植被含水量的遥感反演方式

合集下载

土壤水分供给量的遥感定量反演方法

土壤水分供给量的遥感定量反演方法

土壤水分供给量的遥感定量反演方法
随着耕地资源减少及人口增加等现实情况的变化,农业系统的土壤水分供应面临着越来越大的短缺,以及更复杂的气候变化问题,管理官员和农民正在面临严峻挑战。

为此,以遥感技术为基础的定量反演技术在农业领域出现,可以用于和新技术配合,以改善土壤水分供应和农业生产。

遥感定量反演是以空间和时间形式识别和推算某一大地物理参数为基础,把被乘(受检)查的物理量在空间分布上的变量和空间环境上的其他地理因素进行匹配和综合的统计方法。

如可以利用遥感数据与非遥感数据,结合定性问题数据和定量数据,实现量化土壤水分供给量的反演识别。

此外,遥感定量反演技术可以让实际农田实时估算和把握水分资源,进而实现对土壤水分库的及时控制,从而确保土地的可持续性有效利用。

通过强大的数据库和智能优化计算处理,可以有效解决土壤水分供给量的检测问题,提供可靠的土壤水分管理模型,为农业的健康发展提供强有力的技术支持。

因此,遥感定量反演技术可被视为一种新型的智能农业管理工具,可迅速响应农业生态系统中出现的变化及其结果,以确保农产品可持续生产。

基于集成学习的土壤含水量遥感反演研究

基于集成学习的土壤含水量遥感反演研究

摘要土壤含水量是水文、气象、农业和生态等领域中的关键指标,对土壤含水量进行宏观、动态的监测可为土壤旱情分析、区域洪涝预警、土地退化预报以及生态环境评估等提供有效信息。

遥感技术的发展为获取大范围、长时间的土壤含水量实时信息提供了有效途径。

遥感反演是指基于地物电磁波产生的遥感影像特征去反推目标的实时状态参数,即将遥感数据转变为各种地表实用参数的过程。

目前针对土壤含水量的遥感反演模型可分为经验模型与物理模型:经验模型构造简单、便于实践但监测精度有限;物理模型具有坚实的理论基础,但涉及参数过多、适用性较差。

本文综合了经验模型与物理模型的优点,选择青藏高原为研究区、以MODIS卫星传感器资料作为主要数据源,构建了基于集成学习的土壤含水量遥感反演模型。

本文主要研究内容和结论如下:(1)土壤含水量相关光谱参数的提取收集并整理了青藏高原地区土壤温湿度实测数据以及遥感数据,完成了相关数据预处理操作;基于土壤的光谱反射特性,以MODIS地表反射率产品MOD09A1为数据源,对植被指数、植被覆盖度和叶面积指数等土壤含水量相关的光谱参数进行了反演。

(2)基于随机森林的地表温度重建基于随机森林算法对MODIS地表温度产品MOD11A1进行了重建,削弱了植被和地形的干扰,同时对数据缺失值进行了补充。

验证结果表明重建后的地表温度与实测地表温度具有良好的相关性,其准确性与空间连续性均得到了提升。

(3)基于温度-植被干旱指数的土壤湿度状况评估基于归一化差分植被指数(NDVI)反演结果与地表温度(LST)重建结果,构建NDVI-LST特征分布空间;针对特征分布空间中主体边界处的散点干扰问题,考虑各离散点的分布频率并进行干湿边方程的拟合;最后反演得到温度-植被干旱指数(TVDI),并据此对青藏高原地区土壤湿度状况进行评估。

结果表明TVDI在一定程度上能够反映土壤湿度分布的一般规律,但却不能对土壤含水量进行定量表述,其结果仍然具有局限性。

土壤含水量测量与反演方法综述

土壤含水量测量与反演方法综述

土壤含水量测量与反演方法综述摘要:目前土壤水分研究方法分为两大类:土壤水分直接测量法和反演法,反演法包括遥感监测法和模型模拟法。

本文系统分析了应用较广泛的几种农田土壤水分研究方法原理,研究发现,土壤水分直接测量法是当前研究土壤水分的主要方法,遥感监测法是未来研究土壤水分的发展趋势。

1 土壤水分直接测量法直接测量法包括烘干法、酒精灼烧法、中子仪法、张力计法、时域反射法、频域反射法、介电法、驻波法和电容电阻法等。

本文主要介绍烘干法、中子法和介电法。

1.1 烘干法烘干法包括经典烘干法和快速烘干法。

该方法的操作过程为:在田间地块选择代表性的取样点,按照观测规范要求深度分层取得土样,将土样放入铝盒并立即盖好,以减少水分蒸发对测量结果的影响。

对装有土样的铝盒进行称重,记为w1;打开盖子,置于烘箱内,将温度设为105~110℃对土样烘干6~8h,直至土样质量不再变化,对干土及铝盒进行称重,记为w2,则所测土层的土壤质量含水量的计算公式可表示为(1)2.2 中子法中子法的原理是中子从1个高能量的中子源发射到土壤中,与土壤中氢原子(绝大部分存在于水分子中)碰撞后,能量衰减,这些能量衰减的中子可被检测器检测到,通过标定建立检测到的中子数与土壤含水率的函数关系,便可转化得到土壤含水率。

利用中子仪测量土壤水分含量,只需预先埋设,测量时不破坏土壤结构,测量速度快,测量结果准确,可定点连续观测,且无滞后现象,但中子法并不能實现长期大面积动态监测。

由于中子法测量的实际上是半径约几到几十厘米的球体含水量,其半径随着土壤含水率大小而改变,所以土壤处于干燥或湿润周期时,或对于层状土壤以及表层土壤,中子法的测量结果并不可靠。

2.3 介电法利用土壤的介电特性来测量土壤含水量是一种行之有效、快速便捷,准确可靠的方法。

目前得到普遍认可的三种土壤水分介电测量方法——时域反射法、频域反射法和驻波率法。

(1)时域反射法(TDR)TDR 是近年来出现的测量土壤含水量的重要仪器,是通过测量土壤中的水和其它介质介电常数之间的差异原理,并采用时域反射测试技术研制出来的仪器,具有快速、便捷和能连续观测土壤含水量的优点。

基于光谱形状信息的植被叶片与冠层水分含量高光谱遥感反演模型研究

基于光谱形状信息的植被叶片与冠层水分含量高光谱遥感反演模型研究

促进遥感技术在生态学、环境科学、 地球科学等领域的应用和发展,为相 关学科的研究提供新的思路和方法。
为农业生产、水资源管理和环境保护 等实际应用提供科学依据和技术支持 ,有助于实现资源节约和环境友好型 社会的发展目标。
02
研究现状与问题
研究现状
01
植被叶片与冠层水分含量是植物生长状况和生态系统功能的关键参数,对于农 业生产和生态监测具有重要意义。
结果分析
通过对验证结果进行分析,发现该模型在反演水分含量 时具有较好的稳定性和可重复性,能够满足实际应用的 需求。
06
结论与展望
结论
建立了基于光谱形状信息的植被叶片与冠层水分含 量高光谱遥感反演模型,为精准农业和水资源管理
提供了有效的技术手段。
通过实验验证,该模型可以准确反演植被叶片与冠 层水分含量,为研究植物生理生态过程和环境变化
将结合更多的遥感数据源,如卫星、无人机等,实现更 大范围、更高精度的植被水分含量监测,为生态保护和 农业生产提供更加全面的技术支撑。
THANKS
谢谢您的观看
提供了新的途径。
模型在多个地区的应用结果表明,其具有较好的 稳定性和普适性,能够为实际生产和研究提供支
持。
展望
未来将继续优化模型算法,提高反演精度和稳定性,为 精准农业和水资源管理提供更加可靠的技术支持。
将进一步研究不同类型植物、不同生长条件下的水分含 量高光谱遥感反演模型,以扩大模型的应用范围和适用 性。
结合其他遥感数据和环境因素,进一步分析植被叶片 与冠层水分含量的空间分布和时间变化特征。
04
实验设计与结果分析
实验设计
实验目的
该研究旨在建立基于光谱形状信息的植被叶片与冠层水分含量高光谱遥感反演模型,通过 对叶片和冠层的高光谱数据进行分析,实现对植被叶片和冠层水分含量的准确估算。

