Characterizing drought response patterns of Central European grasslands based on four decades of Landsat and Sentinel-2 data

Climate change will lead to more frequent, longer, and more severe drought and heat periods, with unforeseen consequences for ecosystems globally. In Central Europe, for instance, grasslands deteriorated immediately in response to unprecedented drought and heat in recent years with major impacts on vegetation productivity. However, drought impacts can vary considerably in space and time, suggesting a complex network of underlying drivers. Factors such as soil characteristics, topography, species composition, and land-use modify the severity and duration of vegetation drought in grasslands on local to regional scales, yet our understanding is still underdeveloped. To better understand the complex drivers of grassland response to drought, it is indispensable to characterize drought impacts covering large environmental gradients in a spatially explicit way. While challenging, this task can be addressed with dense satellite-borne multispectral time series. In this study, we investigated how grasslands respond to meteorological and soil moisture drought and how this relationship varies with environmental and land management gradients in Central Europe. We used four decades of remote sensing time series from Landsat/Sentinel-2 to quantify vegetation drought at 30m spatial resolution across all grasslands in Germany. We applied a modeling approach developed in previous studies (Kowalski et al., 2023, 2022) for estimating time series of green vegetation, dry vegetation and soil ground cover percentages. We then derived monthly time series of the Normalized Difference Fraction Index (NDFI), which contrasts dry vegetation and soil relative to green vegetation, thereby providing a physically grounded indicator tracking grass dieback over the growing season. We calculated mean NDFI anomalies from June to September for each growing season from 1984-2021 using the 1984-2021 average as a baseline. We assessed the relation of NDFI anomalies to vapor pressure deficit, climatic water balance, and soil moisture anomalies derived from monthly ERA-5 Land time series. Moreover, we investigated how these relations varied spatially by stratifying grasslands according to environmental (e.g., precipitation, temperature, topographic derivatives, soil available water capacity) and land management factors. For the 38-year timespan, we found several single- and multi-year vegetation drought events including the strongest events in 2003 and 2018. The 2018 event featured the most severe NDFI anomaly of +0.32, translating into 32% higher than average dry vegetation and soil cover across all grasslands in Germany. NDFI anomalies varied spatially with a tendency for highest anomalies in the central uplands and northern lowlands, while grasslands in the southern Alpine region were less affected. NDFI anomalies had consistent moderate to strong correlations with meteorological and soil moisture drought. The overall highest correlations occurred in July and August indicating short time lags of NDFI anomalies. Our results confirm strong and spatially heterogenous impacts of meteorological and soil moisture droughts on grasslands. Drought periods in the next decades will thus pose substantial challenges for grassland vitality and productivity in Central Europe. Our study further shows the value of remote sensing for analyzing vegetation dynamics across grassland ecosystems, thereby enhancing our knowledge on fundamental processes in these complex systems.

 

Kowalski et al. 2022. https://doi.org/10.1016/j.rse.2021.112781

Kowalski et al. 2023. https://doi.org/10.1016/j.rse.2022.113449