Deriving plant phenology from remote sensing

Abstract

Plant phenology is the study of the timing of periodic vegetation cycles and their connection to climate. Examples are the date of emergence of leaves and flowers or the date of leaf colouring and fall in deciduous trees. It is an independent measure on how ecosystems are responding to climate change and therefore experiencing renewed interest from the scientific research community. This paper describes a method to derive plant phenology indicators from time series of satellite images. The satellite images are Normalized Difference Vegetation Index (NDVI) images from the MODIS sensor, which encompass the European continent from 2000 onwards. The Harmonic Analysis of NDVI Time Series (HANTS) algorithm is used to process and analyse the time series of satellite images for each individual year. The resulting amplitude and phase values are translated into commonly understandable phenology indicators like start of growing season, which can be linked again to the biological definitions of plant phenology. The indicators are validated with field observations, recorded by a volunteer's network in the Netherlands and Germany. Conclusions are that the method produces consistant maps, which correlate well with the crop type. However, on average the remote sensing derived start of season is 14 days earlier than the observed values.