific findings.
Many of the latest CMIP6 GCM simulations include models with varying degrees of spatial details. Most simulations of the 21st century climate evolution have a horizontal spatial resolution of approximately 100 to 250 km. There is also a small subset of GCMs with resolutions around 50 km, but their simulations end in the mid-21st century. The individual GCMs also differ from each other in the complexity of their description of processes in the climate system, the way in which smaller-scale phenomena are parametrized, and the formulation and numerical solution of the underlying physical equations. It is inevitable that the simulated climate diverges to some extent from the reality and that this divergence varies in space, time and across physical quantities. Therefore, GCMs that best represent the climate of Central Europe were preferred for the simulations of the future climate of Central Europe. At the same time, it is necessary to ensure that the preferred GCMs, which are only a subset of all the available GCMs, affect the future climate evolution in the same way, with the same degree of uncertainty, as the full set of available GCMs. That is, that the selected subset of GCMs does not represent models, which under the same conditions, expect, for example a higher temperature increase (or changes in precipitation, wind, sunshine, etc.) than models that are outside the selection. For this purpose a selection methodology has been proposed (for narrowing down the ensemble of models) described in:
Meitner, J., Štěpánek, P., Skalák, P., Dubrovský, M., Lhotka, O., Penčevová, R., Zahradníček, P., Farda, A., Trnka, M. (2023): Validation and Selection of a Representative Subset from the Ensemble of EURO-CORDEX EUR11 Regional Climate Model Outputs for the Czech Republic. Atmosphere 2023, 14, 1442. https://doi.org/10.3390/ atmos14091442
In accordance with this methodology, those models that were unable to reliably simulate the climate of Central Europe in the recent past were excluded from the set of approximately twenty CMIP6 GCMs on the basis of validation. From the remaining models, 7 GCMs were then selected so that this shortlist was representative in its statistical properties of the entire original set of models, but allowed working with fewer simulations. One of the reasons for narrowing down the ensemble of GCMs is when the individual GCMs are used as a source of input meteorological data for hydrological models or models simulating the impact of climate change on the landscape and its management. In this case, the need to integrate the entire original GCM set presented enormous demands on computing power and such a task was often practically unsolvable. The selection of GCMs was made taking into account all the basic meteorological elements that are further analyzed and used for the calculation of reference evapotranspiration and soil moisture by the SoilClim model, respectively. The selection of models together with the available climate change scenarios is presented in the following table. GCMs with finer spatial resolution (100 km or less versus 250 km) were preferred.
| Model | Available climate change scenarios | Model spatial resolution in km |
|---|---|---|
| CNRM-CM6-1-HR | SSP126, SSP585 | 50 |
| CMCC-ESM2 | SSP126, SSP245, SSP370, SSP585 | 100 |
| EC-EARTH3 | SSP126, SSP245, SSP370, SSP585 | 100 |
| GFDL-ESM4 | SSP126, SSP245, SSP370, SSP585 | 100 |
| MPI-ESM1-2-HR | SSP126, SSP245, SSP370, SSP585 | 100 |
| MRI-ESM2-0 | SSP126, SSP245, SSP370, SSP585 | 100 |
| TAIESM1 | SSP126, SSP245, SSP370, SSP585 | 100 |
Climate change scenarios act as a source of the so-called boundary conditions for GCMs and reflect different possible future trajectories of the world not only in terms of emissions or the resulting concentrations of greenhouse gases in the atmosphere but also in terms of different economic and social developments on the planet. The latest IPCC AR6 works with socio-economic development scenarios, the so-called Shared Socioeconomics Pathways (SSP).
In simple terms, the different climate change scenarios used as input to GCM simulations can be interpreted as follows:
GCM outputs, unless we are only concerned with the relative change in meteorological elements, cannot be used directly. They are subject to a systematic error (e.g. underestimation of temperature by 1 °C or overestimation of precipitation by 25%, etc. in Central Europe) which must first be corrected. Alternatively, one can work with climate change resulting from climate model simulations, which is related directly to the observed data. The latter approach is referred to as the 'incremental method' or 'direct modification' and is traditionally used in the Czech Republic for modelling the impacts of climate change on, for example, the hydrological balance, as this method is more robust than using climate model simulations with bias correction. To use the 'incremental method' in the daily step, it is suitable to apply transformations that consider not only changes in averages but also changes in variability. This is made possible, for example, by the Advanced Delta Change ("ADC") method. The ADC method makes it possible to include the change in variability in the transformation. This simply means that the extremes can change differently than the mean, which correctly reflects the situation as we observe it in the real world. When deriving changes in precipitation from a climate model, the ADC method also considers systematic errors in the simulation, which may not be linear. Further details can be found in van Pelt et al. (2012).
