Proceedings Article | 18 October 2019
KEYWORDS: Earth observing sensors, Landsat, Climate change, Environmental sensing, Climatology, Spatial resolution, Satellites, Near infrared, Sensors, Data centers
A general decrease of wetland is observed at a global due to climate change and human activity of socio-economic developments. Landsat has been giving us the long record of the Earth's surface in moderate spatial resolution using a multispectral sensor, which can help us better to analyze and attribute the changes in wetland.
This study use the multi-temporal observing data since 1979 from Landsat to reveal and evaluate the changes of long time in wetland in the studying area of Baiyangdian wetland in China which is surrounded by the farmland and rural residents. The results of studying could better understand the varying pattern and changing driven factors as to support the management, conservation and restoration of the wetland for government.
We collected a total of 191 periods Landsat (L1T) cloud-free images, including MSS, TM, ETM+ and OIL images from the USGS from 1979 to 2017 in the studying area. Moreover, GF-2 data with spatial resolution of 4m in 2017, which is a series of observing land satellites with high spatial resolution released by the China Resources Satellite Applications Center, is also used to get the land cover and verification of results.
Firstly, open waters, which is water body apparently without covered by aquatic plants in wetland, are extracted using the dynamic threshold method from 1984 to 2017 using the top of atmosphere reflectance (TOA) data of Landsat TM, ETM+ and OIL in NIR, SWIR and Green bands. The threshold values extracting water in NIR, Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water Index (NDWI) are tested through ROI data then the water pixels are extracted using the multiple thresholds for each period. The verification and optimizing the thresholds are carried by visual interpretation. Lastly we obtained the distribution of open water in a long time from 1984 to 2017. Secondly, we take a clustering analysis of multi-temporal NDVI data derived from the collected Landsat data using an iterative self-Organizing data (Isodata) algorithms. We investigate how the change of precipitation and artificial water recharge, which are main sources of water supply for wetland, affect the variations of open water during 33 years.
As a result, it is found that the artificial water recharge is an indispensable water supply of keeping of the ecological water volume for the Baiyangdian wetland as the natural water supply from rivers and precipitation are dwindling. Moreover, the artificial uses, such as paddy field, golf course, extending of settlement and road etc., have been decreasing the water and induced the degeneration of aquatic plants in wetland. These results will not only increase our knowledge for the change of wetland but also provide the assistant decision-making of the ecological conservation of wetland for the management strategies.