植被覆盖地表土壤水分遥感反演

植被覆盖地表土壤水分遥感反演

植被覆盖地表土壤水分遥感反演一、概述植被覆盖地表土壤水分遥感反演是当前遥感科学与农业科学交叉领域的重要研究方向。

随着遥感技术的不断进步,利用遥感手段对植被覆盖地表下的土壤水分进行反演,已经成为监测土壤水分动态变化的有效手段。

本文旨在深入探讨植被覆盖地表土壤水分遥感反演的基本原理、方法进展及实际应用,以期为相关领域的研究和实践提供有益的参考。

植被覆盖地表土壤水分遥感反演的基本原理在于,通过遥感传感器获取地表植被和土壤的综合信息,进而利用特定的反演算法提取出土壤水分含量。

这一过程中,植被覆盖对遥感信号的影响不可忽视,如何有效去除植被覆盖的影响,成为植被覆盖地表土壤水分遥感反演的关键问题。

在方法进展方面,近年来国内外学者提出了多种植被覆盖地表土壤水分遥感反演方法,包括基于植被指数的反演方法、基于热惯量的反演方法、基于微波遥感的反演方法等。

这些方法各有特点,适用于不同的研究区域和植被类型。

随着深度学习等人工智能技术的快速发展,其在植被覆盖地表土壤水分遥感反演中的应用也逐渐受到关注。

在实际应用方面,植被覆盖地表土壤水分遥感反演在农业、生态、环境等领域具有广泛的应用前景。

通过实时监测土壤水分状况,可以为农业生产提供科学的灌溉指导,提高水资源的利用效率也可以为生态环境监测和评估提供重要的数据支持,有助于维护生态平衡和可持续发展。

植被覆盖地表土壤水分遥感反演是一项具有重要意义的研究工作。

随着遥感技术的不断进步和反演算法的不断优化,相信这一领域的研究将会取得更加丰硕的成果。

1. 背景介绍:植被覆盖地表土壤水分的重要性及其在农业、生态和环境监测中的应用。

植被覆盖地表的土壤水分是地球水循环的重要组成部分,它直接影响着植被的生长和生态系统的平衡。

在农业领域,土壤水分是作物生长的关键因素之一,其含量和分布直接影响着作物的产量和品质。

准确获取植被覆盖地表的土壤水分信息,对于指导农业生产、优化水资源管理具有重要意义。

在生态方面,土壤水分与植被覆盖度之间存在着密切的相互作用关系。

《典型草原不同植被条件下土壤水分遥感反演研究》

《典型草原不同植被条件下土壤水分遥感反演研究》

《典型草原不同植被条件下土壤水分遥感反演研究》篇一一、引言在农业、生态学以及地理学等众多领域中,土壤水分的测量和评估扮演着重要的角色。

特别是对于草原地区,其生态环境的脆弱性及土地资源的有限性使得土壤水分的动态监测尤为关键。

传统方法通常需要地面实测或取样分析,这不仅效率低下,还可能无法实现大面积的连续监测。

而遥感技术的引入为解决这一问题提供了新的途径。

本文旨在探讨典型草原不同植被条件下,如何利用遥感技术进行土壤水分的反演研究。

二、研究区域与数据源本研究选取了具有代表性的草原地区作为研究对象,该地区植被类型多样,包括草地、灌木丛、稀树草原等。

数据源主要来自卫星遥感数据和地面实测数据。

卫星遥感数据包括多光谱、高分辨率以及热红外等不同类型的数据,用于获取地表信息及土壤水分的间接估计。

地面实测数据则用于验证遥感反演结果的准确性。

三、遥感反演方法本研究采用了多种遥感反演方法,包括植被指数法、归一化水体指数法、温度植被干旱指数法等。

这些方法根据不同的植被类型和土壤水分特性,通过分析地表光谱特征、植被覆盖度、地表温度等因素,间接估算土壤水分。

同时,还结合了地理信息系统(GIS)技术,对反演结果进行空间分析和可视化表达。

四、不同植被条件下的土壤水分反演1. 草地条件下的土壤水分反演在草地条件下,采用植被指数法进行土壤水分的反演。

首先,根据多光谱数据计算归一化植被指数(NDVI),然后结合地面实测数据建立NDVI与土壤水分之间的回归模型。

通过该模型,可以估算出草地条件下的土壤水分含量。

2. 灌木丛条件下的土壤水分反演在灌木丛条件下,采用归一化水体指数法进行土壤水分的反演。

由于灌木丛地区存在一定量的地表水体,通过分析水体的光谱特征和空间分布,可以估算出该地区的土壤水分含量。

同时,结合高分辨率遥感数据,可以更准确地识别地表水体的分布和变化。

3. 稀树草原条件下的土壤水分反演在稀树草原条件下,采用温度植被干旱指数法进行土壤水分的反演。

《典型草原不同植被条件下土壤水分遥感反演研究》范文

《典型草原不同植被条件下土壤水分遥感反演研究》范文

《典型草原不同植被条件下土壤水分遥感反演研究》篇一一、引言随着遥感技术的不断发展,其在生态环境监测、土地资源管理等领域的应用越来越广泛。

其中,土壤水分作为草地生态系统的重要参数之一,其准确获取对于草原生态保护、草地资源管理和草地畜牧业发展具有重要意义。

然而,传统的土壤水分测量方法往往存在费时费力、空间分辨率低等问题。

因此,利用遥感技术进行土壤水分的反演研究成为了当前研究的热点。

本文旨在研究典型草原不同植被条件下土壤水分的遥感反演方法,以期为草原生态保护和土地资源管理提供科学依据。

二、研究区域与数据本研究选取了我国北方典型草原区作为研究对象,该区域植被类型丰富,包括草原、草甸、荒漠等多种类型。

研究所需数据包括遥感数据、气象数据和地面实测数据。

其中,遥感数据主要包括Landsat、Sentinel-2等卫星数据,气象数据来自中国气象局,地面实测数据包括土壤水分、植被指数等。

三、方法与技术本研究采用遥感反演的方法进行土壤水分的反演研究。

首先,通过分析不同植被类型下的光谱特征,提取出与土壤水分相关的特征参数。

其次,结合地面实测数据和气象数据,建立土壤水分与遥感参数之间的定量关系模型。

最后,利用该模型对遥感数据进行处理,得到土壤水分的空间分布情况。

在技术上,本研究采用了遥感图像处理软件ENVI和GIS分析软件ArcGIS等工具。

同时,结合机器学习和统计学习方法,建立了土壤水分反演模型。

四、结果与分析通过对比不同植被类型下的土壤水分反演结果,发现不同植被类型对土壤水分的反演结果具有显著影响。

其中,草原区域的土壤水分反演结果较为准确,而荒漠等植被稀疏区域的反演结果存在一定的误差。

这可能与植被覆盖度、土壤类型、地形等因素有关。

在建立土壤水分与遥感参数之间的定量关系模型时,我们发现植被指数、地表温度等参数与土壤水分之间存在显著的相关性。

通过机器学习和统计学习方法,我们建立了基于这些参数的土壤水分反演模型,并进行了验证。

《典型草原不同植被条件下土壤水分遥感反演研究》范文

《典型草原不同植被条件下土壤水分遥感反演研究》范文

《典型草原不同植被条件下土壤水分遥感反演研究》篇一一、引言随着遥感技术的不断发展,其在生态学、环境科学和农业科学等领域的应用越来越广泛。

其中,土壤水分的遥感反演研究是近年来关注的热点之一。

典型草原作为我国重要的生态系统之一,其植被覆盖类型多样,土壤水分状况对草原生态系统的稳定性和可持续性具有重要影响。

因此,本文旨在研究典型草原不同植被条件下土壤水分的遥感反演方法,为草原生态保护和可持续发展提供科学依据。

二、研究区域与数据本研究选取了我国北方某典型草原为研究对象,该地区植被类型多样,包括草地、灌木、森林等。

研究所用的数据包括遥感数据、气象数据和实地采样数据。

遥感数据主要包括Landsat、Sentinel-2等多光谱卫星数据;气象数据来自当地气象局提供的气象观测数据;实地采样数据则用于验证遥感反演结果的准确性。

三、土壤水分遥感反演方法1. 植被指数法:通过计算多光谱卫星数据中的植被指数(如NDVI、EVI等),分析植被覆盖情况对土壤水分的影响。

在此基础上,建立植被指数与土壤水分之间的回归模型,实现土壤水分的遥感反演。

2. 物理模型法:基于土壤-植被-大气之间的能量平衡和水热传输过程,建立物理模型,通过模型参数的估计和优化,实现土壤水分的遥感反演。

3. 机器学习方法:利用机器学习算法(如随机森林、支持向量机等),对多光谱卫星数据、气象数据等进行分析和训练,建立土壤水分与遥感数据之间的非线性关系模型,实现土壤水分的遥感反演。

