A combined model to simulate CO₂ and H₂O gas exchange at the leaf scale was parameterized using data obtained from in situ leaf-scale observations of diurnal and seasonal changes in CO₂ and H₂O gas exchange. The Farquhar et al.-type model of photosynthesis was parameterized by using the Bayesian approach and the Ball et al.-type stomatal conductance model was optimized using the linear least-squares procedure. The results show that the seasonal physiological changes in photosynthetic parameters (e.g., V_cmax25, J_max25, R_d25 and g_m25) in the biochemical model of photosynthesis and m in the stomatal conductance model should be counted in estimating long-term CO₂ and H₂O gas exchange. Overall, the coupled model successfully reproduced the observed response in net assimilation and transpiration rates.

Hyper-arid regions are expected to undergo climatic change, but only a few research works have so far been conducted on the dynamics of carbon dioxide (CO₂) fluxes and their consequent responses to various bioclimatic factors, which is mainly attributable to a limited set of flux observations. In this study, the CO₂ fluxes exchanged between the forest and the atmosphere have been measured continuously by the eddy covariance approach from June 2013 to December 2016 in a riparian forest, which is a primary body of natural oases located within the lower reaches of inland rivers in China. The present results revealed that the climatic conditions characterized by relatively high mean air temperatures (T_a) with fluctuating annual precipitation (P) during the prescribed study periods were comparable to the historical mean value. The annual net ecosystem productivity (NEP) ranged from approximately 278 g C m⁻² year⁻¹ to 427 g C m⁻² year⁻¹, with a mean value of 334 g C m⁻² year⁻¹. The mean annual ecosystem respiration (Re) and the gross primary productivity (GPP) were found to be 558 and 892 g C m⁻² year⁻¹, respectively. The results also ascertained that the high inter-annual variations in NEP were attributable to Re rather than to GPP, and this result was driven primarily by T_a and the groundwater depth under similar eco-physiological processes. In addition, the CO₂ fluxes were also strongly correlated with the soil temperature and photosynthetically active radiation for the present study site. In conclusion, the desert riparian forest is a considerably significant carbon sink, particularly in the hyper-arid regions.