|其他摘要||The increasingly highlighted contradiction of water supply and demand, serious waste of agricultural water and water pollution have become the bottleneck of agriculture sustainable development as well as the national economic development. These problems are the major strategic issues that can affect national security. Quite a long time in the future, China's water shortage problems have to be resolved through water conservation. Crop water requirement rules could provide strong fundamental theory supports and technique supports for the development of agricultural water-saving techniques.The amount of water required to compensate the evapotranspiration loss from the cropped field is defined as crop water requirement. Although the values for crop evaportranspiration and crop water requirement are identical,crop water requirement refers to the amount of water that needs to be supplied,while crop evaportranspiration refers to the amount of water that is lost through evaportranspiration. At present,crop evaportranspiration can be calculated by the equation ET=Kc * ET0. As a result, ET0 has become the key point to calculate the crop water requirement.
In this paper, the inter-annual and inter-monthly variation trends of meteorological factors from 1989 to 2014 were analyzed using meteorological data observed by Aksu station in Tarim river basin. Eight alternative ET0 methods，the FAO-56 Penman-Monteith、 1972 Kimberly Penman、FAO PPP-17 Penman、FAO-24 Radiation、1985 Hargreaves、1948 Penman、Priestley-Taylor and 1957 Makkink were evaluated for application in oasis of the Traim Basin. Average monthly reference evapotranspiration in Aksu area were predicted by BP neural network and generalized regression neural network using meteorological data from 1989 to 2014. The results showed：
（1）Accumulation of annual average temperature in Aksu were decreased from 1989 to 1996 and were increased from 1996 to 2014; Accumulation of annual maximum temperature were decreased from 1989 to 1996 and were incresed from 2006 to 2014,and showed a appear flat fluctuating trend from 1997 to 2005; Accumulation of annual minimum temperature were negative except the year of 1989,1996,2014 ; Accumulation of annual sunshine appeared to be a 2-year period prior downward trend; Accumulation of annual wind speed were incresed from 1989 to 1997 and 2005 to 2014,while were decreased from 1998 to 2004 ; Accumulation of annual relative humidity were decreased from 1989 to 1994 and increased from 1995 to 2003, and appeared to be a gentle fluctuation from 2005 to 2014.
（2）The Rate changes of average temperature in March,April, June,August were positive from 1989 to 2014,while others are negative. Overall, The Rate changes of average temperature in winter was decreased, and were increased in spring, summer, autumn; The Rate changes of average monthly maximum temperature are positive in the month of January, February, September, November, December,others are negative; The Rate changes of average monthly minimum temperature were decreased in the month of January,February,May, July, November, December, others were increased. Winter substantially decreased, spring, summer and autumn are on the rise; The Rate changes of average monthly sunshine hours were negative in the month of August and November; The Rate changes of average wind speed are positive in February, August, November, December,other month were negative; The Rate changes of average relative humidity were negative in January,March,April, August the other months were negative.
（3）Reference crop evapotranspiration in Aksu station tended to decrease with time from 1989 to 2014, sunshine and wind speed are the main meteorological factors that influence Reference crop evapotranspiration. Judging from the season, reference crop evapotranspirationthe was influenced most by sunshine in spring, by relative humidity in Autumn, by maximum temperature in winter.
（4）1972 Kimberly Penman produced ET0 estimates which were very close to the results estimated by FAO-56 Penman-Monteith method in comparison with other methods, however, FAO-24 Radiation estimated ET0 which was very far from FAO-56 Penman-Monteith. 1972 Kimberly Penman can be used for the calculation of reference evapotranspiration in oasis of Traim Basin.
（5）The BP model and generalized regression neural network model were suitable for forecasting ET0 when the input were highest temperature lowest temperature ,sunshine duration and days。|