基于MODIS影像反演地表瞬时温度场和气象站点30年平均气温实测数据，结合数字高程模型，研究了云南山地气温随海拔变化的规律。结果表明，降低或消除了坡向与坡度、纬度与经度变化的影响后，在云南全区域尺度上，海拔对气温作用而产生的气温直减率为0.53℃；在典型山地地貌类型区，山地气温直减率由高山区向低山区逐渐增大，滇西北高山区直减率为0.47℃，滇中、滇东中山区为0.51℃，滇南宽谷低山区为0.54℃，滇西北高山峡谷和滇南低山深切沟谷地带为0.54℃；由阳坡、半阴半阳坡至阴坡，山地气温直减率依次增大：阳坡直减率平均为0.52℃，半阴半阳坡平均为0.54℃，阴坡平均为0.55℃。除海拔之外，坡向、坡度等微地形因子亦是微格局气温场的主要地理响应因子。利用遥感数据反演的瞬时温度场研究山地气温直减率，是一种研究山地气温变化规律的新方法。

Studies on the patterns of temperature changes with elevation have important implications for constructing the microscale temperature of mountain areas and for practical applications in agriculture, forestry, prevention and control of natural disasters, and ecosystem management. Long-term observation data and instrument data logging based on the ground meteorological network are the most conventional methods for obtaining temperature change patterns along the elevation gradient. However, limited by the number of meteorological stations and insufficient coverage over large geographical regions, the observed and recorded results are easily distorted by accidental factors. This implies that the results may be inadequate for constructing spatial distribution models of land surface temperatures in synchronized time phases for contiguous microscale mountain environments over large regions. Choosing Yunnan Province as the study area, this paper endeavored to detail mountainous temperatures with elevation changes across different spatial scales using the Instantaneous Land Surface Temperature (ILST) and Digital Elevation Model (DEM). First, a raster of ILST was retrieved from thermal infrared bands of Moderate Resolution Imaging Spectroradiometer (MODIS) using the split-window algorithm. Second, the mean temperature data for November from three decades recorded at conventional meteorological stations were used to calibrate the remotely sensed ILST to attain a raster of the temperature. Third, the effects of latitude changes, changes in slope drop, slope aspect, water body and latitudinal variations were eliminated or reduced. Then, the Temperature Lapse Rates (TLRs) with elevation for the entire Yunnan province and for different types of mountain landforms and slope aspects were calculated using random sampling, spatial division and statistical regression. The results include: (1) after reducing or eliminating the effects of aspect, slope, longitude, latitude and other microterrain factors, TLR of elevation was 0.53℃ at the entire regional scale. (2) TLR increased gradually from the high to low mountain area at the local regional scale of typical landforms. TLR was 0.47℃ in the alpine region of northwest Yunnan, 0.51℃ in the middle mountainous region of central and eastern Yunnan, 0.54℃ in the low mountainous region alone the southern border of Yunnan, and 0.54℃ in the alpine canyon region of northwest Yunnan and river valley region of southern Yunnan. (3) TLR increased from sunny to shady slope: 0.52℃ on sunny slope, 0.54℃ on semi-shady slope, and 0.55℃ on shady slope. The zonal patterns in mountain temperature changes, acquired by retrieving the instantaneous land surface temperature based on the thermal infrared bands in MODIS images, complement the constraints of unavailable synchronized data in conventional methods. Through statistical computation of TLRs, the zonal patterns of temperature changes at different spatial scales can be acquired in detail. Elevation is the main factor affecting mountain temperature, and it demonstrates outstanding geographical patterns in relation to temperature. In addition to elevation, other microterrain factors, such as aspect and slope are also affecting mountainous temperatures in micropatterns, and this needs to be further researched. It is a relatively new method to research into the laws of temperature change in the mountain regions with DEM and ILST using remotely sensed data.