Using Sentinel 2 to analyze the spatial distribution of damage after the 2018 summer drought

Frank Weiser1, Carl Beierkuhnlein1
1 Biogeography,

O 1.1 in Plant and soil responses to drought: Interdisciplinary approaches and views

10.10.2019, 11:15-11:30, H36, NW III

Introduction 

The Summer Drought of 2018 caused extensive damage to German forests. After the dry summer, the damaged trees were especially susceptible to damage from e.g. storms or bark beetle outbreaks. June and July 2019 were again very hot and dry, adding further pressure to the already stressed forest ecosystems.This combination of factors caused significant damage.

Remote Sensing products have been used extensively to monitor the reaction of vegetation to drought. Many frequently used data sets have either low spatial resolution (e.g. MODIS) or rather long revisit times. The new Sentinel 2 constellation from the EU Copernicus program offers a high 10m spatial resolution with a five-day revisit time. The datasets are therefore well suited to monitor drought impact with a high spatial and temporal resolution.

The goal of this study is to use Sentinel 2 data to analyze the spatial extent and distribution of damage to German forests after the 2018 drought.

Material and Methods

820 Sentinel 2 images covering all of Germany were preprocessed. The Normalized Difference Vegetation Index (NDVI), a good measure for vegetation condition, was calculated for all scenes. Three mosaics (May 2018, September 2018, May 2019) were created to examine the decline in NDVI due to drought damage. Auxiliary data sets, e.g. an elevation model or soil datasets were used to analyze spatial patterns.

Results

Broadleaved forests under drought stress showed a quick decline in NDVI values. First analysis indicate distance to forest edge as the factor most important for suffered drought damage, with trees at the forest edge more susceptible.

Conclusions

Sentinel 2 is well suited to detect drought damage. The spatial resolution allows for good detection of damaged tree groups. Using images from two time steps (before and after drought) includes errors due to phenology effects, which will most likely be solved by a longer time series of images.




Keywords: remote sensing, forest, drought, Sentinel 2

Export as iCal: Export iCal