Water Depth Extraction From Remote Sensing In Longpaozi Of Zhalong Wetland

Water depth research from remote sensing data has gained great achievements inocean, rivers and lakes, but in wetlands which has not been adequately studied.Wetlands have played an important role on maintaining ecological balance and etc al.With the development of economics, wetlands have been impaired and the protectionof wetlands is extremely important. Water is the key for wetland to form, develop,evolve, recess and recycle, so keeping track water depth timely and precisely plays acrucial role on wetland resource conservation. Taken Longpaozi in ZhanglongWetland as study area, the multispectral model has been established by in-situmultispectral data and ground water depths,and the hyperspectral model has beenestablished by in-situ hyperspectral data and ground water depths, and two differentmodels have been compared. We try to inverse the water depths rapidly by usingremote sensing method and the water depth inversion model has been established torealize dynamic monitoring of water level. The main content includes:1．With GPS positioning,two measurements were processed on board. We gained84samples of water depth.There are26samples of water depth pacing withhyperspectral data,which were analysed with remote sensing image and hyperspectraldata respectively.This laid the foundation for this study.2．Multispectral model has been built by high resolution remote sensing imageQuickbird and in-situ water depth. According to the correlation between water depthsand radiance in different bands as well as band combinations, the inversion factors ofwater depths are selected and multispectral water depth inversion model areestablished. Based on verification and precision analysis, four rather good models areused to inverse the water depths for the whole water. Among them, the linear modelhas a good inversion effect, which indicate the trend from decrease to increase aroundcattails in Longpaozi.3．Ten lakes are considered as interest area in Zhalong Wetland, the lake waterdepth is calculated by multispectral water depth inversion model that adopts threeimages in1988,1999,2007from Landsat5/7. The single band model has a badinversion effect, which shows that it is hard to apply in regional water depths inversion. The linear model that involves four bands has a good effect This modelreflected the changing trend of water depths in Zhalong Wetland during20recentyears.4．Hyperspectral model for water depths has been established by water depthsfrom filed measurement and synchronization measurement of hyperspectral data.Compared with multispectral remote sensing, the hyperspectral data has features asmulti bands and high spectral resolution, which could spot tiny variations of complexspectral features of water. After the first order differential deposals, the correlationbetween water albedo and depths enhances greatly. The single band and multi bandmodel are built up respectively, which has the highest fitting degree R2of0.860andthe best number of bands is four.