The Change Analysis Of Water Stage And Beach Wetland In Poyang Lake With The Aid Of Multi-temporal Active And Passive Remote Sensing Data

As the largest freshwater lake in China, Poyang Lake shows a dramatic variation of its water level. The tempo-spatial dynamic characteristic of Poyang lake "as a river in low water level and as a lake in high water level" make it impossible to be understood its variation pattern in water level and beach wetland only depend upon the gauge data. Remote sensing is a new technique to acquire the large-extent information of earth in the same time. It is a good complement to monitoring the beach wetland areas, in which the in situ data is difficult to be obtained for its bad accessibility. The remote sensing data of high temporal resolution makes itself more important in the dynamic wetland monitoring. This paper choose the active satellite altimeter data and passive high temporal resolution remote sensing images to investigate the change of the water level and beach wetland in Poyang Lake, the main achievement is shown as follows:1. Tempo-spatial pattern of water level change of Poyang Lake. The general temporal characteristics of water level variation were obtained by the extreme value, average and the statistics about frequency calculation of the daily water level data from 1955 to 2005. The representing gauge in the Poyang Lake was chosen by the calculation in the absolute difference and slope of water level among a specified gauge with others in the same day. The dynamic of water level of every gauge was discussed based on the monthly variation and the daily extent. The result showed that the high temporal resolution is a necessary definition to accurately monitor the beach wetland in Poyang Lake. The general change trend was depicted through the series time analysis and its reason was discussed.2. Integrated water level monitoring with the altimeter data. ICESat-GLAS and Envisat-RA2 were selected as the main data source to monitor the water level of wetland areas in Poyang Lake based on the comparison of acquisition ability among different altimeters. ICESat as altimeter with high spatial resolution was tested to simulate the slope of water surface in Poyang Lake, and the simulated result of continuous monitoring points in a track showed a similar general trend as the gauge observations. The higher temporal resolution altimeter Envisat-RA2 data was designed as a virtual gauge to monitor the non-gauged wetland area. The corrections of RA2 were combined to meet the inland water characteristics and the result showed a satisfied accuracy and the virtual showed a similar rule of water level as the nearest gauge "Duchang".3. Dynamic variation rules of water area in Poyang Lake. The first two bands of MODIS were selected to extract the water information in Poyang Lake after the comparison of spatial and temporal resolution and recognition ability among different remotely sensed images. To assure the stability of the further analysis, the detailed extent of Poyang Lake were defined based on the historic natural and artificial bank. The 439 scenes images from 2003-2008 were used to extract the water area, and then the area daily change rate and seasonal change pattern were analyzed through the results of extracted water area. The water covered frequency was compared to discuss the possible wetland types in different areas of Poyang Lake. With the aid of correspond 217 daily gauge observations from 2003 to 2005, six relationship models between the area of water covered and water level were constructed. The best model is Logistic through the accuracy comparison among six models. To assess the effect of the number of sample day impacting the stability of model, the slope parameter of the linear model were compared through random run 1000 times by selected 5 to 217 sample days, the result showed more than 30 days sample is necessary to make the model stable.4. Inundation pattern change of beach wetland. Based on the gauge observations in 2003-2005 and the lake terrain data in 1950s, the great difference was shown between the simulated inundated areas results derived from the observations of Tangyin gauge and Hukou gauge. It is obvious that Hukou gauge as a usual representing gauge in many research to simulate the inundated beach areas of Poyang lake will cause serious inaccurate. The tension spline interpolation model was introduced to simulate the inundated beach more accurately according to the attribute of gauge observations. The spatial pattern of terrain change was obtained to compare the simulated result of old terrain data and the extracted result based on the recently remotely sensed data. The drainage activity and annual sediment balance of input from five branches and output to Yangtze River were taken account to discuss the reason of the lake terrain change. Traditional classification method is difficult to distinguish different beach wetland for their similar spectral attribute. The high dynamic property of beach wetland make the classification result based on a simple temporal image could not represent the general characteristic of the beach wetland. So the temporal spectral was introduced to distinguish several different beach wetland. MODIS NDVI of the specified periods in a year were selected to combined to 4-bands temporal spectral image based on the season pattern of different beach wetlands. The two phrase classification result of temporal spectral images in 2003 and 2007 were compared and the result showed that there is a obvious dry trend in Poyang Lake.