四、不同植被条件下土壤水分反演结果与分析1. 草地条件下的土壤水分反演结果:在草地条件下,采用植被指数法、物理模型法和机器学习等方法进行土壤水分反演。

通过对比分析,发现机器学习方法在草地条件下的反演效果较好,能够较好地反映土壤水分的空间分布和变化趋势。

2. 灌木条件下的土壤水分反演结果:在灌木条件下,由于植被覆盖度较高,采用植被指数法进行土壤水分反演的效果较好。

同时,物理模型法也能够较好地反映土壤水分的状况。

植被覆盖地表土壤水分遥感反演

植被覆盖地表土壤水分遥感反演

1007-4619 (2010) 05-959-15 Journal of Remote Sensing 遥感学报Received : 2009-11-20; Accepted : 2010-04-20Foundation : Project of National Scientific Foundation of China (No. 40861020); the Open Fund of State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing (No. 09R03); Key Projects of Sciences Research of Education Ministry and Natural Science Foundation of XinJiang, China (No. 200821128).First author biography : ZHOU Peng, (1980— ), male, Post graduate of College of Resources and Environment Science, Xinjiang University, maily engaged in the study of resources, environment and remote sensing application in arid area. E-mail: zp5226@Corresponding author : DING Jianli, E-mail: watarid@Retrieval methods of soil water content in vegetation covering areasbased on multi-source remote sensing dataZHOU Peng, DING Jianli, WANG Fei, Guljamal.Ubul, ZHANG ZhiguangLab for Oasis Ecosystem of Xinjiang , College of Resource and Environmental Science , Xinjiang University , Urumqi 830046, ChinaAbstract: This paper takes the delta oasis of Weigan and Kuqa rivers in Xinjiang as the study area. Fusion image of SAR(Radar-sat image) combined with visible spectrum remote sensing image (TM image) is used to extract soil and vegetation water content in arid oasis. Based on the Normalized Difference Moisture Index extracted from homochronous visible spectrum re-mote sensing data, this thesis utilizes “water-cloud model” to wipe off vegetation influence from total backscattering coefficient of radar data and sets up the relationship between soil backscattering coefficient and soil moisture. Correlation coefficient for HH Polarization is R 2 =0.5227, for HV Polarization is R 2 =0.3277. Result shows that in arid and semi-arid area where the main crops are cotton and corn, the combination of C- band HH polarization radar data with visible image performs well in the study of removing vegetation influence while retrieving soil water content in medium vegetated areas.Key words: soil moisture, remote sensing, vegetable, backscatter coefficient, water-cloud model CLC number: TP751.1 Document code: A1 INTRODUCTIONSoil moisture is a very important component of the earthecosystems, which is the tie of the surface water and ground-water. Soil moisture plays a very important role in the global water cycle, it is also an important parameter in the hydrologi-cal, meteorological and agricultural research (Yuan et al., 2004). Large-scale soil moisture monitoring is an important content of agricultural water management and crop drought forecast-ing .At the same time, in regional and even the global-scale, soil moisture is also an essential parameter in the research of the land surface processes model, Which plays an important role in improving the regional and global climate models (Gao et al., 2001). Recently, the monitoring of regional scale soil moisture helps to resolve the problems of hydrological model in arid drainage basin and monitor the growth of crops. Tradition-ally, the estimation of soil moisture requires intensive labour operations in the field and needs to do some complex post-treatment processes, which are not only time-consuming, but also difficult to obtain a wide range of synchronous soil moisture information. The development of remote sensing technology provides an effective means of accessing the re-gional scale soil moisture information (Cashion et al., 2005; Urso & Minacapilli, 2006). At present, the methods of moni-toring soil moisture based on remote sensing are mainly ther-mal inertia, thermal infrared, crop water stress index, anomalies of vegetation index and microwave remote sensing (Chen et al., 1999; Guo & Zhao 2004). However, microwave remote sensing has the characters of all day and night, high repeated coverage, the penetrability through some surface objects and not restricted by weather conditions, which make microwave remote sensing monitoring of soil moisture widely used in arid and semi-arid area.Soil moisture monitoring of microwave remote sensing has experienced 30-year history, and has been established many backscattering coefficient models such as small perturbation model, Kirchhoff model, two-scale model and the integral equation model and so on. In addition, there are experience and semi-empirical models (Baghdadi et al., 2002; Liao et al., 2002; Wickel et al., 2001; Xiong & Shao, 2006). However, in the vegetated areas of arid and semi-arid oasis, the application of these retrieval models are confined by surface roughness and vegetation coverage. Higher vegetation coverage will cause lower estimation on soil moisture and the higher estimation on surface roughness (Liu et al., 2005) which make the acquisition of soil moisture became complex. The key question to the study of soil moisture is how to eliminate the influence of surface roughness and separate the vegetation scattering and absorption960Journal of Remote Sensing 遥感学报 2010, 14(5)from the soil moisture. In these models, vegetation water con-tent (VWC) is an important parameter. As it is very difficult tomeasure the parameter in the field, people usually use the opti-cal image to predict and then establish the relationship between normalized difference moisture index (NDMI) and VWC. Based on the use of optical image data, this paper uses NDMI to estimate the study area’s vegetation water content, then, em-ploys multi-polarization satellite-based radar data together with microwave scattering Water-Cloud Model eliminating the im-pact of the vegetation layer and isolating the contributions of vegetation scattering and absorption from the total backscat-tering coefficient, Which is the research on the estimation of surface soil moisture in vegetated areas in arid oasis.2 STUDY AREA AND DATA SOURCE 2.1 Status of study areaThe study area in this paper locates in the north of Tarim Basin, the lower reaches of Weigan and Kuqa Rivers, the mid-dle of Tianshan and the north of Taklamagan desert. The aver-age elevation is 920—1100m, which belongs to landwarm-temperate zone extreme arid climate. The average annual evaporation is 2420.23 mm, the average annual precipitation is 43.1mm, and the ratio of evaporation to precipitation is about 54 to 1. The image range of study area is that the coordinate of east longitude is from 82º15′ to 82º53′ and north latitude is from 41º15′ to 42º36′, which is confirmed according to the samples in field. Farm belt is dominant in study area, and the types of vegetation is not complex and is mainly composed of cotton (55%) and corn (15%) combining with other halophytes shrubs and salt secretion plants such as Tamarix, bulrush, Populus eupfratica and Alhagi sparsifolia Shap.2.2 Status of satellite dataRadarsat-2 satellite was successfully launched in December 14, 2007 at space launch base in Kazakhstan's Baikonur, it can provide 11 kinds of beam patterns, the highest resolution is 3m, the maximum imaging width is 500 km, and the maximum data rate is 445.4Mb/s, the range of incidence angle is 10°—60°. The C-Band can effectively extract soil moisture of 0—5cm soil layer. The primary parameters of RADARSAT-2 System are showed in Table 1.Table 1 Primary parameters of RADARSAT-2 systemCarrier frequency Polarization mode Bandwidth/MHz Antenna dimension/m 2Antenna quality/kgActive antenna Proposed isolation/dB C-Band (5.405GHz) HH, HV VH, VV11.6, 17.3, 30, 50, 10015×1.5750C-Band T/R>25The multi-polarization data of Radarsat-2 in September 10,2008 is chosen in this study, and the polarization is HH and HV. The homochronous visible data is Landsat-5 TM data which includes mid-infrared and near-infrared band for calculating NDMI. The image of study area is as follows:Fig. 1Polarization composite image of RADARSAT 2 HH/HVFig. 