Attribution of recent evapotranspiration trends in Spain using multivariate statistics

  • Andreas Baum (Main supervisor)
  • Monica Garcia (Supervisor)

Activity: Examinations and supervisionSupervisor activities

Description

Climate change threatens the integrity of Earth’s ecosystems. In order to measure the loss in function suffered by these ecosystems, evapotranspiration, which serves as a holistic indicator of ecosystem resilience, is widely used.

In this observational study, the contribution of different environmental factors to detected trends in evapotranspiration in Spain is examined. The study models vineyards in the BSk climate area, permanent crops, and forests by which land covers and related Köppen-Geiger climate areas are included. In order to examine the contribution of the factors, we utilize general linear models using different methodologies, such as cross validation and regularization, to avoid overfitting. Based on performance tests, Lasso regression is used throughout the analysis. To interpret results, point estimates and associated confidence intervals of model coefficients are computed using non-parametric bootstrap, and used to show general tendencies of the included factors.

The results show how positive trends in leaf area index, precipitation and net radiation affects the evapotranspiration trend positively. Additionally, a decrease in the actual water vapor pressure trend will enhance the effect of the net radiation trend on the evapotranspiration trend. In general, there are no distinctive effects on the evapotranspiration trend depending on climate zones, but forests are shown to be more resilient than permanent crops in terms of changes in the trend of evapotranspiration.
Period21 Mar 202020 Jun 2020
Degree of RecognitionNational