Modelling tropical dry forest deciduousness using spatially downscaled TRMM data
Increases in the intensity and spatial extent of dry season deciduousness in the tropical dry forests of the Mexican Yucatán may impact biosphere-atmosphere interactions. Issues of data scale affect characterization of the relationship between precipitation and vegetation leaf canopy condition using remotely sensed measurements of precipitation. This paper examines the use of a set of spatial and topographical methods to downscale rainfall data to account for observed differences in total monthly rainfall measurements at weather stations (N=22) and measurements from the Tropical Rainfall Measuring Mission. Each is evaluated by the resulting increase in spatially-averaged coefficient of determination from a per-pixel (0.01 deg.) linear regression model of MODIS EVI and contemporaneous and 1-month-lagged precipitation image time series (2000-2001). Increases in model explanatory power are observed for all downscaling techniques, with ΔR2ranging from 0.024 to 0.046. Results suggest spatial variability of sensitivity to water-scarce conditions within semi-deciduous forests in the area.
International Geoscience and Remote Sensing Symposium (IGARSS)
Cuba, Nicholas and Rogan, John, "Modelling tropical dry forest deciduousness using spatially downscaled TRMM data" (2014). Geography. 643.