2 RGB composite image of Landsat5 TM 5432.3 Collection and analysis of samplesIn this study, Landsat-5 TM data obtained in Sep 27, 2007 is used as a reference map to choose the typical samples in the middle and the southwest of the oasis of Weigan and Kuqa rivers combining with GPS positioning technology. Sampling points are selected regularly distribution as much as possible and com-prehensive considering of the local soil, vegetation type and other factors. Sampling time is from sep 16, 2008 to sep 26, 2008 and there are 30 located sampling points in this study. Set 5 sampling points as Plum-shape in the range of 30m of each located points and profile mining each sample points and the profiles of each sample are 0—5cm. Borrow these soil samples and vegetation samples around of the soil sample dots back to laboratories, and dry them with the weighing method to obtain the soil avoirdupois moisture and vegetation water content.3 PREPROCESSINGImage data pre-processing mainly includes image registra-tion, geometric correction, noise reduction and color enhance-ment and so on. As it is need to find out the relationship be-tween normalized difference moisture index (NDMI) and vege-tation water content (VWC), the atmospheric correction of TM data is absolutely necessary. The level of SAR image data ob-tained is SLC, so it should do noise reduction, radiometric cali-bration and geometric correction. 3.1 Optical image processingIn order to keep the aboriginality, the image only did at-ZHOU Peng et al .: Retrieval methods of soil water content in vegetation covering areas based on multi-source remote sensing data 961mospheric correction and geometric correction neglecting otherpre-processing, which assure the objectivity of estimation. Be-cause of the impacts of atmosphere on visible-near-infrared and thermal infrared are different, it is necessary to separately carry out atmospheric correction using 6S model (Liu & Zhao, 2002). While geometric correction was performed to do multinomial correction of TM image using ground control points.3.2 Radar image processing 3.2.1 Radiometric calibrationAs the radiometric error caused by the difference of the dis-tance between ground scattering cell and radar, it is need to becorrected.Radiometric calibration equations is :20sin()nD Kσα= (1) where, D n 2 is the pixel intensity value, α is the radar incidence angle of the target location, K is the absolute calibration factor of the image. The values of K and α are obtained from the header files. SIGMA parameter correction is selected. Figure out the distance orientational incident angle of every pixels, then RADARSAT-2 satellite image data is changed to backscattering coefficient using the below calibration expres-sion:212()10log()/10log(sin())n D A A σθθ°=++ (2) where, D n is the gray-level value of radar image, A 1, A 2 are automatic gain control coefficient of radar system, θ is the dis-tance orientational incident angle of every pixels.3.2.2 Noise treatmentThe interrelated treatment in the processing of Radar-sat imaging causes a large number of spots (Speckle), which pro-duce an obstacle on feature extraction, so the noise treatment is used to eliminate the impact. After several rounds of test analy-sis, 5 × 5 enhanced Lee filter to the original SAR image has the best filtering effect, which can eliminate the most spots. 3.2.3 Geometric calibrationIn this thesis, the study area is flat, so cubic polynomial is selected as the correction method.4 RETRIEVAL OF SOIL MOISTURE CONTENTIN VEGETATED AREASThe vegetation water content (VWC) is extracted based on a choice of appropriate microwave scattering model. As it is definitely difficult to do large-scale area survey and extracting, which also definitely destroys vegetation, this paper uses Landsat-5 TM data to figure out normalized difference moisture index (NDMI) and find out the relationship between NDMI and VWC, then validates the vegetation water content. With that, multi-polarization satellite-based radar data together with mi-crowave scattering water-cloud model are employed eliminat-ing the impact of the vegetation layer on radar backscattering and finally retrieves the soil water content of study area.4.1 Choice of microwave scattering model of soil moisturecontent in vegetated areasIn the study on soil moisture of Microwave remote sensing,vegetated areas will interfere with soilbacks cattering signal. Therefore, it is necessary to establish a rational vegetation scat-tering model and remove the impact of the vegetation in soil moisture retrieval algorithm. A lot of foreign scholars havebeen studied on vegetation scattering model: a group of Michi-gan State University in USA, proposed “MIMICS” Model in 1990, MIMICS Model (Fung et al., 1992)was established fortall plants such as arbor, it took into account three levels ofvegetation canopy, tree trunks and the bottom surface ,whichmore realistically simulate the surface microwave backscat-tering. In 1978, Attma, Ulaby etc. using crops as the research object, proposed “Water-Cloud” model. “Water-Cloud” model assumes vegetation as homogeneous scatter and ignores themultiple scattering between vegetation layer and the surface. The overall backscattering of vegetation coverage area is sim-ply divided into two parts, namely, the direct reflection by the vegetation back scattering and the ground backscattering after the double attenuation of crop. Because there are mostly crops and other low vegetation in study area, “Water-Cloud” model is simple and practical. Be-sides, it uses very few parameters.The model is expressed as follows:0020can veg soil ()()()()σθσθγθσθ=+ (3)where, 0can()σθ is the total radar backscattering coefficient in vegetated areas, 0veg ()σθ is the direct backscattering coeffi-cient of vegetation layer, 0soil ()σθ is the direct backscatteringcoefficient of surface, γ2(θ) is the double attenuation factor of radar wave penetrating crops, where: 02veg veg ()cos()(1())A m σθθγθ=⋅⋅⋅− (4)22pp veg ()exp(2sec())Bm γθθ=− (5)In the expressions above, A and B are parameters depending on the type of vegetation which can be respectively obtained through regression arithmetic after the utilization of MIMICS model to simulate the surface parameters. m veg is vegetation water content (kg/m 3), θ is the angle of incidence of radar wave.4.2 Extracting of characteristic parameters and vegetationwater content estimationThe vegetation water content is defined as the water content of plants per unit and per area. As a significant input parameter of Water-Cloud model, VWC performs very crucial effect on the retrieval of soil moisture in vegetated areas (Liu et al., 2008). Now, the study of the regional retrieval of VWC using visible image has been already mature, and the relativity estab-lished between spectrum index and VWC is high(Jackson et al.,962Journal of Remote Sensing 遥感学报 2010, 14(5)2004; Rosnay et al., 2006; Wang et al., 2008). So, this study uses visible remote sensing to retrieve and field survey data ofsamples to validate the VWC in study area. In view of the spec-trum characteristic of soil, vegetation and water between 0.8 to1.7μm, Landsat5 TM(sep 16,2008) homochronous with fieldsurvey date is chose which includes near- infrared band 4(0.76μm)and mid-infrared band 5(1.55μm). The spectrumbandwidths are respectively 25 nm and 20 nm. According to the characteristic of vegetation that it has higher reflectivity in near-infrared band and lower reflectivity in mid-infrared band because of the absorbing effect of leave’s water content, NDMI (Gao,1996; ZARCO-TEJADA et al., 2003)is applied to extractvegetation water content information. As the vegetation instudy area are most cotton and other low salt secretion plants,based on relevant research (Chen et al., 2005), the equation isshowed as expressions 6, while expressions 7 is established to gain VWC according to the relationship between vegetationwater content of field survey and NDMI. NDMI (NIR MIR)/(NIR MIR)=−+ (6) VWC 2.15NDMI 0.32=+ (7) There are rivers, lakes and reservoirs in study area whichdisturb the extracting of vegetation water content. So, the ef-fects of these water bodies must be removed in the process of the retrieval work. This paper uses MNDWI index (Xu, 2008)which is improved a little