ASOADeK mapping of long-term average monthly precipitation over Canadian Rocky Mountains area using a moving-frame approach

Precipitation estimate in mountainous terrain is difficult due to the geographic complexity. ASOADeK (Auto-Searched Orographic and Atmospheric Effects Detrended Kriging) is a recently developed geostatistical model particularly for precipitation mapping in mountainous areas. The model has two components: multivariate regression and de-trended residual kriging. The regression auto-searches some regional and local climatic settings (i.e. the spatial gradient in atmospheric moisture distribution and the effective moisture flux direction), which are useful for precipitation mapping. ASOADeK only requires point precipitation measurements and DEM (Digital Elevation Model).

Previous applications of the ASOADeK model were performed on relatively small areas (with 300 km in length and width). In this study, ASOADeK is applied on a much larger and complicated mountainous area, with unevenly distributed gauge stations, using a moving frame approach for the regression. In this approach, for each target point (pixel), the ASOADeK regression is performed in a frame centered at the target point. Hence, the regression model changes with each target point. The optimum regression was achieved by auto searching the regional and local climatic settings within the frame. Unreasonable regressions were eliminated at the auto searching step. The ordinary kriging is applied with the residuals at the gauge points (gauge observations - regression estimates).The size of moving frame is determined by a minimum area and a minimum number of gauges in the frame.

The new approach gives a better understanding of the mountainous area precipitation comparing to other traditional approaches.