High-Resolution 1-arc min Climate Model Data: Downscaled and Bias-Corrected CMIP6 Model Outputs for Three BioValue Pilot Countries at Weekly Time Resolution (version-1.0)

Abstract

As part of the H2020 project BioValue: Fork-to-farm agent-based simulation tool augmenting BIOdiversity in the agri-food VALUE chain, a weekly-aggregated high-resolution climate dataset was developed. This dataset contributes to project deliverable D6.4 "Climate change – biodiversity interlinks" that was prepared by the group of Justus Liebig University Giessen. The dataset comprises of statistically downscaled and bias-corrected climate model simulations of three CMIP6 models: MPI-ESM1-2-HR (von Storch et al. 2017), EC-Earth3 (Döscher et al. 2022), and CNRM-CM6-1-HR (Aurore et al. 2019). The climate variables are mean daily temperature (ta), maximum daily temperature (tx), minimum daily temperature (tn) and daily precipitation (pr). The statistical downscaling approach combines the Perfect Prognosis (PP) method with the analog method. Specifically, method M6, as described in Bedia et al. (2019), was applied. Predictors for precipitation include specific humidity at 700 and 850 hPa (hus700, hus850), mean sea-level pressure (mslp), 500 hPa geopotential height (zg500), and 850 hPa air temperature (ta850). For temperature (mean, minimum, and maximum air temperature: ta, tn, tx), only the zg500 was used as a predictor. The predictand observational data consisted of the European Meteorological Observations at 1arcmin (EMO-1) daily dataset (1990–2014) (Gomes et al. 2020), while ERA5 reanalysis data (Hersbach et al. 2017) served as the perfect predictors. Following the downscaling processing, bias correction was performed using the Quantile Delta Mapping method (Canon et al. 2015) to correcting systematic biases between model outputs and observations. The post-processed model simulations were elaborated at a daily resolution. For the sake of ease and overcoming the challenge of extensively large data, the dataset is prepared at a weekly scale by an aggregation into non-overlapping 7-day blocks. For precipitation, 7-day sums were calculated and for temperature variables, 7-day means. NOTE: Since aggregation was carried out per calendar year, the final "week" of the year may be shorter than seven days. The dataset includes the following climate variables: • pr: Precipitation total (7-day sum, unit: mm) • ta: Mean air temperature (7-day mean, unit: °C) • tn: Minimum air temperature (7-day mean, unit: °C) • tx: Maximum air temperature (7-day mean, unit: °C) The countries currently included are Cyprus, Estonia, and Germany. These represent a first selection among the BioValue pilot site areas, which ultimately cover seven EU countries and Turkey. Importantly, the dataset focuses on specific months of interest, as defined by the project partners and pilot areas, that are critical to crop development, the yield, and regional climate characteristics at each country. Three of the BioValue pilot sites are provided in the current version of the dataset, namely: • Cyprus: March to July • Estonia: May to October • Germany: March to October This tailoring ensures that the climatic information is closely aligned with the critical periods for agricultural production and biodiversity considerations in each region. The data are provided in NetCDF format, with one file per country, model, and time period. Two time periods are considered, namely the historical period 1990–2014 and the future period 2015–2050 under the scenario SSP5-8.5. All post-processed datasets are based on a longitude/latitude regular grid with a spatial resolution of approximately 1 arcminute, allowing detailed analysis across the selected regions. Users can explore and process the files using tools such as CDO (Climate Data Operators - https://code.mpimet.mpg.de/projects/cdo ) or ncview (https://cirrus.ucsd.edu/~pierce/software/ncview/quick_intro.html) for quick visualization. However, for more advanced analysis and handling, it is highly recommended to work with the data in Python (e.g., using xarray, netCDF4) or R (e.g., using the ncdf4 or terra packages). NetCDF files may be stored in compressed format. Be aware that they can grow significantly in size once decompressed, so ensure you have sufficient disk space before unpacking. The naming convention of the NetCDF files follows a clear structure: Country_Model_Variable_bc_weekly7b-(sum/mean) _Period_final.nc where, • Country refers to the Bio-Value-pilot-cite, • Model refers to the CMIP6 climate model used (e.g., MPI-ESM1-2-HR, EC-Earth3, CNRM-CM6-1-HR), • Variable refers to meteorological variables like precipitation (pr), ta, tn and tx, • Period denotes the time span (e.g., 1990-2014 or 2015-2050). For example, a file named Germany_MPI-ESM1-2-HR_pr_bc_weekly7b-sum_1990-2014_final.nc contains downscaled and bias corrected precipitation historical climate data for Germany based on the MPI-ESM1-2-HR model.