| Recent developments in hydrological modeling and biomass retrieval in complex mountain areas have heightened the need for accurate precipitation data at high spatial resolution. The Tropical Rainfall Measuring Mission (TRMM) provides rainfall estimates for a certain number of climate models in mountain ranges where rain gauges are lacking. TRMM precipitation estimates, however, inherently have large uncertainties because of its coarse spatial resolution of 0.25°. And the TRMM 3B43 precipitation product only contains the precipitation data from 1998. Moreover, for the complex mountainous terrain (ex. the Tibetan Plateau), not only ground observation stations are scarce but also satellite data is not so good.In this study, first, we investigate a statistical downscaling-calibration procedure to derive the high spatial resolution maps of precipitation over the Tibetan Plateau using satellite-based dataset Enhanced Vegetation Index (EVI) from the Moderate Resolution Imaging Spectroradiometer (MODIS), the Digital Elevation Model (DEM) from the Shuttle Radar Topography Mission (SRTM), and the TRMM 3B43 product. The spatial downscaling from 0.25° to 1 km was achieved by using the non-parametric statistic relationship between precipitation and EVI, altitude, slope, aspect, latitude, and longitude. Second, we investigate the extent to which Tropical Rainfall Measurement Mission (TRMM) precipitation and Advanced Very High Resolution Radiometer (AVHRR) Normalized Distribution Vegetation Index (NDVI) product can be used to map historical precipitation at high spatial resolution with a Random Forest algorithm and calibration method. The regression method is mainly based on the relation between the annual TRMM precipitation and AVHRR NDVI of the coarse resolution; the geographical differential analysis (GDA) method corrects the gap between pixel precipitation and in-situ measurements. A dataset of annual precipitation over the Tibetan Plateau is being constructed with the integrated method at 10 km spatial resolution for a 30-year period from 1982 to the present. Third, we get the spatial and temporal variation of precipitation over the Tibetan Plateau by analyzing all precipitation data acquired before.Conclusions of this paper are as follows:(1) The quality of TRMM 3B43 precipitation product over the Tibetan Plateau is not so good. Similar to other mountain areas, the TRMM 3B43 precipitation product over the Tibetan Plateau underestimated the low precipitation and overestimated the high precipitation. So the precipitation product needs to be calibrated.(2) The 1 km annual precipitation data for 2001-2012 over the Tibetan Plateau were generated through downscaling and addictive calibration for most cases. The results show the method improves the accuracy of rainfall estimates. The 1 km monthly precipitation was also obtained by disaggregating 1 km annual un-calibrated downscaling estimates with monthly fraction function. The monthly precipitation prediction is in good agreement with rain gauge data (the best one:R2=0.77, RMSE=27mm, MAE=15 mm, Bias=0.16). The calibration of the monthly product with rain gauge data significantly reduces the Bias value from 0.16 to 0.04. Overall, we conclude that the methodology is useful for areas characterized by varied climate condition and complex topography.(3) A dataset of annual precipitation over the Tibetan Plateau is being constructed with the integrated method at 1 km spatial resolution for a 30-year period from 1982 to the present. This reconstructed dataset shows good agreement with the rain gauge data over the research region for the years 1982-1997. This research has practical implications for researches who are interested in mapping forest aboveground biomass, watershed planning, and stream flow simulation.(4) According to a brief analysis of the temporal and spatial variation of precipitation over the Tibetan Plateau, the results show that the precipitation over the Tibetan Plateau increase from Northwest to Southeast. The climate of Tibetan Plateau is temperate continental climate. And the precipitation concentrated during the summer months, especially in July. |
PDF Full Text Request