Exploring drought-induced tree mortality in Luxembourg using aerial imagery and environmental data
Wednesday, August 4, 2021
Link To Share This Presentation: https://cdmcd.co/bYGj4D
Selina Schwarz, Christian Werner and Nadine K. Ruehr, Institute of Meteorology and Climate Research - Atmospheric Environmental Research (IMK-IFU), Karlsruhe Institute of Technology, Garmisch-Partenkirchen, Germany, Rosa Haffter and Fabian Fassnacht, Institute of Geography and Geoecology (IfGG), Karlsruhe Institute of Technology, Karlsruhe, Germany
Institute of Meteorology and Climate Research - Atmospheric Environmental Research (IMK-IFU), Karlsruhe Institute of Technology Garmisch-Partenkirchen, Germany
Background/Question/Methods The changes in the climate system over the last decades have led to an increase in the frequency of droughts globally. A recent example are the summers of 2018 and 2019 in Central Europe, during which forests suffered under harsh drought conditions. Forests are pivotal for humankind, they cover about 31% of the global land area and are a net sink of about 20% of anthropogenic CO2 emissions. Thus, increases in tree mortality in response to drought extremes can result in large, yet uncertain, feedback to the climate system. Here, we study drought-related tree mortality in response to the European 2018 drought event. More specifically, we seek to identify site and stand characteristics that might lead to higher mortality rates. We use high-resolution orthoimages of the country of Luxembourg and deep learning classification algorithms to create a data set of tree mortality between 2017 and 2019 for the whole of Luxembourg at individual tree level. To train the image classification algorithm we use a hand delineated data set of ~8,000 dead trees in a catchment area of Luxembourg. We then select data sets of environmental variables related to site characteristics, including Temperature, Precipitation, Soil Type, Ground Water, Forest Type and Topography to explain the variance in tree mortality between 2017 and 2019, using a Generalized Additive Model (GAM). Results/Conclusions First results show that we are able to explain 33.7% of the observed variance in tree mortality using data from the catchment area only. Influential predictors are the distance to forest edge, tree height and the Topographic Wetness Index (TWI). We expect, that we will be able to explain a higher percentage of tree mortality by including additional site characteristics like competition, direction of forest edge and distance to already dead trees for the whole of Luxembourg. The results of the study contribute to an improved understanding of which forest sites in central Europe are particularly threatened by increasingly severe drought events under climate change.