according as the regional circs toextract water body information, the equation is as follows:MNDWI (1+0.5)(GREEN-MIR)/(GREEN+MIR)= (8) After that, masking the original image and the water body effects are removed. Analysis and validation indicate that NDMI performs well in the retrieval of vegetation water con-tent.4.3 Calculating of backscattering in vegetated areasThe soil moisture estimation of initiative microwave remotesensing is mainly affected by vegetation coverage and surface roughness. The soil backscattering in vegetated area is com-posed of body scattering from vegetation, surface scattering from earth and the multiple scattering between vegetation layer and the surface. These factors are taken into account in the estimation of soil moisture content (Li et al., 2002). As the hypsography in study area is flat, and mostly is covered by low plants, vegetation is considered seriously while the effect of surface roughness is not taken into account when calculating the soil backscattering coefficient.In the study of soil moisture estimation in vegetated area, it is necessary to use “Water-Cloud” model to remove the contri-bution of vegetation in the backscattering. Parameters A and B are experience constants and they are changed by regional limit. A relatively mature method is applied to calibrate the parame-ters (Chen et al., 2007; Gao et al., 2008). Firstly, Water-Cloud and MIMICS model are used to simulate the backscattering coefficient of vegetated area then obtain the surface direct scat-tering and the variational relationship between surface scatter-ing and incident angle. Follow that, parameters of the dominate plants in study area is showed in Table 2. Based on the simu-lating results, calibrate the vegetation parameters A and B men-tioned in Water-Cloud model through non-linear least squares method.Table 2 Parameters input to MIMICS modelClass Parameters ValueFrequency /GHz 5.4Sensor Polarization mode HH/HVWater content 0.17 Root mean square height /cm 0.817 Land surface Correlative length /cm 7.528 Water content 0.39Radius /cm 2.6 Thickness /cm 0.04 LeavesHeight /m 0.9 Radius /cm 0.3 Length /cm 21Branch Density /m 38.6Climate Temperature /(°)22 After the analysis of regression, A is 0.0019 and B is 0.137. In term of Eq.(4) and Eq.(5), bare soil backscattering is ac-quired with Water-Cloud model removing the vegetation influ-ence, the expressions are as follows:02veg veg ()0.0019cos()(1())m σθθγθ=×××− (9)2veg ()exp(20.137sec())m γθθ=−× (10)00veg 0soil 2()()()()σθσθσθγθ−=(11)The total backscattering coefficient of surface can be calcu-lated from Eq. (2) and soil backscattering coefficient after eliminating vegetation effect is got from Eq.(11). In order to clear up the connection between HH and HV , there separately does a correlation analysis of backscattering coefficients of HH and HV before and after eliminating vegetation effect. Consid-ering that the change of soil moisture can not be reflected effec-tively from radar backscattering coefficient in some high vege-tated area, ambiguous dots are canceled in the analysis, results show:Fig. 3 and Fig. 4 indicate that: whatever before or after eliminating vegetation effect, correlation coefficients of HHFig. 3 Simulated curve between the backscattering coefficient of HHand HV polarization before eliminating vegetation effectZHOU Peng et al.: Retrieval methods of soil water content in vegetation covering areas based on multi-source remote sensing data 963Fig. 4 Simulated curve between the backscattering coefficient of HHand HV polarization after eliminating vegetation effectand HV polarization backscattering coefficients are all very lowwhich are respectively 0.2534 and 0.2884. Therefore, the nextwork is to discuss the soil water content retrieval of HH, HVpolarization.4.4 Analysis of sample dots dataIn order to analyze the effect factors of backscatteringcoefficient in vegetation coverage area, we got 30 sample dotsfrom field survey which include water content in the soil layerof 0—5cm and the coordinates of these samples. Backscatteringcoefficient and water content of soil of the samples are showedin Table 3. According to the data in Table 3, respectively ana-lyze the relationship between HH, HV polarization backscat-tering coefficient and water content of soil before and afterremoving vegetation influence and discuss the impact tobackscattering coefficient of vegetation.Table 3 Backscattering coefficient and water content of soilSamples HH polarization/dBHV polarization/dBWater content in 5cm%C 1 −13.2507 −17.7713 5.1470 C 2 −10.7600−19.8877 4.9754 C 3 −11.5946−18.4416 5.0036 C 4 −11.3997−17.61538.4150 C 5 −12.2587−18.45157.6217 C 6 −12.5959−15.775814.7993 C 7 −11.5938−17.819411.7455 C 8 −10.7801−17.229313.1799 C 9 −14.3954−19.3163 3.6656 C 10 −9.6117−21.013910.7864 C 11 −12.8598−19.3812 4.3598 C 12 −15.9316−17.5230 3.2380 C 13 −14.9207−18.9558 2.2208 C 14 −9.7728−16.767318.8682 C 15 −7.8132−16.776320.1818 C 16 −13.1535−17.999312.1819 C 17 −12.8540−19.6966 4.9591 C 18 −14.1582−16.0084 4.1625 C 19 −13.8883−19.6414 4.0697 C 20 −13.5933−20.4478 4.8147 C 21 −12.8223−22.6523 3.1433 C 22 −10.5719−29.8080 1.1135 C 23 −9.2110−22.3394 6.7921 C 24 −8.8493−22.076813.1973 C 25 −12.9640−22.8122 1.3152 C 26 −11.4630−18.05148.5359 C 27 −15.0914−14.8482 2.0410 C 28 −11.0910−20.98437.2859 C 29 −9.4889−19.748315.7987 C 30 −8.8780−19.678310.3772Obtain the corresponding backscattering coefficient of the 30 samples in the image according as their coordinates, the relationship between before and after removing vegetation in-fluence of HH, HV polarization backscattering coefficient dis-played in Fig. 5.Fig. 5 Graph of relationship between HH, HV polarizationbackscattering coefficient(a) HH polarization; (b) HV polarizationIt is showed in Fig. 5 that, HH, HV polarization backscat-tering coefficient are becoming attenuation when separating the scattering and absorbing of vegetation using water-cloud model. The change of HH polarization data is thin, the reason is that the cropland of these samples is over the irrigation time and the evaporation is high, which results in lower water content of leaves and rhizome. Thus, the impact to scattering and absorb-ing of vegetation is tiny. HV polarization data is influenced by the scattering of plant canopy and branch. The study area is mostly covered by low crops and shrub. It is supposed that the vegetation is a orbicular scattering body covered with the earth’s surface and ignoring its size, figure and the distributing character of the orientation, which results in a tiny change after removing the influence of vegetation.The influencing analysis of vegetation on backscattering co-efficient 0soil()σθof bare soil:The backscattering coefficient of vegetation 0veg()σθ is acquired from Water-Cloud model, then the contribution of vegetation’s scattering and absorbing can be separated from the total backscattering of radar, and then the backscattering coeffi-cient 0soil()σθ of bare soil is obtained. The relationship be-tween HH, HV polarization backscattering coefficient and soil water content before and after eliminating vegetation effect are respectively showed in Fig. 6 and Fig. 7.964Journal of Remote Sensing 遥感学报 2010, 14(5)Fig. 6 Scatter plot of HH, HV polarization backscattering coefficient and soil water content before eliminating vegetation effect(a) HH polarization; (b) HV polarizationFig. 7 Scatter plot of HH, HV polarization backscattering coefficient and soil water content after eliminating vegetation effect(a) HH polarization; (b) HV polarizationFig. 