巢湖地区土壤相对湿度气象预测模型研究

侍永乐 章超 靳青春

摘要 [目的]描述气象影响因子与土壤相对湿度的关系，预测巢湖地区逐旬10～100 cm土壤相对湿度，定量评估土壤旱涝情况。[方法]通过分析巢湖地区各层土壤相对湿度逐月分布特征，讨论各气象影响因子与土壤相对湿度的相关关系，并分别基于多元线性回归和BP神经网络方法建立预测模型。[结果]气象因子和土壤相对湿度存在相关关系，相关系数从大到小分别为蒸发量、气温、日照时数、相对湿度和降水量。随着土层的深入，两者的相关系数和显著程度呈下降趋势。[结论]2种方法均能较好地预测出巢湖地区各层土壤相对湿度逐旬动态变化，可为了解巢湖地区籼稻主生长周期内的土壤相对湿度变化提供依据，为该地区作物旱涝评估提供参考。

关键词 巢湖;土壤相对湿度;多元线性回归;BP神经网络

中图分类号 P426.1 文献标识码 A 文章编号 0517-6611（2021）01-0218-05

doi：10.3969/j.issn.0517-6611.2021.01.058

Abstract [Objective]In order to describe the relationship between meteorological factors and soil relative humidity， predict the soil relative humidity of 10-100 cm of ten days in Chaohu Region and evaluate the soil drought and flood quantitatively. [Method]The relationship between meteorological factors and soil relative humidity was discussed by analyzing the monthly distribution characteristics of soil relative humidity in Chaohu Region， and the prediction models were established based on multiple linear regression and BP neural network. [Result]There is a correlation between meteorological factors and soil relative humidity. The correlation coefficients from large to small are evaporation， temperature， sunshine duration， relative humidity and precipitation. With the depth of soil layer， the correlation coefficient and significance of the two have a downward trend. [Conclusion]The two methods can predict the dynamic change of soil relative humidity in every layer of Chaohu Region， and provide the basis for the change of soil relative humidity in the main growth cycle of Indica Rice in Chaohu Region， and provide a reference for crop drought and flood assessment in this area.

Key words Chaohu;Relative soil moisture;Multiple linear regression;BP neural network

土壤湿度即土壤水分，是气候系统关键变量之一，作为陆面过程中一个重要的物理量，是影响作物生长、发育及产量的重要因素[1]，它通过改变地表反照率、热容量、地表蒸发、植被生长状况等途径影响着气候的变化[2]。土壤相对湿度是土壤水分状况的综合体现[3]，大气降水、灌溉、地下水等有助于土壤相对湿度增加，而土壤蒸散、植物吸收、蒸腾、土壤水分渗漏和径流则加快了土壤相对湿度降低[4]。关于土壤相对湿度的预测模型研究，大多学者采用降水量、土壤含水量、作物旱情等指标来研究[5-6]，运用模拟、替代、实测资料从空间分布、季节变化、垂直分布等不同层面、不同尺度上分析了土壤湿度变化特征及其影响因素。……