A global prognostic scheme of leaf onset using satellite data

Publisher Website
Google Scholar
Abstract
Summary: Leaf phenology describes the seasonal cycle of leaf functioning. Although it is essential for understanding the interactions between the biosphere, the climate, and biogeochemical cycles, it has received little attention in the modelling community at global scale. This article focuses on the prediction of spatial patterns of the climatological onset date of leaf growth for the decade 1983–93. It examines the possibility of extrapolating existing local models of leaf onset date to the global scale. Climate is the main variable that controls leaf phenology for a given biome at this scale, and satellite observations provide a unique means to study the seasonal cycle of canopies. We combine leaf onset dates retrieved from NOAA/AVHRR satellite NDVI with climate data and the DISCover land‐cover map to identify appropriate models, and determine their new parameters at a 0.5° spatial resolution. We define two main regions: at temperate and high latitudes leaf onset models are mainly dependent on temperature; at low latitudes they are controlled by water availability. Some local leaf onset models are no longer relevant at the global scale making their calibration impossible. Nevertheless, we define our unified model by retaining the model that best reproduced the spatial distribution of leaf onset dates for each biome. The main spatial patterns of leaf onset date are well simulated, such as the Sahelian gradient due to aridity and the high latitude gradient due to frost. At temperate and high latitudes, simulated onset dates are in good agreement with climatological observations; 62% of treated grid‐cells have a simulated leaf onset date within 10 days of the satellite observed onset date (which is also the temporal resolution of the NDVI data). In tropical areas, the subgrid heterogeneity of the phenology is larger and our model's predictive power is diminished. The difficulties encountered in the tropics are due to the ambiguity of the satellite signal interpretation and the low reliability of rainfall and soil moisture fields.