Vegetation response to rainfall pulses in the Sonoran Desert as modelled through remotely sensed imageries

摘要

In northwestern Mexico in an annual cycle comprised months 2008-2009, we estimated the effects of cumulative rainfall and previous vegetation status on current vegetative greenness adjusted for soil reflectance [via the Transformed Soil Adjusted Vegetation Index (TSAVI)] and canopy water content [via the Normalized Difference Infrared Index (NDII)]. Sample kernels (540 of <0.4 ha) were stratified in nine areas by latitude, distance to the Pacific Ocean and slope across a 15 000 km(2) area on the Baja California peninsula. Rainfall data were accumulated over 2-, 4- and 6-week periods from daily satellite estimates [via the Tropical Rainfall Measuring Mission (TRMM)]. TSAVI and NDII were calculated from Landsat 5 TM images. When restricted by season, multiple regressions results were 0.59 r(2)(adj) < 0.96 for TSAVI and 0.45 < r(adj)(2) < 0.85 for NDII. Rainfall alone accounted for up to an exceptional 70% of the variance in vegetation indices, while the most widely significant factors were the previous vegetation status and its interaction with precipitation. For TSAVI, direct dependence on precipitation was generally much greater for the 4- and 6-week accumulation than for the 2-week periods. However, none of the factors was universally significant for either vegetation index. Results also showed the importance of scale in the spatial heterogeneity of climate and topography in this arid landscape.