Sentinel-2 time series reveal species-specific responses in temperate conifer dieback

Abstract

Silver fir (Abies alba Mill.), Norway spruce (Picea abies (L.) H. Karst) and Scots pine (Pinus sylvestris L.) tare major European conifers undergoing severe dieback. Large-scale, high-frequency monitoring with multispectral satellites using Sentinel-2 images and ground observations (2018-2023) over the French mountains for 742 pure and 949 mixed stands, we investigated the interest of analysing dieback detection at the species level. We used time series analysis from seven raw bands and eight vegetation indices and integrated spectral anomalies for different periods prior to the ground survey. Results differed among species: silver fir was best detected with a 15-month period and Inverted Red-Edge Chlorophyll Index (IRECI), Norway Spruce with a 1-month period and Continuum removal of the shortwave-infrared spectrum (CRSWIR), and Scots pine with a 4-month period and red-edge (CRRE). In pure stands, dieback detection was most accurate for Norway spruce (R2 = 0.54, OA = 77%) followed by silver fir (R2 = 0.34, OA = 61%) and Scots pine (R2 = 0.16, OA = 64%). Dieback was difficult to detect in its early stages and in mixtures, mainly for Norway spruce and Scots pine. This study highlights the importance of developing species-specific methods for large-area monitoring of dieback.