Potential of Landsat spectral indices in regard to the detection of forest health changes due to drought effects
2019
Martin Hais | Kateřina Neudertová Hellebrandová | Vít Šrámek
Because of climatic variability that has been increasing in last decades a higher drought risk seriously influences the forest vitality from regional to global scale. Despite there are of many studies that describe the spectral response of forest stands to the water stress, there is still a lack of information concerning the full understanding of forest reaction to the water deficiency over a longer time period. We hypothesize that the various severity and/or frequency of drought periods will result in different spectral responses of forest stands. The forest response was detected using two spectral vegetation indices (normalized difference moisture index - NDMI, wetness) which are widely used for the detection of forest health changes. These indices were calculated on the basis of Landsat (TM, ETM+ and OLI) imagery which includes 105 scenes from the 2005-2016 period. The area of our interest includes 300 forest stands (dominated with Norway spruce) in the Czech Republic, Moravia. These stands were identified as damaged by drought that occurred during the 2012-2017 period. To document the climatic water deficiency, two climatic indices were calculated (AWBPE, standardized precipitation evapotranspiration index). Despite high correlation of both spectral indices, the NDMI has high sensitivity to the drought events. However, both indices significantly decreased in reaction to the drought events. In case of the 2012 drought event the decrease was one year delayed, probably due to the lower severity of drought effect. The both groups of spectral and climatic indices bring valuable information in regard to the description and understanding of drought effect on the spruce forest stands.
Mostrar más [+] Menos [-]Palabras clave de AGROVOC
Información bibliográfica
Este registro bibliográfico ha sido proporcionado por Directory of Open Access Journals