6 and Fig. 7 illustrate the correlation coefficient of HH, HV polarization backscattering coefficient and soil water con-tent before eliminating vegetation effect are respectively 0.3769 and 0.1919,while them rise to 0.5227 and 0.3277 with the use of Water-Cloud Model after eliminating vegetation effect. Therefore, the relativity of soil backscattering coefficient and soil water content is increasing while eliminating the influence of vegetation. Research indicated HH polarization is muchmore sensitive to soil water content (Bao et al., 2006), data in this paper validate that HH polarization combining with Wa-ter-Cloud Model performs well in the separation of vegetation’s scattering and absorbing from the total backscattering of soil. Based on the above conclusion, establishing the connection model of soil water content and HH polarization backscattering coefficient, the regress equation is as follows:Water5=112.91e 0.1906δ (R 2=0.5227,n =30) (12)Use the Eq. (12) to retrieve the samples’ soil water content,results is as Fig. 8.Fig. 8 Soil water content mapField survey shows some of the river ways inside of the oa-sis are dry. The top left corner of the image is oasis-desertecotone, where the soil water content is less than 8%, as the vegetation cover area is out of the irrigating time, the soil water content are mostly between 7% and 15%, and only the soil water content of a hand of marsh are a little higher. The results of field survey are consistent with the detection of the soil moisture.5 CONCLUSION AND DISCUSSIONIn this paper, a case study is the Delta oasis of Weigan and Kuqa rivers in the Xinjiang where provided with predominance of terrain characteristic. Expert the advantages of initiative microwave remote sensing Radarsat-2 data and visible remotesensing TM data to extract water content. Firstly, using TM datato extract water content of plant, then combining with Wa-ter-Cloud Model to eliminate the impact of vegetation to back-scattering of soil, finally, simulating the relationship between backscattering coefficient and soil water content and validating and analysis the results, conclusions show:(1) Comparing with soil water content data and backscat-tering coefficient of SAR image, visible data combing with radar data performs well to extract the information of vegeta-tion and soil water content, and HH polarization backscattering coefficient is much closer to the change of soil water content. (2) Years’ difference has influence on character of vegetation. Although the year and the month of the soil water content data from field survey are contemporaneous with the image, the days are not, Which are likely to lead to differences of waterZHOU Peng et al.: Retrieval methods of soil water content in vegetation covering areas based on multi-source remote sensing data 965ingredient, leaves and rhizome water content and soil water content and them will disturb the extracting of plant water con-tent. The continuing study can try to relatively transform the spectrum data and construct some index to eliminate the effect of exoteric factors.(3) The hypsography in study area is flatness, and mostly covered by low plants, accordingly, the effect of surface roughness is not taken into account when calculating the soil water content. While in practical application the impact of sur-face roughness to radar echo is not neglected, which is a prob-lem in this research. The continuing work is to add the analysis of the influence of surface roughness.Presently, it has been a big progress of retrieving soil water content using microwave remote sensing, but it is also difficult to establish a universal arithmetic to retrieve soil moisture, which is mainly induced by the complexity of factors effecting microwave backscattering coefficient such as surface roughness and vegetation coverage and the uncertainty of the relationship between these factors. So how to eliminate these effects is the emphases of the continuing work and in latter research multi–source data should be comprehensively utilized and combining with kinds of retrieval model to found a regional initiative microwave remote sensing soil water content retrieval arithmetic in order to improve the precision of soil moisture inversion.REFERENCESBaghdadi N, King C and Clumsy A. 2002. An empirical calibration of the integral equation model based. SAR data ,soil moisture and surface roughness measurement over Bare soils. International Journal of Remote Sensing, 23(20): 4325—4340Bao Y S, Liu L Y, Wang J H and Li X W. 2006. Estimation of soil water content and wheat coverage with ASAR image. Journal of Remote Sensing, 10(3): 263—271Cashion J,Lakshmi V and Bosch D.2005.Microwave remote sensing of soil moisture:evaluation of the TRMM microwave imager(TMI) satellite for the Little River Watershed Tifton,Georgia. Journal of Hydrology, 307: 243—253Chen D Y, Huang J F and Jackson T J. 2005. Vegetation water content estimation for corn and soybeans using spectral indices derived from MODIS near-and short-wave infrared bands. Remote Sensing of Environment,98: 225—236Chen H L and Mao L X.1999. A review: theories, methods and devel-opment on soil moisture monitoring by remote sensing. Remote Sensing Technology and Application, 14(2): 55—65Chen Q, Li Z, Wang L and Wei X L. 2007. The study of estimating soil moisture using ERS wind scatterometer. Journal of Remote Sens-ing, 11(6): 803—810Fung A K, Li Z Q and Chen K S. 1992. Backscattering from a randomly rough dielectric surface. IEEE Transactions on Geoscience and Remote Sensing, 30(2): 356—369Gao B C. 1996. NDWI-A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sensing of Environment, 58: 257—266 Gao F, Wang J M, Sun C Q and Wen J. 2001. Advances in study on microwave remote sensing of soil moisture. Remote Sensing Technology and Application, 16(2): 97—102Gao S, Niu Z and Liu C Z. 2008. The estimation of tropical plantation forest leaf area index based on Radarsat SAR data. Remote Sens-ing for Land & resources, 78(4): 35—38Guo G M and Zhao B R. 2004. Monitoring soil moisture content with MODIS data.Soils, 36(2): 219—221Jackson T J, Chen D and Cosh M. 2004. Vegetation water content map-ping using Landsat data derived normalized difference water index for corn and soybeans. Remote Sensing of Environment, 92: 475—482Liao J J, Guo H D, Shao Y, LI X W and Veneziani N. 2002. Method and model of surface feature detection in arid to Semi-arid Area Using SAR Interferometry. Journal of Remote Sensing, 6(6): 431—434Li Z, Guo H D and Shi J C. 2002. Measuring the change of soil mois-ture with vege tation cover integration passive and active micro-wave data. Journal of Remote Sensing, 6(6): 481—484Liu W X,Liu X L and Wang J.2008.Romote sensing retrieval of soil moisture using ENVISAT-ASAR and MODIS images in vegetated areas of Huanan. Agricultural Research in the Arid Areas, 26(3): 39—43Liu W, Shi J C and Wang J M. 2005. Applying the decomposition tech-nique in vegetated surface to estimate soil moisture by multi- temporal measurements. Remote Sensing Information, (4): 3—6 Liu Z H and Zhao Y S.2002. The application study of MODTRAN and 6S model on atmospheric correction of MODIS image. Journal of Remote Sensing, 6: 217—222Rosnay P D, Calvet J C and Kerr Y. 2006.SMOSREX:A long term field campaign experiment for soil moisture and land surface processes remote sensing. Remote Sensing of Environment, (102): 377—389 Urso G D and Minacapilli M. 2006. A semi-empirical approach for surface soil water content estimation from radar data without a-priori information on surface roughness. Journal of Hydrology, 321: 297—310Wang J and Xu R S. 2008.Methods and research developments for retrival of vegetable water content by remote sensing. Remote Sensing Information, (1): 100—105Wickel A J, Jackson T J and Wood E F. 2001. Multitemporal monitoring of soil moisture with radarsat SAR during the 1997 Southern Great Plains hydrology experiment. International Journal of Re-mote Sensing, 22(8): 1571—1583Xiong W C and Shao Y. 2006. Applying multi-temporal synthetic aper-ture radar(SAR) to evaluating soil-water and salt content based on IEM in arid areas. Journal of Remote Sensing, 10(1): 111—117Xu H Q. 2008. Comment on the enhanced water index(EWI): a discus-sion on the creation of a water index. Geo-Information Science, 10(6): 777—780Yuan W, Li Z Q and Liu N. 2004. Analysis of data sets with different microwave remote sensing mode in soil moisture retrieval. Engi-neering Science, 6(6): 50—56Zarco-Tejada P J, Rueda C A and Ustin S L. 2003.Water content esti-mation in vegetation with MODIS reflectance data and model in-version methods. Remote Sensing of Environment, 85: 109—124。

