Abstract:

The headwaters of the Minjiang River are on the eastern edge of the Tibetan plateau. Canopy rainfall interception plays an important role in the water balance at the regional-scale. Many studies on canopy rainfall interception have been carried out at the stand level but less effort has been devoted towards understanding canopy interception at large scale, neither in the Minjiang River basin nor other areas. In this study, modeling canopy rainfall interception in subalpine forests and meadows in the upper reaches of the Minjiang River was carried out by using field surveys, MODIS data, and RS，GPS and GIS technologies. LAI (leaf area index), vegetation cover and canopy capacity per unit leaf area were the main parameters used in the model. LAI was derived from the vegetation index and measured using a LAI-2000 in the forests and LAI-3000 in the sub-alpine meadows. The LAI of coniferous stands were multiplied by a correction factor because of the clumped arrangement of needles in the crown. Normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) were composed by red, near-infrared and blue reflectances from the 500 m 32-day composites available from the MODIS level 3 surface reflectance (MOD09A1). The results indicated that LAI was non-linearly correlated to NDVI and EVI. EVI was preferable to NDVI as NDVI saturates in well-vegetated areas and the degree of correlation between LAI and EVI is higher than that between LAI and NDVI. The results showed that the LAI of vegetation in the upper reaches of the Minjiang River were in the following categories: 28.57% between 0 and 2, 63.06% between 2 and 4.5, and 8.37% above 4.5. LAI was estimated using EVI, and the results showed that LAI could better reflect the spatial distribution of the vegetation. LAI in the upper watershed was lower than down river due to a large number of trees in the down river. Vegetation cover was derived from NDVI. The spatial distribution of canopy capacity per unit leaf area was modeled on the basis of a vegetation-classification map (1∶1000000). Canopy rainfall interception in the well-vegetated areas was higher than that in other areas. The model was validated using field measurements made in Wolong and Miyaluo and some additional sites in the upper watershed of the Minjang River. Empirical expressions to describe evaporation from the wet canopy were derived from additional sites and evaporation from the wet canopy was closely correlated to rainfall. Based on the empirical expressions, simulation results showed that there was a 15.4 percent error in Wolong and a 19.4 percent error in Miyaluo.