河南省夏季土壤湿度反演模型研究

发布时间：2018-08-06 14:54

【摘要】：干旱灾害是一种在全球范围内频繁发生的复杂的自然灾害,对全球的自然生态环境和人类社会经济活动的影响巨大。因此干旱灾害一直是研究的热点之一。河南省受到季风影响,全年降水分布不均,在夏季受到干旱灾害影响频繁且严重。土壤湿度是干旱的重要评价指标之一,土壤湿度与各个因子之间有较强的非线性耦合关系,需要充分考虑影响监测的各个因子,建立一个满足实际需要的监测模型。本文将以河南省作为主要研究区域,利用2007-2012年夏季MODIS遥感数据,基于TVDI指数方法和神经网络算法两种方面对土壤湿度的反演模型进行研究,从而获得研究区域的干旱灾害动态监测模型。取得的主要研究成果如下:(1)利用MODIS数据、DEM和实测数据,基于TVDI指数方法和神经网络算法对土壤湿度进行反演,建立河南省夏季土壤湿度反演模型,获得土壤湿度高分辨率的空间分布。两个反演模型能客观反映土壤湿度空间分布的变化情况,但两种方法在反演值上有所出入,其中通过TVDI指数方法反演的土壤湿度值较小。(2)分析基于神经网络的反演模型和基于遥感指数的反演模型与实测土壤湿度之间的误差和相关性关系。发现结合土壤湿度影响因子的神经网络土壤湿度反演模型精度相对TVDI指数反演模型的精度更精确。说明算法模型能更好地描述土壤湿度的空间分布特征。(3)以基于神经网络构建的土壤湿度模型为基础,构建2012年河南省夏季干旱灾情等级分布图。结果表明,2012年夏季旱灾灾情较轻,其中6月灾情较为严重,8月较轻,7月几乎不发生干旱。而豫北、豫西和豫中地区的干旱发生频率较高。以上研究工作建立了一个结合土壤湿度影响因子的高分辨率的土壤湿度动态反演模型,实现土壤湿度高分辨率的精细化表达,达到监测研究区域内干旱灾害发展变化的目的。
[Abstract]:Drought disaster is a kind of complex natural disaster frequently occurring in the world, which has a great impact on the global natural ecological environment and the social and economic activities of human beings. Therefore, drought disaster has been one of the hotspots of research. Henan Province is affected by the monsoon, the precipitation is uneven in the whole year, and it is frequently and seriously affected by the drought disaster in summer. Soil moisture is one of the important evaluation indexes of drought. There is a strong nonlinear coupling relationship between soil moisture and various factors. It is necessary to fully consider the factors that affect the monitoring and establish a monitoring model to meet the actual needs. In this paper, Henan Province is taken as the main research area, using MODIS remote sensing data from 2007 to 2012, based on TVDI index method and neural network algorithm, the inversion model of soil moisture is studied. The dynamic monitoring model of drought disaster in the study area is obtained. The main research results are as follows: (1) based on MODIS data Dem and measured data, soil moisture inversion model in summer of Henan Province is established based on TVDI index method and neural network algorithm. The spatial distribution of soil moisture with high resolution was obtained. The two inversion models can objectively reflect the spatial distribution of soil moisture, but there are some differences between the two methods in the inversion value. Among them, the soil moisture value retrieved by TVDI exponent method is small. (2) the error and correlation between the inversion model based on neural network and the inversion model based on remote sensing index and the measured soil moisture are analyzed. It is found that the precision of the neural network model combined with the influence factors of soil moisture is more accurate than that of the TVDI exponent inversion model. It shows that the algorithm model can better describe the spatial distribution characteristics of soil moisture. (3) based on the soil moisture model based on neural network, the distribution map of summer drought disaster in Henan Province in 2012 is constructed. The results show that the drought in summer of 2012 is relatively light, in which June is more serious, August is lighter, and there is almost no drought in July. The drought frequency is higher in north Henan, western Henan and central Henan. In this paper, a dynamic inversion model of soil moisture with high resolution combined with the influence factors of soil moisture was established, which realized the fine expression of high resolution of soil moisture, and achieved the purpose of monitoring the development and change of drought disasters in the study area.
【学位授予单位】：南京信息工程大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：S152.71;S423

【参考文献】