《典型草原不同植被条件下土壤水分遥感反演研究》范文

《典型草原不同植被条件下土壤水分遥感反演研究》范文

《典型草原不同植被条件下土壤水分遥感反演研究》篇一一、引言随着遥感技术的快速发展,其在农业、环境监测等领域的应用越来越广泛。

其中,土壤水分作为草原生态系统的重要参数,其监测与反演技术对于草原生态保护、草地资源管理和农牧业生产具有重要意义。

本文针对典型草原不同植被条件下的土壤水分遥感反演进行研究,旨在为草原生态环境监测提供技术支撑。

二、研究区域与方法2.1 研究区域本研究选取我国北方典型草原区作为研究区域,该区域植被类型丰富,生态环境脆弱,对土壤水分的变化较为敏感。

2.2 研究方法本研究采用遥感技术手段,结合地面实测数据,对典型草原不同植被条件下的土壤水分进行反演研究。

具体方法包括:(1)遥感数据获取与处理:收集研究区域的遥感数据,包括多时相、多光谱的卫星遥感数据,进行辐射定标、大气校正等预处理。

(2)植被指数计算:根据预处理后的遥感数据,计算归一化植被指数(NDVI)等植被指数,用于表征植被覆盖度和生长状况。

(3)土壤水分反演模型构建:结合地面实测土壤水分数据,构建土壤水分反演模型,包括统计模型、物理模型和机器学习模型等。

(4)不同植被条件下的土壤水分反演:根据构建的反演模型,对不同植被条件下的土壤水分进行反演,分析土壤水分的时空分布规律和变化趋势。

三、不同植被条件下的土壤水分反演结果3.1 草地类型对土壤水分的影响研究发现在草地类型对土壤水分的影响显著。

草地类型主要包括草原、草甸、荒漠草原等。

不同草地类型的植被覆盖度、根系分布、土壤类型等均存在差异,导致土壤水分的分布和变化规律也不同。

通过遥感反演技术,可以有效地监测不同草地类型的土壤水分状况,为草原生态保护和草地资源管理提供科学依据。

3.2 季节变化对土壤水分的影响季节变化对土壤水分的影响也不可忽视。

春季和夏季是草原生长的主要时期,植被覆盖度较高,土壤水分相对较充足;而秋季和冬季,草原生长减缓或进入休眠期,植被覆盖度降低,土壤水分相对减少。

通过遥感反演技术,可以实时监测季节变化对土壤水分的影响,为农牧业生产提供决策支持。

《典型草原不同植被条件下土壤水分遥感反演研究》

《典型草原不同植被条件下土壤水分遥感反演研究》

《典型草原不同植被条件下土壤水分遥感反演研究》篇一一、引言随着遥感技术的不断发展,其在生态环境监测、土地资源管理等领域的应用越来越广泛。

其中,土壤水分作为生态系统的重要参数之一,其准确获取对于草原生态保护、农业发展等具有重要意义。

典型草原作为我国重要的生态屏障和农业生产基地,其不同植被条件下的土壤水分遥感反演研究显得尤为重要。

本文旨在通过遥感技术手段,探究典型草原不同植被条件下土壤水分的分布特征和变化规律,为草原生态保护和农业生产提供科学依据。

二、研究方法本研究采用遥感技术手段,结合地理信息系统和统计分析方法,对典型草原不同植被条件下的土壤水分进行反演研究。

具体包括以下步骤:1. 数据收集:收集典型草原区域的遥感数据、气象数据、植被数据等。

2. 数据处理:对遥感数据进行预处理,包括辐射定标、大气校正、图像融合等。

3. 植被分类:利用遥感数据对典型草原进行植被分类,划分出不同植被类型区域。

4. 土壤水分反演:采用遥感反演模型,对不同植被类型区域的土壤水分进行反演。

5. 结果分析:对反演结果进行统计分析,探究土壤水分的分布特征和变化规律。

三、不同植被条件下土壤水分的分布特征1. 草地土壤水分分布草地是典型草原的主要植被类型之一,其土壤水分分布受到气候、地形、植被等多种因素的影响。

通过遥感反演,我们发现草地土壤水分在空间上呈现出一定的分布规律。

在降雨较多的季节,草地土壤水分较高,而在干旱季节则相对较低。

同时,草地土壤水分的分布还受到地形的影响,如在山谷、河谷等低洼地区,土壤水分相对较高。

2. 灌木林土壤水分分布灌木林是典型草原的另一种重要植被类型,其土壤水分的分布与草地有所不同。

由于灌木林的根系较为发达,能够更好地吸收和保持土壤水分,因此其土壤水分相对较高。

同时,灌木林的分布也受到地形的影响,如在山腰、山坡等地区,灌木林的土壤水分相对较高。

3. 其他植被类型土壤水分分布除了草地和灌木林之外,典型草原还存在着其他植被类型,如荒漠、湖泊等。

《典型草原不同植被条件下土壤水分遥感反演研究》范文

《典型草原不同植被条件下土壤水分遥感反演研究》范文

《典型草原不同植被条件下土壤水分遥感反演研究》篇一一、引言草原生态系统对环境变化非常敏感,其健康状态对气候变化和生态安全具有重要意义。

土壤水分作为草原生态系统的重要组成部分,是草地生长和植被覆盖的重要影响因子。

传统的土壤水分测量方法,如土壤烘干法、土壤水势测定等,虽能得到准确数据,但工作量大、时间周期长,无法满足快速动态监测的需求。

随着遥感技术的不断发展,利用遥感技术进行土壤水分的反演已经成为研究热点。

本文旨在研究典型草原不同植被条件下土壤水分的遥感反演方法,以期为草原生态环境的保护和恢复提供科学依据。

二、研究区域与方法本研究选取了我国典型草原区作为研究对象,包括内蒙古、新疆等地的草原区域。

采用遥感技术手段,结合地面实测数据,对不同植被条件下的土壤水分进行反演研究。

在方法上,首先收集了研究区域的遥感数据,包括多时相、多光谱的卫星遥感数据。

其次,结合地面实测的土壤水分数据,建立土壤水分与遥感数据之间的关系模型。

最后,利用该模型对研究区域的土壤水分进行反演,并分析不同植被条件下的土壤水分变化规律。

三、结果与分析1. 土壤水分与遥感数据的关系模型通过分析遥感数据与地面实测的土壤水分数据,我们发现土壤水分与遥感数据中的植被指数、地表温度等指标存在一定的关系。

其中,归一化植被指数(NDVI)与土壤水分的关系最为密切。

随着NDVI值的增加,土壤水分含量也呈现出增加的趋势。

2. 不同植被条件下的土壤水分变化规律在典型草原区,不同植被条件下的土壤水分存在明显的差异。

草地植被覆盖度较高的区域,土壤水分含量相对较高;而裸地、沙地等区域的土壤水分含量较低。

此外,季节性降雨也会对土壤水分产生影响,雨季时土壤水分含量较高,旱季时则较低。

3. 遥感反演结果分析利用建立的模型对研究区域的土壤水分进行反演,得到的结果与地面实测数据具有较好的一致性。

在不同植被条件下,遥感反演结果能够反映出土壤水分的空间分布和变化趋势。

此外,通过分析时间序列的遥感数据,还可以监测到季节性降水对土壤水分的影响。

多源遥感数据反演土壤水分方法.

多源遥感数据反演土壤水分方法.

多源遥感数据反演土壤水分方法张友静1,王军战2,鲍艳松3(1 河海大学水文水资源与水利工程科学国家重点实验室,江苏南京 210098;2 中国科学院寒区旱区环境与工程研究所,甘肃兰州 730000;3 南京信息工程大学大气物理学院,江苏南京 210044摘要:基于A S AR A PP 影像数据和光学影像数据,根据水云模型研究了小麦覆盖下地表土壤含水量的反演方法。

利用TM 和M OD IS 影像构建的植被生物、物理参数与实测小麦含水量进行回归分析,发现T M 影像提取的归一化水分指数(N D W I反演精度较好,相关系数达到0 87。

根据这一关系,结合水云模型并联立裸露地表土壤湿度反演模型,建立了基于多源遥感数据的土壤含水量反演模型和参数统一求解方案。

反演结果表明:该方案可得到理想的土壤水分反演精度,并可控制参数估计的误差。

反演土壤含水量和准同步实测数据的相关系数为0 9,均方根误差为3 83%。

在此基础上,分析了模型参数的敏感性,并制作了研究区土壤缺水量分布图。

关键词:土壤含水量;多源遥感数据;水云模型;A S AR;多尺度中图分类号:P338 9 文献标志码:A 文章编号:1001 6791(201002 0222 07收稿日期:2009 03 09基金项目:国家自然科学基金资助项目(40701130;40830639作者简介:张友静(1955-,男,江苏南京人,教授,主要从事遥感机理与方法研究。

E m a i:l zhangy @j hhu edu cn 土壤含水量是地表和大气界面的重要状态参数,并直接影响地表的热量和水量平衡,因而受到水文、气象和农业灌溉等多个学科的关注。

微波土壤水分遥感研究始于20世纪80年代,其中最具代表性的是U laby 利用试验数据得出土壤后向散射系数的主导因素为粗糙度和含水量[1]。

80年代后,Dobson 和U laby 利用车载、高塔、航空平台的微波数据研究了土壤湿度反演的最佳工作模式,并一致认为小角度入射后向散射系数对土壤湿度最敏感[2]。

基于集成学习的土壤含水量遥感反演研究

基于集成学习的土壤含水量遥感反演研究

摘要土壤含水量是水文、气象、农业和生态等领域中的关键指标,对土壤含水量进行宏观、动态的监测可为土壤旱情分析、区域洪涝预警、土地退化预报以及生态环境评估等提供有效信息。

遥感技术的发展为获取大范围、长时间的土壤含水量实时信息提供了有效途径。

遥感反演是指基于地物电磁波产生的遥感影像特征去反推目标的实时状态参数,即将遥感数据转变为各种地表实用参数的过程。

目前针对土壤含水量的遥感反演模型可分为经验模型与物理模型:经验模型构造简单、便于实践但监测精度有限;物理模型具有坚实的理论基础,但涉及参数过多、适用性较差。

本文综合了经验模型与物理模型的优点,选择青藏高原为研究区、以MODIS卫星传感器资料作为主要数据源,构建了基于集成学习的土壤含水量遥感反演模型。

本文主要研究内容和结论如下:(1)土壤含水量相关光谱参数的提取收集并整理了青藏高原地区土壤温湿度实测数据以及遥感数据,完成了相关数据预处理操作;基于土壤的光谱反射特性,以MODIS地表反射率产品MOD09A1为数据源,对植被指数、植被覆盖度和叶面积指数等土壤含水量相关的光谱参数进行了反演。

(2)基于随机森林的地表温度重建基于随机森林算法对MODIS地表温度产品MOD11A1进行了重建,削弱了植被和地形的干扰,同时对数据缺失值进行了补充。

验证结果表明重建后的地表温度与实测地表温度具有良好的相关性,其准确性与空间连续性均得到了提升。

(3)基于温度-植被干旱指数的土壤湿度状况评估基于归一化差分植被指数(NDVI)反演结果与地表温度(LST)重建结果,构建NDVI-LST特征分布空间;针对特征分布空间中主体边界处的散点干扰问题,考虑各离散点的分布频率并进行干湿边方程的拟合;最后反演得到温度-植被干旱指数(TVDI),并据此对青藏高原地区土壤湿度状况进行评估。

