Abstract:
Canopy stomatal conductance indicates the intensity of photosynthesis and respiration of a plant, particularly for the evapotranspiration, agricultural production and ecosystem function. The estimated accuracy of canopy stomatal conductance varies greatly among different vegetation types. However, the influence of vegetation types on canopy stomatal conductance has rarely been considered in most previous studies. This study aims to develop a canopy stomatal conductance model considering vegetation types, thereby to achieve higher accuracy than before. According to a widely-used model of canopy stomatal conductance proposed by Leuning in 2008, two improvements were focused mainly on: 1) Stomatal length and density were introduced into the new model to calculate the maximum stomatal conductance, which differed among vegetation types but usually been calculated via parameter optimizing by minimizing the cost function. 2) Satellite-derived photosynthetically active radiation data was used to calculate the flux density of visible radiation at the top of canopy, instead of the net absorbed radiation minus the soil heat flux in previous studies, where the flux can contain the absorption by non-vegetation component. Ground observations were used to validate the model via the inversion of Penman-Monteith equation, including eddy covariance and automatic meteorological data from 15 sites in the middle reaches of Heihe Basin (100°20′-100°24′ E, 37°45′-42°40′ N) in China. An attempt was made to partition the evapotranspiration into soil evaporation and plant transpiration, where only plant transpiration was selected to derive canopy stomatal conductance, ruling out the influence of soil evaporation. Evaluation indexes included the determination coefficient, Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). The results were as follows: 1) The improved model achieved higher accuracy in the land cover of maize and bell pepper, where the ground observed data was used to validate the models before and after partitioning soil evaporation and plant transpiration by the inversion of Penman-Monteith equation. Specifically, the determination coefficient of maize improved from 0.49 to 0.68 after improvement, while the RMSE reduced from 3.99 to 2.67 mm/s, and the MAE reduced from 3.34 to 2.06 mm/s, after partitioning soil evaporation and plant transpiration. Before that, the determination coefficient of maize improved from 0.38 to 0.46 after improvement, with both RMSE and MAE reduced. 2) The new improved model was applied to the middle reaches of Heihe Basin from July to September 2012 using HJ-1 satellite data, including the leaf area index, photosynthetically active radiation, and land cover data. The model showed a good ability in estimating the canopy stomatal conductance of different vegetation types. The average value for the canopy conductance of maize on July 27 was 15.12 mm/s, while that of spruce was 3.84 mm/s. The determination coefficient improved from 0.51 to 0.70, and the RMSE reduced from 4.91 to 3.32 mm/s, comparing with the model before and after improvement. The commission error was calculated to evaluate whether the new model was sensitive to stomatal characteristics, indicating that there was an obvious difference, only when the vegetation was wrongly classified. It demonstrated that the remote sensing data can be used for the better accuracy of new model in various vegetation types. A better improvement was achieved in estimating canopy stomatal conductance using the new model considering vegetation types, compared with the traditional models without considering that. The finding can bring great convenience to scientific research, such as surface energy balance and carbon-water cycle, as well as the specific practical applications in precision agriculture, water resources management, and ecology.