Spatial heterogeneity poses a major challenge for the appropriate interpretation of eddy covariance data. The quantification of footprint climatology is fundamental to improving our understanding of carbon budgets, assessing the quality of eddy covariance data, and upscaling the representativeness of a tower flux to regional or global scales. In this study, we elucidated the seasonal variation of flux footprint climatologies and the major factors that influence them using the analytical FSAM (Flux Source Area Model), KM (Kormann and Meixner, 2001), and H (Hsieh et al., 2000) models based on eddy covariance measurements at two and three times the canopy height at the Qianyanzhou site of ChinaFLUX in 2003. The differences in footprints among the three models resulted from different underlying theories used to construct the models. A comparison demonstrated that atmospheric stability was the main factor leading to differences among the three models. In neutral and stable conditions, the KM and FSAM values agreed with each other, but they were both lower than the H values. In unstable conditions, the agreement among the three models for rough surfaces was better than that for smooth surfaces, and the models showed greater agreement for a low measurement height than for a high measurement height. The seasonal flux footprint climatologies were asymmetrically distributed around the tower and corresponded well to the prevailing wind direction, which was north-northwest in winter and south-southeast in summer. The average sizes of the 90% flux footprint climatologies were 0.36-0.74 and 1.5-3.2 km 2 at altitudes of two and three times the canopy height, respectively. The average sizes were ranked by season as follows: spring > summer > winter > autumn. The footprint climatology depended more on atmospheric stability on daily scale than on seasonal scale, and it increased with the increasing standard deviation of the lateral wind fluctuations.