Skip to main content
Dryad logo

Satellite data reveal differential responses of Swiss forests to unprecedented 2018 drought

Citation

Sturm, Joan T. (2022), Satellite data reveal differential responses of Swiss forests to unprecedented 2018 drought, Dryad, Dataset, https://doi.org/10.5061/dryad.bk3j9kdd8

Abstract

The summer drought of 2018 caused major damages to forest ecosystems. In this dataset, various environmental variables (precipitation anomaly, temperature anomaly, climatic water balance anomaly, elevation, slope, aspect, exposition, potential direct incedent radiation, distance to the forest edge, tree type, and species heterogeneity) have been stratified into ten sections and the proportion of forest pixels with severe changes (equal or more than 10% either positive or negative) in the normalized difference water index (NDWI) of the forest canopy from 2017-2018, 2018-2019, and 2017-2019 in Switzerland declared.

Methods

All Copernicus Sentinel-2 level-1C satellite imagery covering Switzerland has been collected for August 2017, 2018, and 2019. Atmospheric correction was applied using the Sen2Cor processor to generate level-2A data. Furthermore, an empirical line-based radiometric co-registration using pseudo-invariant features for each tile has been applied to increase the consistency between the images of the time series. Clouds and cloud shadows have been masked with simple band thresholds in band 2 (blue; disregarding pixels with more than 5% reflectance) and band 8 (near infrared; disregarding pixels with less than 15% reflectance). To avoid artificial atmospheric effects and other sensor errors the median per pixel and spectral band was calculated from the pre-processed and masked images acquired for August of a given year (20172019). From this, the normalized difference vegetation index (NDWI) has been calculated using band 8 and band 11(shortwave infrared). To assess drought impacts on forests in Switzerland to the 2018 drought, the NDWI changes between 2017-2018 (resistance), 2018-2019 (recovery), and 2017-2019 (resilience) have been calculated as ratios. The continuous change measurements were converted to binary data using a 10% threshold (strong positive and negative changes).

Eleven environmental variables (precipitation anomaly, temperature anomaly, climatic water balance anomaly, elevation, slope, aspect, exposition, potential direct incedent radiation, distance to the forest edge, tree type, and species heterogeneity) have been resampled to fit the NDWI change maps for Switzerland. For the analysis, the range of each variable has been divided into ten sections and an additional division into ten biogeographic regions in Switzerland leads to 100 strata (sample size). For each stratum, the mean of the environmental variable has been calculated and the total number of forest pixels, as well as the numbers of forest pixels with strong positive and negative changes for the resistance, recovery, and resilience measure, have been counted. This data can be found in the "JS_DataTable_stratified.xlsx" file.

Furthermore, to assess the effect and the interactions of the most significant environmental variables we aggregated the data for five chosen environmental variables (elevation, aspect, exposition, distance to forest edge, and tree type) using three intervals and the ten biogeographic regions to a multidimensional table with 2430 strata (2108 non-empty). For each stratum, the mean for each environmental variable has been calculated and the total number of forest pixels counted. Additionally for each measure, resistance, recovery, and resilience, the mean and standard deviation have been calculated and the number of forest pixels with either positive or negative change of 5%, 10%, and 15% has been counted. This data can be found in the "JS_DataTable_grouped.xlsx" file. 

Usage Notes

Full legends explaining the variable names can be found in the second tab of the excel files.