Low-resolution remote sensing data are kinds of very important data source to global change research, which is with of high temporal resolution and large coverage.However,multi-source low-resolution data have their own characteristics,such as geographical coverage,data accumulated time,which could not satisfy the demands of global change research by single low-resolution data.Therefore,it is necessary to combine multi-source large-scale low resolution remote sensing data together being mutually complementary to meet its requirement.It is very necessary to make the multi-source remote sensing data be consistent with geo-location firstly,so a series of analysis and experiments of geometric correction was carried out. MOD09A1,as a kind of standard data product with higher accuracy of geo-location,is used as the base data. Relative geometric accuracy evaluation between the base data and L1Bdata of NOAA/AVHRR,FY-3/ VIRR,FY-3/MERSI,FY-2/VISSR was done respectively.The result shows that the difference in geo-location between the base data and the L1Bdata mentioned is significant,which is not good for the combination of them.Meanwhile,multi-order polynomial geometric correction based on sparse and evenly distributed Ground Control Points(GCPs)of the data mentioned was carried out,since the selection of GCPs on low spatial resolution data was difficult.The result indicates that low-order polynomial geometric correction could make a remarkable improvement on geometric accuracy of the multi-source remote sensing data and be consistent in geo-location with the MODIS base data,which would meet the requirement of the combination of multi-source remote sensing data in global change research.