Simulated total evapotranspiration of winter wheat with modified Shuttleworth-Wallace model in different stages in Nanjing

Abstract: Winter wheat is an important food crop in Nanjing. In order to investigate the water management and increase the yield, it is crucial to investigate the precise water consumption of winter wheat and its influencing factors. Using the data of measured evapotranspiration (ET) with weighing lysimeter and meteorology factors from automatic weather station during 2011-2012 and 2013-2014, we analyzed the seasonal variation of winter wheat ET and simulated the ET using Penman-Monteith (PM) and Shuttleworth-Wallace (SW) model in 4 different periods from sowing to harvest in the experiment site of Nanjing University of Information Science & Technology (118.8 °E, 32.0 °N, with an altitude of 32 m). The site belonged to subtropical monsoon climate, with an average annual temperature of 15.6 ℃ and an average annual rainfall of 1 106 mm. The soil was yellow brunisolic soil with a field capacity of 0.30 cm3/cm3. The weighing lysimeter had an effective evapotranspiration area of 4 m2, with an accuracy of 0.1±0.01 mm. Jarvis model was modified to calculate the canopy resistance which was considered as the most sensitive variable in PM and SW models. In addition, we discussed the relationships between ET and environmental factors. The results showed that winter wheat ET increased gradually to 0.25 mm/h in the sowing and tillering stage, kept at a low level of about 0.20 mm/h in the winter period, and increased rapidly in the regreening and shooting period. The peak ET was 1.37 mm/h in April when the canopy was well developed. The PM model was suitable for calculating the ET in developed canopy cover, but the underestimate of the ET occurred in the beginning of the stage and the overestimate of the ET occurred in the end of the growing season. The analysis of modeling ET in sowing-tillering, tillering completion-regreen stage and jointing-filling stage indicated that the SW model performed well with a minimum stomatal resistance of 70 s/m, however, there was still large errors in milky stage. Lower soil water content occurred in the end of the growing season for winter wheat in Nanjing area, as a result the minimum stomatal resistance increased, which might be the direct reason that caused the obvious error in the ET simulation. Thus, we corrected the minimum stomatal resistance to a high value of 150 s/m for the Jarvis model in milky stage. The comparison between the ET from SW model and the measured ET by lysimeter in milky stage indicated that the modified SW was better than the unmodified SW model, for the modified and unmodified models, the slope was 1.12 and 3.57, the mean absolute error (MAE) was 0.05 and 0.18 mm/h and the agreement index was 0.62 and 0.33, respectively, in 2011-2012 and the MAE was 0.09 and 0.12 mm/h respectively in 2013-2014. The combined SW model obtained better results than each single model when evaluating the seasonal ET because of the optimized canopy resistance in different seasons. The modified SW model increased the determination coefficient (R2) to 0.87 and 0.67, decreased the MAE by 0.02 and 0.01 mm/h and apparently increased the agreement index to 0.82 and 0.70, respectively, in 2011-2012 and 2013-2014. The values of whole season ET were 408 and 453 mm using the modified SW model respectively during 2011-2012 and 2013-2014, and the ratio of soil evaporation to total ET was 28%. ET was mainly controlled by net radiation, vapor pressure deficit, and air temperature and the order of determination coefficient of the environmental factors was net radiation > vapor pressure deficit > air temperature. It was concluded that we could use a larger canopy resistance in SW model in milky stage in order to evaluate the ET during the whole season. The results can improve the evaluation precision of the ET throughout the whole season and be used as a reference for water management of winter wheat in Nanjing.