Rice evapotranspiration at the field and canopy scales under water-saving irrigation

Abstract

Evapotranspiration (ET) is an important process of land surface water and thermal cycling, with large temporal and spatial variability. To reveal the effect of water-saving irrigation (WSI) on rice ET at different spatial scales and understand the cross spatial scale difference, rice ET under WSI condition at canopy (ETCML) and field scale (ETEC) were measured simultaneously by mini-lysimeter and eddy covariance (EC) in the rice season of 2014. To overcome the shortage of energy balance deficit by EC system, and evaluate the influence of energy balance closure degree on ETEC, ETEC was corrected as $${\text{ET}}_{\text{EC}}^{*}$$ by energy balance closure correction according to the evaporative fraction. Seasonal average daily ETEC, $${\text{ET}}_{\text{EC}}^{*}$$ and ETCML of rice under WSI practice were estimated as 3.12, 4.03 and 4.35 mm day−1, smaller than the values reported in flooded paddy fields. Daily ETEC, $${\text{ET}}_{\text{EC}}^{*}$$ and ETCML varied in a similar trends and reached the maximum in late tillering, then decreased along with the crop growth in late season. The value of ETEC was much lower than ETCML, and was frequently 1–2 h lagged behind ETCML. It indicated that the energy balance deficit resulted in underestimation of crop ET by EC system. The corrected value of $${\text{ET}}_{\text{EC}}^{*}$$ matched ETCML much better than ETEC, with a smaller RMSE (0.086 mm h−1) and higher R 2 (0.843) and IOA (0.961). The time lapse between $${\text{ET}}_{\text{EC}}^{*}$$ and ETCML was mostly shortened to less than 0.5 h. The multiple stepwise regression analysis indicated that net radiation (R n) is the dominant factor for rice ET, and soil moisture (θ) is another significant factor (p < 0.01) in WSI rice fields. The difference between ETCML and $${\text{ET}}_{\text{EC}}^{*}$$ ( $${\text{ET}}_{\text{CML}} - {\text{ET}}_{\text{EC}}^{*}$$ ) were significantly (p < 0.05) correlated with R n, air temperature (T a), and air vapor pressure deficit (D), and its partial correlation coefficients to R n and T a were slightly greater than D. Thus, R n, T a and D are important variables for understanding the spatial scale effect of rice ET in WSI fields, and for its cross scale conversion.