Simulating canopy conductance of the Haloxylon ammodendron shrubland in an arid inland river basin of northwest China

Accurate estimates of canopy conductance (g(c)) are essential for quantifying the carbon, water and energy fluxes of ecosystem and understanding the patterns of water utilization of vegetation in arid regions. To this end, g(c) of Haloxylon ammodendron community, dominated by three desert shrub species, was calculated using the inversed Penman-Monteith equation from measurements of sap flow and concurrent microclimate over two main growing seasons (2014-2015). Then, it was used to examine the Jarvis-Stewart (JS) models which comprised different response functions and the simplified process-based model (BTA) to select the best one for our study species and climate. Photosynthetically active radiation and vapor pressure deficit typically covary throughout the day and are known have opposite effects on g(c). When this effects was taken into account, both the JS model and the BTA model produced better g, fittings. Selection of proper vapor pressure deficit function and air temperature function significantly improved the performance of the JS model. The best JS model given a correlation coefficient of 0.89, RMSE of 1.99 mm s(-1) and average percent error of 19% in comparison with the PM-calculated g(c) while the best BTA model outperformed this model, reflected by higher correlation coefficient (0.90), and lower RMSE (1.93 mm s(-1)) and average percent error (9%). The average decoupling coefficient was 0.28, indicating canopies of H. ammodendron community were well coupled from the atmosphere. These findings addressed the importance of selection of stress function and consideration of air temperature for improving g, estimation in arid region, and gain new knowledge on the environmental control on canopy conductance.