In the application of the ADC method, the daily values of meteorological elements are processed in a weekly step to preserve their annual runs, and smoothing of the transformation parameters is performed. Air temperature is transformed linearly, unlike precipitation. Other meteorological variables (global radiation, relative humidity and wind speed) are modified by multiplying by the ratio of the averages over the GCM control run period and the GCM scenario simulation period. The transformation parameters are again smoothed.
In the case of the Czech Republic, the input data of station measurements are applied in the form of technical data series. These are time series of measurements that have undergone a thorough quality control, error correction, homogenization and filling in of data gaps in the measurements (Štěpánek et al., 2011, 2013). The data coming from the stations of the Czech Hydrometeorological Institute (CHMI) are interpolated into a fine spatial grid with a resolution of 500 m before applying the ADC method. The same approach is applied to the climate model data in a daily time step. The resulting information regarding future climate change is therefore in the form of a map with a spatial resolution of 500 m and a temporal resolution of 1 day.
In case of processing for Central Europe, the E-OBS dataset (version v27.0e) of gridded station observations (https://cds.climate.copernicus.eu/cdsapp#!/dataset/insitu-gridded-observations-europe?tab=overview) was used instead of the CHMI station measurements. This dataset is based on station measurements collected in the ECA&D database (https://www.ecad.eu/).
With respect to the interpretation of the results, it is important to note that in addition to the 1981-2010 reference period, we are working with 30-year time frames for the future climate: 2015-2044 (referred to as "2030"), 2035-2064 ("2050"), 2055-2084 ("2070"), and 2070-2099 ("2085"). The periods overlap with each other. Within these time frames, statistical characteristics (including outliers) can be evaluated over the period. Similar to climate model simulations, it does not make sense to analyze and present individual days or years, but only statistics for the whole period. Long-term trends can then be evaluated by relating individual (moving) periods in the future climate.
As mentioned above, the basis for the outputs presented on this website are 500m resolution maps for the Czech Republic and grid layers at the spatial resolution of the E-OBS dataset (approx. 10 km) for the Central Europe region. There are 4 SSP scenarios describing the projected future evolution of the world and a set of 7 CMIP6 GCM models accurately representing the original larger ensemble of about 20 models. Daily data of basic meteorological variables (air temperature, precipitation, wind speed, humidity, sunshine and radiation) are available from which the necessary characteristics can be derived, including those describing extremes.
Considering the detailed temporal (daily data) and spatial resolution (500 m for the Czech Republic and 10 km for Central Europe), both temporal and spatial data aggregation was performed for presentation purposes. The aim of this application is to provide users with relevant information regarding the potential problems that may arise at a regional level in the context of climate change. Upon requests for more detailed temporal resolution or a more precise location, the creator of this application can provide more detailed information beyond the scope of this public presentation.
The climate change information is aggregated in the form of long-term characteristics, which were derived using all the 7 selected GCMs and all SPP scenarios. From the model ensemble, in addition to the most likely future climate development, the range (margins) within which this development can take place is also assessed. This processing has been carried out for each administrative unit as mentioned below.
Basic processing was performed for 30-year periods for both the present (1981-2010, denoted as 1995 and 1991-2020, denoted as 2005 in the outputs) and the expected future climate. Since, after correcting the GCM outputs, the statistical properties of these outputs are consistent with the present station measurements (or the values in the E-OBS database that is based on the station measurements), it is possible to combine both the station and GCM outputs (e.g., for 30-year periods centered on the present years, e.g., 2023 and 2025).
Aggregations for administrative units have been created to present the outputs on this website.
In the case of the Czech Republic, these are cadasters; in the case of data for Central Europe, they are based on NUTS regions or other administrative units appropriate for the particular country. Specifically, for the Czech Republic these are municipalities with extended competence, in Slovakia they are districts, in Germany they are NUTS3 regions, in Poland they are powiats, in Austria they are districts (one level more detailed than NUTS3) and in Hungary again they are districts (also one level more detailed than NUTS3).
The spatial data of the regions of the European domain and the corresponding national boundaries have been geometrically simplified with respect to the volume of data and smooth operation of the website. This may lead to spatial distortions of the detailed scale boundaries and inaccuracies with respect to the actual position of administrative boundaries.
The form of data simplification used: Douglas-Peucker algorithm with a tolerance of 0.001° (PEUCKER, T. K. (1976): A theory of the cartographic line. International yearbook of cartography, pp. 134-143). This algorithm simplifies the boundary lines of each region, by segmenting the lines into zones of a predetermined size. The boundary breakpoints of these zones are then used as new breakpoints of the simplified line and breakpoints located inside the zone are dropped. This simplifies the line progressively.