In order to reduce the uncertainty of homogeneity test results and improve the credibility of homogeneity test conclusion, the possible causes of inhomogeneous temperature time series are discussed based on reliable evidences. The monthly average temperature series of 77 stations in Shaanxi are tested by using TwoPhase Regression (TPR), Penalized Maximal 〖WTBX〗F〖WTBZ〗 test (PMFT), Penalized Maximal t test (PMT). The test results are judged based on the detailed station metadata and satellite remote sensing image data, and the causal analysis of the influence on the homogeneity temperature series is discussed. The results show that station metadata can support 78% of the detected inhomogeneous breakpoints using two methods. In addition to 66.7% of inhomogeneity caused by station migration, the replacement of equipment and calculation method change for daily average temperature are the other reasons, accounting for 22.2% and 11.1%, respectively, in the 78% of the inhomogeneous breakpoints supported by station metadata. Using the land use/cover change (LUCC) distribution image data of the phase at the different time and different buffers obtained by remote sensing (high resolution Landsat images) combined with GIS technology, the panorama view and composite diagram of the observation field, a comprehensive analysis is made of the inhomogeneous breakpoints occurred in 1993 and the 8month average temperature at Tongguan influenced by small environment changes around the station. Finally, due to the satellite remote sensing images can fill the limitations in the aspect of station observation environmental assessment of traditional station metadata. It is suggested that satellite remote sensing images can be the supplement of stations metadata to provide a more intuitive and objective evidences for station observation environmental assessment.