Dynamic Simulation of Vegetation Abundance in a Reservoir Riparian Zone using a Sub-Pixel Markov Model
Document Type
Article
Publication Date
3-2015
Keywords
Vegetation abundance, LSMA, Sub-pixel Markov, Reservoir riparian zone
Digital Object Identifier (DOI)
https://doi.org/10.1016/j.jag.2014.09.004
Abstract
Vegetation abundance is a significant indicator for measuring the coverage of plant community. It is also a fundamental data for the evaluation of a reservoir riparian zone eco-environment. In this study, a sub-pixel Markov model was introduced and applied to simulate dynamics of vegetation abundance in the Guanting Reservoir Riparian zone based on seven Landsat Thematic Mapper/Enhanced Thematic Mapper Plus/Operational Land Imager data acquired between 2001 and 2013. Our study extended Markov model's application from a traditional regional scale to a sub-pixel scale. Firstly, Linear Spectral Mixture Analysis (LSMA) was used to obtain fractional images with a five-endmember model consisting of terrestrial plants, aquatic plants, high albedo, low albedo, and bare soil. Then, a sub-pixel transitive probability matrix was calculated. Based on the matrix, we simulated statuses of vegetation abundance in 2010 and 2013, which were compared with the results created by LSMA. Validations showed that there were only slight differences between the LSMA derived results and the simulated terrestrial plants fractional images for both 2010 and 2013, while obvious differences existed for aquatic plants fractional images, which might be attributed to a dramatically diversity of water level and water discharge between 2001 and 2013. Moreover, the sub-pixel Markov model could lead to an RMSE (Root Mean Square Error) of 0.105 and an R2 of 0.808 for terrestrial plants, and an RMSE of 0.044 and an R2 of 0.784 for aquatic plants in 2010. For the simulated results with the 2013 image, an RMSE of 0.126 and an R2 of 0.768 could be achieved for terrestrial plants, and an RMSE of 0.086 and an R2 of 0.779 could be yielded for aquatic plants. These results suggested that the sub-pixel Markov model could yield a reasonable result in a short period. Additionally, an analysis of dynamics of vegetation abundance from 2001 to 2020 indicated that there existed an increasing trend for the average fractional value of terrestrial plants and a decreasing trend for aquatic plants.
Was this content written or created while at USF?
Yes
Citation / Publisher Attribution
International Journal of Applied Earth Observation and Geoinformation, v. 35, Part B, p. 175-186
Scholar Commons Citation
Gong, Zhaoning; Cui, Tianxiang; Pu, Ruiliang; Li, Chuan; and Chen, Yuzhu, "Dynamic Simulation of Vegetation Abundance in a Reservoir Riparian Zone using a Sub-Pixel Markov Model" (2015). School of Geosciences Faculty and Staff Publications. 1344.
https://digitalcommons.usf.edu/geo_facpub/1344
