Predicting monthly precipitation with multivariate regression methods using geographic and topographic information
Multivariate regression models that integrate topographic and geographic information are developed to predict monthly precipitation in the Daqing Mountains of northern China. Five geographic and topographic factors, including longitude, latitude
, elevation, slope, and aspect, are taken into account in the model development. The data are acquired from a 100 m resolution DEM of the national topographic databases. Measured precipitation data at 56 stations between 1955 and 1990 are used for model
development, and a leave-one-out cross-validation method is used for model evaluation. The model explains most of the spatial variability in monthly precipitation, and can also quantify the relative importance of different geographic and topographic variables.
restoration. The results showed that: (1) the topographic wetness index was positively correlated with soil moisture near the surface (0–1 m), but negatively correlated with soil moisture at depths below 2 m; and (2) the negative relationship was found between
biomass and soil moisture content in deep layers. Soil moisture in shallow layers was more likely to be affected by topographic factors. However, it is highlighted that the biomass determines spatial variation of deep soil moisture. Therefore, vegetation