How water and carbon ecosystem services vary over land covers and extreme weather events across Germany?

To ensure sustainable development it is vital to account the stocks and flows of ecosystem services, understand the status quo of these resources and project how resilient or vulnerable they are to future climate and land cover change. In this study, we applied the U.S. Forest Service eco-hydrological model, Water Supply Stress Index (WaSSI), to estimate the monthly spatial dynamics of green & blue water resources and evaluate the ecosystem services (water supply and carbon sequestration) across sixteen German states by land covers and during extreme drought events. The simulated discharge (Q), evapotranspiration (ET) and Gross Primary Productivity (GPP) from upstream watersheds were validated against measurements from gauging stations, eddy covariance (EC) data, and remotely sensed ET and GPP estimates. Our results showed that eleven out of twelve watersheds modeled Q bias and determination coefficient (R²) are within ± 25% and above 0.60, respectively. Similarly, when we compared ET against EC data, ten out of eleven watersheds had R² above 0.60 and seven out of eleven watersheds have Kling-Gupta efficiency above 0.6. The R² between simulated and Moderate Resolution Imaging Spectroradiometer (MODIS) ET was around 0.48 with a gradient of 0.63. The model bias between simulated ET and precipitation minus observed discharge (P-Q _observed) values for all the validated watersheds was within ± 25%. Likewise, modeled GPP was higher than MODIS GPP by 16% and a lower R² (0.37). A comparison to Copernicus GPP (CGLS-GPP) gave a much better R² (0.70) with an overestimation of 7%. Moreover, a land cover specific comparison between simulated GPP and EC observed GPP showed nine out of fourteen watersheds had a model bias within ± 25% and Nash-Sutcliffe efficiency above 0.4, while twelve watersheds had R² above 0.60. Overall, the validation results demonstrate that WaSSI can capture seasonal hydrological and carbon cycles reasonably well. It is estimated that the mean annual ET across Germany is 530 ± 49.5 mm yr^-1, the mean annual water yield is 259 ± 173.5 mm yr^-1, and the mean annual Net Ecosystem Productivity (NEP) is 308.3 ± 78.2 g C m^-2 yr^-1. The annual water yield and carbon sequestration at the German national scale was around 84.86 billion m³ yr^-1 and 106.03 Tg C yr^-1, respectively. We found that Mecklenburg-Vorpommern (-1.91 Tg C/yr) and Thüringen (-0.57 Tg C/yr) were the only two states where anthropogenic CO₂ emissions were less than NEP. Across Germany, cropland and deciduous broadleaf forest are the largest share of water supply and carbon sequestration, respectively.  We found the severe drought events of 2003 and 2018 in Germany caused significant decrease in Q (29.6% & 26.8%), GPP (8.8% & 11.7%), and NEP (18.5% & 24.7%) due to decrease in P (22.7% & 25.5%) and ET (8.7% & 11.7%). In the next step, the potential impacts of different adaptive land cover and climate change scenarios on ecosystem services will be studied.