|Other Abstract||The region of Yili is located in the western part of Tianshan Mountains, northwest China. It is known as “ a wet island’ in arid region for abundant precipitation Under the context of climate change, it is thought provoking that what changes the evaporation would exhibit. Based on the meteorological data of the 8 stations and GLDAS Product, we conducted the study on the evaporation in Yili region. The main objevtives of this study are: (1) to test and select a most suitable method from candidate ones to calculate reference evapotranspiration (ETr) for the Yili region. (2) to examine the spatial and temporal patterns of ETr, pan evaporation (ETp) and actual evaporation(ETa). (3) to analyze the responses of ETr to the meteorological factors. (4) to clarify whether there exists Bouchet’s complementary relationship between potential evapotranspiration(ETo) and ETa in the Yili region.
The present study showed that the method of Irmark-Allen can be satisfactorily used to calculate the ETr in the Yili region where the meteorological data lacked. ETr in the Yili region significantly decreased with a range from 758.9mm to 901.9mm and had a abrupt change in the year of 1985. ETp experienced a non-remarkable decreasing trend with a abrupt period from late 1980s to early 1990s. Both ETr and ETp varied with 3-4 years and 7-8 year cycles; Low ETr and pan evaporation values took place in mountainous regions, while high values were found at low elevations. Spatial distribution characteristics of ETr and pan evaporation were similar to each other.
By relative contribution analysis, it found that average wind speed contributed the most to variation of ETr, followed by the average maximum temperature, sunshine duration and actual water vapor pressure, while the least contribution to the ETr were from average temperature and average minimum temperature. To explore the difference of actual contribution of various meteorological factors to ETr among the particular stations (Xinyuan station, Zhaosu station, Chabuchaer station), the method of actual contribution analysis was used. In Xinyuan station, the positive contributions of increasing temperature and growing sunshine duration to the increasing ETr were higher than the negative contributions of decreasing wind speed and rising actual water vapor pressure. In Chabuchaer station, the positive contributions of decreasing wind speed and increasing actual water vapor pressure to the decreasing ETr were more than the negative contributions of increasing temperature and rising sunshine duration. In Zhaosu station, the positive contributions of increasing wind speed and decreasing actual water vapor pressure to the decreasing ETr exceeded the negative contributions of increasing temperature and rising sunshine duration. Even though the trends of ETr in Zhaosu and Chabuchaer were both decreasing, the mechanism for contributions of meteorological factors to ETr were discriminating.
Sensitivity analysis found that ETr was most sensitive to maximum average temperature, followed by vapor pressure, sunshine duration and average wind speed, with the lowest sensitivity to average temperature and minimum temperature across the Yili region. If the temperature rise by 7%, ETr would increase by 7.7%. In the study period, the temperature showed increasing trends, but the ETr experienced a decreasing trend, resulting in the existence of “pan evaporation paradox” in the region. The increasing effect of temperature on ETo was cancelled by the decreasing impact of the changes in vapor pressure, average speed and sunshine hours, which resulted in an overall decrease in ETo during the study period.
According to the path analysis, the variations of ETr in the Yili region cannot be fully explained by the direct effect of a single factor. Due to the existence of the interactions among meteorological factors, one meteorological factor might influence ETr through another one with an indirect path. One meteorological factor with a little direct effect on the ETr may interact with another factor, which significantly influences on it through an indirect path. The results demonstrated that average wind speed and average temperature not only directly influence the variation of ETr, but also indirectly affect the variation of ETr through interacting with other factors.
ETr is lower than ETp and ETo, and ETp is higher than ETo. However, these three had a high correlation coefficient of over 0.9, respectively. As a conclusion, ETp can be used to forecast ETo and ETr; The results indicated that ETa experienced an increasing trend with a 10.8 mm/10 years during the study period, and an abrupt change was detected in 1986 according to the Mann-Kendall trend test. The spatial distribution of ETa presents a zonal distribution that it gradually decreased from Northwest to Southeast in the Yili region. ETo was close to wet environment evapotranspiration (ETw), while ETa was far from ETw. The ratio of ETa+ETo to ETw was only about 1.49, far less than 2.0, implying that there was a complementary relationship in the Yili region, but it belonged to unsymmetrical complementary relationship.|