Abstract: The Karakoram Highway (KKH) is an important part of the China-Pakistan Economic Corridor. However, geological disasters along the route are frequent, posing a serious threat to the normal operation of the highway. In particular, the geological environment of the Gaizi valley in China is more complicated, under the precipitation and induced earthquakes, geological disasters such as collapse, landslide and mudslide occur frequently. This study uses SBAS-InSAR technology combined with field verification to carry out surface deformation monitoring and early identification of geological disasters in the Gaizi valley, the time serious surface deformation information of the Gaizi valley section is obtained, and the annual average deformation rate and cumulative deformation variables of each deformation point are extracted. The 10 km buffer centered on KKH is selected as the research scope, and the deformation rate of the LOS (Line Of Sight) in the study area from 2016 to 2017 ranging from -76 to 28 mm/a is obtained by the SBAS-InSAR method. Combined with the slope and aspect of the study area and the geometrical attitude of the satellite acquisition data, the LOS deformation is converted to the slope direction, and the maximum deformation rate value along the slope direction is -157 mm/a. Based on the surface deformation rates of the slope direction, 449 disaster points covering the study area are obtained by field investigation, including 31 landslides, 416 unstable slopes and 2 glacial movements. Through remote sensing interpretation and field verification, 23 debris flows are identified in the area. By using the TRMM precipitation data to analyze the time series deformation curve, the occurrence of regional landslides and unstable slopes is related to strong precipitation, and the sliding phenomenon lags behind the occurrence of heavy precipitation. Therefore, we should focus on the occurrence of abnormal precipitation. The results provide a scientific basis for early identification and prevention of disasters.