Research On The Fusion Technology Of Measured Meteorological Data And CMADS Dataset In Small And Medium-Sized Watersheds In Mountainous Areas

As an important replenishment method of Earth’s surface water source,precipitation is an indispensable hydrometeorological variable in the hydrological cycle.It is not only the basic output variable of atmospheric process,but also an important driving factor of hydrological change process.Therefore,the accuracy of precipitation data has an important influence on the accuracy of hydrological forecast.The accuracy of precipitation data plays a decisive role in the simulation results of watershed hydrological models.However,due to the influence of terrain,landform,sea and land locations,atmosphere and other factors,precipitation data are discontinuous in space and time.Therefore,how to obtain high-quality precipitation data becomes the key to accurate hydrological simulation and scientific prevention of flood disasters.Based on this problem,taking four typical mountainous small and medium-sized watersheds in Banan District,Chongqing as the study area,and taking the measured precipitation data of the ground rainfall station and the precipitation data of the CMADS dataset as the research object,the accuracy evaluation of the precipitation data of the CMADS dataset and its integration with the measured precipitation data of the ground rainfall station were carried out,and the main research contents and achievements were as follows:(1)Aiming at the accuracy evaluation of precipitation data of CMADS dataset,six evaluation indicators(including correlation coefficient,root-mean-square error,relative deviation,detectivity,false alarm rate and key success index)were used to evaluate and analyze it in time-scale.And analysis and evaluation were carried out at spatial-scale.The results show that from the perspective of consistency between the precipitation data of the CAMDS dataset and the measured precipitation data of the surface rainfall station,the correlation between them at the day-scale and the monthly-scale is relatively stable,but the precipitation data of the CMADS dataset is underestimated compared with the measured precipitation data of the surface rainfall station at both time scales.From the perspective of the ability of precipitation data of CAMDS dataset to monitor precipitation events,the precipitation data of CMADS dataset can detect precipitation events correctly to a large extent,and the false alarm probability of precipitation events is small.Judging from the spatial interpolation results of the correlation coefficient,root mean square error,and relative deviation at the daily scale,the correlation between the two data is high in most areas of the study area,the error is relatively stable,and the degree of deviation is also small(2)The measured daily precipitation data of the rainfall station and the daily precipitation data of the CMADS dataset in the study area were fused by three methods:geographically weighted regression,geographically weighted regression kriging and geographical difference analysis.Six relevant indicators were used to evaluate the fused data,and its fusion effect on spatial-scale was evaluated.The results show that from the perspective of consistency between precipitation data after fusion and measured precipitation data of surface rainfall stations,the consistency between the fused data of geographically weighted regression(GWR)and the measured precipitation data of surface rainfall stations is not as good as that of precipitation data after fusion using geographically weighted regression kriging method(GWRK)and geographic difference analysis method(GDA).From the perspective of relevant evaluation indicators,the relevant indicators of GDA fusion method are slightly better than those of GWRK fusion method.From the spatial-scale analysis results,the consistency between the precipitation data after the fusion of the geographically weighted regression method(GWR)and the measured precipitation data of the ground rainfall station is not as good as the precipitation data after the fusion of the geographically weighted regression kriging method(GWRK)and the geographic difference analysis method(GDA),and the data after the fusion of GDA fusion method and GWRK fusion method are not much different in high values,but the GDA fusion precipitation data is closer to the ground measured precipitation data value at low values.Therefore,the GDA fusion method is slightly better than the GWRK fusion method.Based on the evaluation of indicators and spatial-scale distribution,the geographical difference analysis method of the three methods has a better effect in this fusion application.(3)In order to further verify the applicability of precipitation data after fusion,Different SWAT models which respectively based on the three kinds of data after fusion,the ground actual precipitation data before fusion and the original precipitation data of CMADS was constructed for runoff simulation.and the fusion effect is reverified through the runoff simulation results.The results show that although the simulation results of CMADS precipitation data are not as good as the simulated results of the measured precipitation data of the ground rainfall station during the simulation period,the simulation results in the rate regular and verification periods are also within the acceptable range,and there is still a certain applicability in the study area,and the CMADS precipitation data can be used as an alternative precipitation data for areas with no data or areas with little data.Among the simulation results of the three fusion precipitation data in the simulation period,the simulation results of the three fusion precipitation data were improved to different degrees than the simulation results of the first two types of precipitation data,among which the simulation results of GDA precipitation data improved more obviously,followed by GWRK precipitation data,and the GWR precipitation data simulation results were improved in the coefficient of determination compared with the first two types of data fused,but its Nash coefficient was not as good as the simulation results of rainfall station precipitation data.The precipitation data fused by the geographical difference analysis method can effectively improve the effect of runoff simulation in the actual runoff simulation process,which indicates that the method has good applicability in this fusion application research.