Abstract. Information on the exchange of energy, momentum and mass (H₂O, CO₂, CH₄, etc.) over complex topography is critical for determining the development of the boundary layer, carbon and water cycles, weather and climate. This information can also improve the numerical modelling of physical atmosphere-land processes. Based on a 12-year (2007–2018) eddy covariance dataset over the Dali agricultural field in the southeastern Tibetan Plateau, we analysed the diurnal, seasonal and inter-annual changes in sensible heat flux (H_s), latent heat flux (LE) and CO₂ flux (F_c) and their meteorological controls on multiple timescales (half-hourly, daily, monthly, and yearly). The results show that both H_s and LE have similar diurnal and seasonal variations, but the amplitude of LE is obviously larger than that of H_s throughout the year, which indicates that the LE plays a dominant role in surface heat exchange. The F_c has a noticeable diurnal cycle, reaching its minimum around noon, and clear seasonal variations, reaching its minimum in the summer. The annual average H_s increased from approximately 6 W m⁻² during 2007–2012 to 19 W m⁻² during 2013–2018, while the LE decreased from approximately 110 W m⁻² during 2007–2013 to 79 W m⁻² during 2014–2018. The Dali observational area is a carbon sink in all years, while the magnitude of net uptake decreases significantly from approximately 739 g C m⁻² yr⁻¹ during 2007–2013 to 218 g C m⁻² yr⁻¹ during 2014–2018. The results also show that wind speed (WS) is the major control of H_s, while the product of WS and vapour pressure deficit (VPD) is the main driver of LE on different timescales. The net radiation (R_n) and soil temperature (T_s) have the largest effects on F_c from the daily to monthly timescales, while the WS has the largest impact on annual total F_c.