Abstract: With the increasing of global change, the change of surface heat flux is becoming more and more attention. But the quality and uncertainty of flux data is seriously affected by the spatial representativeness. Therefore, the paper selected the data during the growing season from four observation stations in northwestern China. Then, the paper calculated the flux source area and the point of maximum flux contribution for each site by FSAM model. In this paper, the main source areas(P=80%, 60%, 40%, 20%)was given for each site under the different condition of atmosphere(Stable & Unstable). It turned out that each station under the condition of stable stratification, the source area are as follows:Yingke-station(190 m, 110 m, 80 m, 60 m); Arou-station(140 m, 115 m, 85 m, 65 m); Guantan-stati-on(580 m, 320 m, 240 m, 180 m); Dunhuang-station(280 m, 140 m, 100 m, 80 m). Under the condition of unstable stratification, the source area of the four stations are as follows:Yingkestation(90 m, 55 m, 35 m, 30 m); Arou-station(120 m, 95 m, 80 m, 65 m); Guantan-station(480 m, 280 m, 200 m, 160 m); Dunhuang-station(140 m, 80 m, 60 m, 40 m). From the results, the source area of Guantan-station is the largest in these three stations, because its underlying surface is the forest ecological system and its instrument installation location also is the highest of all. The source area of Yingke-station is not only the minimum but also in a flat elliptical shape. That is the place of the station was surrounded by windbreaks and had a strong disturbance of human activity, so that the turbulence movement for atmospheric is more complex. As a consequence, the direction of the wind is not always stable but very easy to form a strong side wind, especially under the condition of instability. For Arou-station, by the reason of its lowest instrument installation location and the homogeneous underlying surface, the change of source area is more uniform than the other three stations. The source area of Dunhuang-station is between Guantan-station and Arou-station. Thus, it is concluded that the changes of the flux source area were preliminarily analyzed with the height of the instrument and the different underlying surface, etc. Obviously, the erection height of the instrument had a direct influence on the extent of the flux source and different underlying surfaces form different turbulence and thus indirectly affected the distribution of flux source area. In the present paper, we also calculated the point of maximum flux contribution under the different atmospheric stratification for each site(Yingke 20.7-39.8 m; Arou 44.8-49.6 m; Guantan 109.3-124.7 m; Dunhuang 31.7-49.3 m). Its length is consistent with its source area size. In addition, in combination with characteristics of underlying surface for each station, the major source area of the observation is located in the area where we interested and the observation data of the four stations can represent the underlaying surface properties. It is able to provide the reliable basic data for the research that the matter and energy exchange in continental river basin in northwest China. There is some errors when calculating the tall vegetation due to the model itself. Because it is a simplified analytical model, so it can not describe the turbulence conditions inside of the canopy.

Key words: eddy covariance, footprint analysis, FSAM, northwestern China