Mapping Vegetation Dynamics Under Drought Stress: Integrating Satellite, Meteorological, and Terrestrial Geospatial Data

In an era of climate change, as the intensity and frequency of droughts are expected to increase, effective
climate adaptation strategies become increasingly important for both natural ecosystems and society.
Understanding vegetation's responses to drought and monitoring the impacts of drought on terrestrial
ecosystems are essential for developing reliable decision-support systems. This thesis aimed to develop,
test, and apply methodologies for detecting and quantifying the spatially explicit effects of drought on
Swedish agricultural and forest ecosystems and their productivity. The thesis includes four research
papers. Paper I focused on developing a change-detection framework that quantifies immediate and
delayed drought-related vegetation changes in forest ecosystems and their recovery times. Results
showed that using Sentinel-2 satellite data, successive stages of vegetation drought stress could be
detected, and recovery times could be estimated. Paper II focused on identifying the drivers and mapping
forest areas at high risk of disturbances from the European spruce bark beetle (Ips Typographus L.) due
to drought-related effects. The forest stands predisposed to bark beetle attacks were accurately
identified, and the methodology provided quantifiable information on the importance and dynamics of
environmental features contributing to the predisposition risk. Paper III developed a causal inference
framework to quantify the impacts of drought on crop yields while accounting for confounding variables.
The methodology considered the effects of crop rotation and variations in local soil attributes, predicting
regional and field level yields and identifying factors driving local variations in yield losses. Paper IV
estimated the foliar and radial growth responses of Norway spruce (Picea abies (L.)) and Scots pine
(Pinus sylvestris (L.)) trees to drought at a multidecadal level. Results indicate that the drivers of variance
in annual radial growth were identified, with pine stands exhibiting higher drought resilience and faster
recovery rates than spruce. Overall, this thesis demonstrates that effectively using multi-source
geospatial data can identify and map the risks and effects of drought on forests and agricultural
ecosystems over time and space, thereby enhancing our understanding of vegetation drought responses
and resilience patterns. The methodologies presented in this thesis provide essential information for
policy planning and decision-making supported by scientific knowledge, contributing to a shift towards
more sustainable management practices in a changing climate.