To determine the main meteorological factors affecting the inter-annual variability of ET0 is the basis of accurate estimation of crop water requirement in the future, and is also of great significance in dealing with the climate change for agricultural production. In this paper, we investigated the factors affecting reference crop evapotranspiration (ET0) based on different mathematical statistic methods in Henan province. The evaluation results from the different methods were compared with the actual variation trend of ET0 and each factor. The effective method should be consistent with the trend. From common methods, we selected 5 methods to evaluate the effects of 7 meteorological factors on the inter-annual variability of ET0 based on the meteorological data of 17 stations in Henan province. The 5 methods included the correlation method, partial correlation method, dominant method, stepwise regression analysis, grey correlation analysis based on numerical average, numerical initial and numerical standardization data. The data were on the highest temperature, the lowest temperature, average temperature, relative humidity, precipitation, wind speed and sunshine hours. They were from meteorological stations. The annual average of daily ET0 was calculated by the Penman-Monteith method. The result showed that the influential factors based on the 5 methods were different for each station. By considering the trend of ET0 and each factor during a long term, we obtained the main factors affecting ET0 in Henan. The sunshine was the primary factor for Shangqiu, Xuchang, Lushi, Xixia, Nanyang, Zhumadian, Xinyang, and Gushi stations. The wind speed was the primary factor for Anyang, Xinxiang, Kaifeng, Zhengzhou, Luanchuan and Mengjin station. In the other stations, sunshine and wind speed were both the primary factor. In sum, the sunshine and wind speed were the main factors affecting reference crop evapotranspiration in Henan province, the average wind speed was more important than the other factors in the northern region of the Yellow River, but the sunshine was more important in the southern area of the Yellow River. The impact of the high temperature could not be ignored in the estimation of ET0 at Xinyang and Xixia stations. There were great differences in evaluation results among 5 methods. Grey correlation analysis method was not suitable for the evaluation of the main factors influencing ET0 variation because of the different results with different data transformation. Stepwise regression analysis was not suitable either because there were many differences between actual and prospective trend of ET0 based on the change trend of meteorological elements in each station. Correlation analysis, partial correlation analysis and dominant analysis were suitable to determine the main factors influencing ET0 variation in a given area with small difference in its conclusion and uniform results. Furthermore, dominant analysis method was adopted to rank meteorological factors influencing ET0 variation and its actual ET0 was consistent with the predicting trend of ET0, so it can be used to evaluate the sequence of meteorological factors affecting ET0changes in each station. However, the dominant method should be assisted by the results from the correlation and partial correlation method since it could not obtain the correlation between ET0 and each factor. It was suggested that correlation analysis and partial correlation analysis method could be adopted to analyze the relationship between each factor and ET0 in order to get credible results.%确定影响ET0年际变化的主要气象因子是准确估算未来作物需水的基础,对于农业生产科学应对气候变化具有重要意义.以河南省17个站点为例,分别采用国内外常用的5种数理统计方法评价7个气象要素对参考作物蒸散量(reference crop evapotranspiration,ET0)年际变化的影响程度,结果发现:日照和风速是影响河南省地区参考作物蒸散量的主要因子,黄河以北地区主要为风速,黄河以南地区以日照为主,信阳、西峡两地高温作用不容忽视.5种方法评判结果差异较大,采用灰色关联分析法,利用不同的数据变换方式,其结果大相径庭,认为其不适宜用于评价影响ET0变化主要因子的判定;结合各站气象要素年际变化趋势分析认为,逐步回归分析法得出的结论与各站点气象要素及ET0实际变化趋势存在多处悖理,不适宜用于评价影响ET0变化主要因子的判定;相关分析、偏相关分析、主导分析方法结果较为统一,差异较小,认为采用3种方法综合判定某地区影响ET0的主要因子,其结果较为可信.其中,采用主导分析法对各气象因子的影响排序与各因子对ET0的影响趋势以及ET0实际变化趋势较为一致,建议用于评价影响ET0变化的气象因子排序,但因其无法得到各因子与ET0的相关关系,需借助相关分析与偏相关分析才能得到详实可信的结果.
展开▼
