Monitoring Individual Tree Phenology in a Multi-Species Forest using High Resolution Images

Organised by Laboratory of Geo-information Science and Remote Sensing

Tue 12 April 2022 10:30 to 11:00

Venue Gaia, building number 101
Room 2

By Jasper Kleinsmann

Monitoring tree phenology is important to understand ecosystem functioning and to assess ecosystem responses to the changing climate. Satellite imagery offers an open-access global coverage but is restricted to forest-level analyses due to its coarse spatial resolution. Unmanned Aerial Vehicle (UAV) imagery has the potential to breach this gap by utilising the centimetre-scale resolution for individual tree phenology monitoring. This study has acquired a multispectral UAV time series to evaluate the ability of the Normalised Difference Vegetation Index (NDVI), Enhanced Vegetation Index 2 (EVI2), Optimised Soil-Adjusted Vegetation Index (OSAVI) and Chlorophyll Index Red-Edge (CIRE) to model the seasonal phenology in a diverse forest near Wageningen. A double logistic model was fitted on the vegetation index (VI) observations for each individual tree to derive the start of season (SOS) and end of season (EOS). Individual tree crowns were delineated automatically by a marker-controlled watershed segmentation (MCWS), where the treetops were identified using a Local Maximum Filter (LMF). Automatic segmentation performed best in single-species areas, while it underperformed in complex mixed tree areas. All VIs captured a strong seasonal signal for the deciduous trees and accurately derived the deciduous SOS and EOS consistent with literature and ground observations. General phenological patterns included an early silver birch SOS, quick beech budburst and large within-species phenophase variations for oak trees. On the contrary, seasonal VI variation for coniferous trees was limited, and the resulting phenophase estimates proved uncertain. Overall, phenophase estimates proved most accurate from the NDVI time series. In conclusion, these findings emphasise the capabilities of UAV imagery for individual tree crown phenology monitoring. However, they also show the difficulty of monitoring coniferous phenology by commonly-used VIs and stress the need for further investigation.

keywords: UAV; phenology; individual tree crown; automatic segmentation; multispectral; high resolution