Interannual variation of transpiration and its modeling of a larch plantation in semiarid Northwest China

Abstract

Quantifying the variation of forest transpiration (T) is important not only for understanding the water and energy budget of forest ecosystems but also for the prediction, evaluation, and management of hydrological effects as well as many other ecosystem services of forests under the changes of climate, vegetation, and anthropological impacts. The accurate prediction of T, a key component of water used by forests, requires mechanism-based models describing the T response to environmental and canopy conditions. The daily T of a larch (Larix principis-rupprechtti) plantation was measured through monitoring the sap flow in the growing season (from May to September) of a dry year (2010), a normal year (2012), and a wet year (2014) at a shady slope in the semi-arid area of Liupan Mountains in northwest China. Meanwhile, the meteorological conditions, soil moisture, and forest canopy leaf area index (LAI) were monitored. To get a simple and easily applicable T model, the numerous influencing parameters were grouped into three factors: the atmospheric evapotranspiration demand indicated by the potential evapotranspiration (PET), the soil water supply ability indicated by the relative extractable soil water content (REW), and the vegetation transpiration capacity indicated by the forest canopy LAI. The T model was established as a continuous multiplication of the T response equations to individual factors, which were determined using the upper boundary lines of measured data. The effect of each factor on the T in a dry year (2010) or normal year (2012) was assessed by comparing the measured T in the baseline of the wet year (2014) and the model predicted T, which was calculated through inputting the actual data of the factor (i.e., PET) to be assessed in the dry or normal year and the measured data of other two factors (i.e., REW, LAI) in the baseline of the wet year. The results showed that the mean daily T was 0.92, 1.05, and 1.02 mm; and the maximum daily T was 1.78, 1.92, and 1.89 mm in 2010, 2012, and 2014, respectively. The T response follows a parabolic equation to PET, but a saturated exponential equation to REW and LAI. The T model parameters were calibrated using measured data in 2010 and 2012 (R2 = 0.89, Nash coefficient = 0.88) and validated using measured data in 2014 satisfactorily (R2 = 0.89, Nash coefficient = 0.79). It showed a T limitation in the dry year 2010 for all factors (18.5 mm by PET, 11.5 mm by REW, and 17.8 mm by LAI); while a promotion for PET (1.4 mm) and a limitation for REW (4.2 mm) and LAI (14.3 mm) in the normal year 2012. The daily T model established in this study can be helpful to assess the individual factor impact on T and improve the daily T prediction under changing environmental and canopy conditions.