Comparison And Modification Of Spatial Interpolation Accuracy Of Precipitation With Few Sample Points

Precipitation is one of the main sources of water resources for people's production and living.The precipitation data has important application value in the management of production and water resources with agricultural,the farmland irrigation,the analysis of drought and flood disaster,the hydrology analysis,etc.And it is one of the important data in the process of simulating rainfall runoff.Therefore,it is of great significance to obtain the complete precipitation data.Sometimes the precipitation data is incomplete and not widely covered in the region due to terrain,personnel,technology and funds,etc.To get precipitation data or missing data in the region where there is no observation station.The general method is using space interpolation methods to supplement the precipitation data in this region.In order to study interpolation accuracy of precipitation data by spatial interpolation methods in which the number of observation points of precipitation is few,the monthly precipitation data of the Upper Sangamon river in Illinois,USA and the Bei'an Section of Wuyuer river in Heilongjiang Province China were selected,and the following research work were performed:(1)Probability distribution function,and distribution of temporal and spatial of precipitation data in the two regions were analyzed,and the relationship between precipitation and elevation,longitude and latitude was analyzed by using the scattered plot,individually.In Upper Sangamon river,the precipitation data conform to Extreme I probability distribution,and the precipitation data disperse slightly with time and space.The global spatial auto-correlation of the data is close,and the relationship between precipitation and elevation,longitudes and latitude is weak.In Bei'an Section of the Wuyuer river,the precipitation data not conform to Extreme I probability distribution,and the precipitation data disperse clearly with time and space.The global spatial auto-correlation of the data is estrange,and the between precipitation and elevation,longitudes and latitude is strong.The characteristics of the precipitation data in the two regions are quite different,which provides a comparatively well sample material for the subsequent analysis of the precision of interpolation in precipitation data by different spatial interpolation methods,and it provides the basis for verifying the adaptability of different spatial interpolation methods and correction methods in different regions.(2)Four spatial interpolation methods Inverse Distance Weight(IDW),Ordinary Kriging(OK),Spline and Trend were selected,and the monthly precipitation data was interpolated by the four spatial interpolation methods in the region.The interpolated results and the test data were compared,and the interpolation accuracy of the four methods was analyzed for interpolation and extrapolation points.The IDW is more accurate than the other three interpolation methods.Then the influence of different number of observation points on interpolation accuracy of different method were analyzed by using different observation points which were reduced according to displacement to the test point from far to near.The IDW method is less affected by the number of observation points.The OK method comes second,and the Spline method and Trend method are greatly influenced by the number of observation points.The interpolation accuracy of the two research regions was compared.The mean of mean relative error is 9%~23% in the Upper Sangamon river,and the mean of mean relative error is23%~33% in Bei'an section of the Wuyuer river.Because the precipitation data were distributed evenly and disperse slightly with spatial in Upper Sangamon river,and so the spatial interpolation accuracy of the former is higher than that of the latter.(3)IDW was modified by using three parameters of geographical terrain factors elevation,longitude and latitude.And multi-factor IDW was obtained.Then the interpolation accuracy of this method was verified by using different observation points which were reduced according to displacement to the test point from far to near.The results show that the interpolation accuracy of precipitation increases by 0.7%~1.1% with multi-factor IDW in the Upper Sangamon river,and the interpolation accuracy of precipitation on increase by 2.3%~3.9% with multi-factor IDW in the Bei'an Section of the Wuyuer river.The increase rate of interpolation accuracy in Upper Sangamon river is less than Bei'an Section of the Wuyuer river due to the different influence of geographical terrain factors on these two regions.