B1735
Title: Inferring the relationship between soil temperature and normalized difference vegetation index with machine learning
Authors: Steven Mortier - Antwerp University (Belgium) [presenting]
Tim Verdonck - KU Leuven and UAntwerpen - imec (Belgium)
Tom De Schepper - Antwerp University (Belgium)
Steven Latre - imec (Belgium)
B Didrik Sigurdsson - Agricultural University of Iceland (Iceland)
Ruth P Tchana Wandji - Agricultural University of Iceland (Iceland)
Amir Hamedpour - Svarmi (Iceland)
Bart Bussmann - Antwerp University (Belgium)
Abstract: Changes in climate can greatly affect the phenology of plants, which can have important feedback effects, such as altering the carbon cycle. These phenological feedback effects are often induced by a shift in the start or end dates of the growing season of plants. The normalized difference vegetation index (NDVI) serves as an indicator of the presence of green vegetation in the observed area and can also provide an estimation of the plants' growing season. The effect of soil temperature (ST) is investigated on the timing of the start and peak of the season (SOS and POS) and maximum annual NDVI value (PEAK) in subarctic grassland ecosystems. The impact of other meteorological variables, namely air temperature, precipitation, and irradiance, is also explored in vegetation phenology. Using machine learning and Shapley additive explanations (SHAP) values, the relative importance and contribution of each variable to the phenological predictions are analyzed. The results reveal a significant relationship between ST and SOS and POS, indicating that higher STs lead to an earlier SOS and POS. The other meteorological variables had a varying impact on the SOS and POS depending on the year of study. Ultimately, the results contribute to the knowledge of the relationships between ST, air temperature, precipitation, irradiance, and vegetation phenology, providing valuable insights for predicting and managing subarctic grasslands in the face of climate change.