结果表明TVDI在一定程度上能够反映土壤湿度分布的一般规律,但却不能对土壤含水量进行定量表述,其结果仍然具有局限性。

《典型草原不同植被条件下土壤水分遥感反演研究》

《典型草原不同植被条件下土壤水分遥感反演研究》

《典型草原不同植被条件下土壤水分遥感反演研究》篇一一、引言草原生态系统对环境变化非常敏感,其健康状态对气候变化和生态安全具有重要意义。

而土壤水分作为草地生态系统中关键的水文参数,直接影响植被生长、能量交换等重要生态过程。

然而,传统的土壤水分测量方法通常费时费力,且难以在较大空间范围内进行连续监测。

因此,利用遥感技术进行土壤水分的反演研究显得尤为重要。

本文旨在研究典型草原不同植被条件下土壤水分的遥感反演方法,以期为草原生态环境的监测与保护提供科学依据。

二、研究区域与方法本研究选取了具有代表性的典型草原区域,通过收集该区域的遥感数据、气象数据以及地面实测数据,开展土壤水分的遥感反演研究。

(一)研究区域研究区域位于我国某典型草原区,具有丰富的植被类型和地形地貌特征。

(二)方法与技术路线1. 数据收集:收集遥感数据(包括Landsat、Sentinel-2等卫星数据)、气象数据以及地面实测数据。

2. 遥感图像处理:对遥感数据进行预处理,包括辐射定标、大气校正等,以提高数据的信噪比。

3. 植被指数计算:根据预处理后的遥感数据,计算归一化植被指数(NDVI)等植被指数。

4. 土壤水分反演:利用统计模型、机器学习等方法,建立植被指数与土壤水分之间的关系模型,进行土壤水分的反演。

5. 结果分析:对反演结果进行统计分析、空间分析等,探讨不同植被条件下土壤水分的分布特征及变化规律。

三、植被条件下的土壤水分反演模型构建(一)统计模型本研究采用多元线性回归模型进行土壤水分的反演。

以NDVI等植被指数为自变量,土壤水分含量为因变量,建立统计模型。

通过分析自变量与因变量之间的关系,得出土壤水分含量的估算模型。

(二)机器学习模型除了统计模型外,本研究还采用了机器学习算法进行土壤水分的反演。

利用支持向量机、随机森林等算法,建立植被指数与土壤水分之间的非线性关系模型。

通过训练和验证,得出适用于不同植被条件的土壤水分反演模型。

四、结果分析(一)统计模型结果分析通过对比反演结果与地面实测数据,发现多元线性回归模型的估算精度较高,能够在一定程度上反映实际土壤水分状况。

  1. 1、下载文档前请自行甄别文档内容的完整性,平台不提供额外的编辑、内容补充、找答案等附加服务。
  2. 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
  3. 3、如文档侵犯您的权益,请联系客服反馈,我们会尽快为您处理(人工客服工作时间:9:00-18:30)。
其中C为生化组分含量,N为入选的波段数,α0、αi分 别为回归常数和第i个回归系数,D(λi)为入选的波段光 谱值。
统计分析方法
• 确定了植被水分的敏感光谱后,在统计分析的基础上,前人提出了很多不 同的指数和方法来诊断植物的含水量。这些统计模型可以大致分为以下 三类:
• 1:建立光谱指数 • 2:基于光谱导数变量建立模型 • 3包络线消除法
植被含水量光谱反演原理及 水分的敏感光谱波段
• 1 植被含水量光谱反演原理
• 2 水分的敏感光谱波段
1:植被含水量光谱反演原理
• 每一种物质对不同波长的电磁波的吸收和反射都不同,物质的这 种对不同波段光谱的响应特性叫光谱特性。植被光谱诊断便是基 于植被的光谱特性来进行的。
• 植被反射波谱中某些波长的光谱反射和吸收差异是由植被中化学 组分分子结构的化学键在一定辐射水平的照射下发生振动引起的, 从而产生了不同的光谱反射率,且该波长处光谱反射率的变化对 该化学组分的含量多少非常敏感(故称敏感光谱) 。植被含水量 光谱诊断的实现便是以植被水分敏感光谱的反射率与水含量的相 关关系为基础的。
• 研究表明,FMC、RWC和EWT是表征含水量的三个不相关量,是定量提取含水量 的不同方法。
• 研究发现在用遥感数据反演含水量时,短波红外波段与EWT相关性较好,而与 FMC相关性较差,并且表明用EWT来表征含水量要优于用FMC表示,因为FMC要受 叶子中的干物质影响。
• 在研究EWT和FMC的关系时发现,对于桉树叶,其近红外波段反射光谱与FMC有 很好的相关性,而短波红外波段的反射光谱与EWT高度相关。
建立光谱指数
• 建立的光谱指数一般是两个波段或多个波段的组合,如简单的加减组合、比值或者 是归一化比值,这是根据植被波谱的物理特性和半经验方法提出的。
• 例如Penuelas等发现用水分指数WI(WI=R970/R900)能清楚地指示水分状况的变化.
• Penuelas和Inoue在随后的研究中还表明WI(WI=R900/R970)与NDVI(NDVI=(R900R680)/(R900+R680))的比值WI/NDVI不仅可以用来预测叶片的水分含量,还可以用来 预测植株或冠层的含水量,且显著提高了预测的精度.
• RWC =(FW-DW)/(TW –DW)×100 %
• EWT =(FW-DW)/A
g/ cm^2
• 植物鲜重用FW表示;植物干重用DW表示;植物饱和鲜重用TW表示; 单位都是g
• 叶面积用A表示 单位是cm^2
• 把植物鲜重在80℃下烘干24小时以上直到恒重,就得到植物的干重
• 把新鲜植物水合至饱和就得到了植物饱和鲜重
植被含水量反演方法
统计分析方法
• 叶片生化组分对应特定光谱的吸收特征,利用多元回归可以确定化学组分 和光谱数据相关程度高的波段和波段组合,从而反演出化学组分含量。在 进行回归分析的过程中,采用逐步回归的方法,通过F检验,使对因变量贡 献大的因子随时可以进入方程,贡献小的因子又可以随时剔除,从而建立 最优回归方程。
植被含水量的定义
• 常用含水量表示方法有三种:
• 叶片含水量FMC( Fuel Moisture Content )
• 相 对 含 水 量RWC( Relative Water Content )
• 等效水深EWT ( Equivalent Water Thickness)
• FMC =(FW-DW)/(FW or DW) ×100 %
2:水分的敏感光谱波段
对于MODIS数据而言,1230~1250nm比较适合用于预测含水量。 Penuelas等指出近红外858nm波段是反演水含量的一个好的选择,因为相 对于更长的近红外和短波红外波段,此波段对水含量的变化不敏感,故很 适合用它来进行归一化处理。
Chen等在用Landsat数据反演含水量时发现短波红外位于 1550~1750nm波段较佳,并且发现1640nm、2130nm波段处对水分的吸收很 敏感。
植被含水量反演
介绍
近年来随着成像光谱技术的兴起,如何利用遥感数据监测植被化 学特性,已成为全球变化研究中重要的议题。
水分是控制植物光合作用、呼吸作用和生物量的主要因素之一, 水分亏缺会直接影响植物的生理生化过程和形态结构,从而影响植物 生长和产量与品质,因此植物的水分在农林业的应用中是一个重要的 参数,研究植物水分状况具有重要的意义。利用成像光谱遥感估测植 物水含量有很大的潜力,它可以实时快速准确地监测或诊断出植物水 分状况,从而可有效及时指导精确植物灌溉,有效评价自然干旱情况, 及时预测森林火灾。
2:水分的敏感光谱波段
大量的研究表明植被水分对热红外波段 (6. 0~15. 0μm)、 近红外(700~1300nm)和短波红外(1300~3000nm)波段比较敏感。
自1963年提出以冠层温度指示植被水分亏缺以来,冠层温度法 成为诊断作物水分状况的一个重要手段。
30多年来,有关科学家相继提出了参考温度法、胁迫积温法、 作物缺水指标法以及水分亏指数法等,并在田间以及区域尺度上 展开了大量的应用研究。
2:水分的敏感光谱波段
但利用热红外波段反演植被水分仍受到环境状况的强烈影响, 还不足以说明作物水分状况在时间和空间上随环境的巨大变化而 变化,并且热红外波段更适合于指示植被的蒸腾作用所以对植被 含水量的反演更多的焦点聚集于近红外-短波红外波段。
2:水分的敏感光谱波段
为了明确水分的敏感光谱波段,早在1951年, Curcio就指出820nm、 970nm、1200nm、1450nm和1940nm处是水分的强吸收波段,可以用来诊断 植物的含水量。
在1971年, Thomas就用完全饱和的叶片在室温下逐渐干燥的方法 来获取不同含水量下的反射光谱,并研究了叶片含水量与光谱反射率之间 的关系,结果表明叶片的光谱反射率随叶片含水量的下降而增加,1450nm 和1930nm波Leabharlann 的反射率与叶片的相对含水量显著相关。
Sims等经过研究指出950~970nm ,1150~1260nm和1520~1540nm 波段和冠层水分相关性很好,尤其在960nm和1180nm处没有大气的干涉,是 监测冠层含水量的较佳选择。
相关文档
最新文档