新疆地区参考作物腾发量的灰色模型预测

该文依据新疆地区6个站的长序列逐日气象观测资料，基于Penman-Monteith公式计算了逐日参考作物腾发量（ET0），并应用重标极差法对ET0未来变化趋势进行了分析。运用灰色关联理论计算了各站各气象因子与年ET0间的灰色关联度和关联序。在此基础上，运用灰色系统理论建立灰色不等维递补GM（1，h）模型对6个站的年ET0进行了模拟预测，并与灰色GM（1，1）模型进行了比较。结果表明：各站ET0年内变化均呈抛物线型，4－9月ET0依各站顺序为：若羌＞吐鲁番＞哈密＞喀什＞和田＞伊宁；6站年ET0赫斯特指数均大于0.5，各站未来的趋势与历史呈正相关，依然是波动递减；总体上，平均温度、日照时数、饱和气压差对各站年ET0的影响比较大；灰色不等维递补GM（1，h）模型预测相对误差限为0～7.31%，预测精度明显高于GM（1，1）模型。该研究表明采用灰色模型预测新疆地区参考作物腾发量精度较好。

Prediction of reference crop evapotranspiration with grey model in Xinjiang region

Based on daily meteorological observation data of six stations in Xinjiang region, Penman-Monteith equation was used to calculate daily reference crop evapotranspiration(ET0). The rescaled range analysis method was applied to analyze future tendency of annual ET0. Grey relational theory was used to calculate the correlation degree and order between meteorological factors and annual ET0 for each station. Based on that, grey information renewal GM(1, h) model of unequal dimensions which was established by the grey systematic theory was used to simulate and predict annual ET0 of the six stations, and GM(1, h) model was compared with it. The results showed that the tendency of month ETo was parabola, and the sequence of ET0 variation with month was Ruoqiang > Tulufan > Hami > Kashi > Hetian > Yining, respectively. From April to September, the Hurst indice of annual ET0 in six stations were all larger than 0.5, the future tend of stations were still volatility decreasing which were positively related to their history. In general, the annual ET0 values of various stations were mostly affected by average temperature, sunshine hours, saturated pressure difference. The relative error limits for the predictions using grey information renewal GM(1, h) model of unequal dimensions ranged from 0 to 7.31%, with obviously high accuracy of prediction than GM(1, 1) model. This study shows that accuracy of predicting reference crop evapotranspiration in Xinjiang region is relatively good.