| Reservoir riparian zones are essential ecologic land areas located between terrestrial and aquatic ecosystems, and play important roles in the exchange of water, carbon, and energy between the two types of ecosystems. Vegetation, one of the crucial components of reservoir riparian zone ecosystems, plays a significant role in stabilizing and mitigating ecological problems caused by the dramatic changes of environments. As a critical indicator for measuring the growth status of vegetation, vegetation abundance is also important for the evaluation of reservoir riparian zone eco-environment.Due to the effective application in ecological studies, Remote Sensing (RS), Geographic Information System (GIS), and Mathematical models were used in this study. Seven cloud-free Landsat TM/ETM+/OLI images acquired in years2001,2004,2007,2008,2009,2010and2013were used to derive vegetation abundance of Guanting Reservoir Riparian Zone, to assess the accuracy of a process model, to determine its applicability in simulating vegetation abundance, and to predict the dynamic succession of vegetation abundance in the near future. Meanwhile, environmental factors including meteorological and hydro geomorphic data were used to access their impacts to the dynamic succession of vegetation abundance. Some of the conclusions can be made as results of this study as follows.(1) Spectral mixture analysis is one of the effective way to estimate vegetation abundance. For Landsat images, a maximum of four endmembers are usually used due to the strong correlations among visible bands. Our study indicates that four endmembers may be inadequate to spectrally characterize the complex and heterogeneous landscapes of reservoir riparian zones. While it is possible to allow the use of a fifth endmember by extending the spectral dimensionality of the Landsat data, which can capture the spectral variability of our study region more effectively than the four-endmember model. The study also indicates that the FCM (Fuzzy c-Means) classification based on the nonlinear assumption can handle mixed pixels more effectively than LSMA (Linear Spectral Mixture Analysis).(2) Unlike traditional crisp classifications, sub-pixel classification makes it possible to add the spatial component to the Markov chain analysis as the transitive probabilities can be estimated on a per-pixel basis, which extends Markov model’s application from traditional regional scale to sub- pixel scale. The study indicates that sub-pixel Markov model can produce a reasonable result under the assumption that the transitive probabilities remain constant in a short period.(3) The dynamic succession of vegetation abundance from2001to2013were determined by calculating the average fractional values of both terrestrial plants and aquatic plants of the Guanting Reservoir riparian zone from each stage. The result shows that the average fractional value of terrestrial plants increased dramatically between2001and2008with0.029as its annual growth amount. However, the average fractional value of aquatic plants decreased0.007per year during this period, which may be attributed to the shrinking water area and the continuous sedimentation. Both the increasing and decreasing trends were interrupted in2009. After2009, the average fractional values of terrestrial and aquatic plants experienced the trends of increase and decrease, respectively. However, the dynamic succession procedure became slower than the years before2008. Moreover, the simulation of future vegetation abundance changes based on a sub-pixel Markov model showed a continuing trend of increase in terrestrial plants, and a decline in aquatic plants.(4) A number of environmental factors were used to disclose their merits in driving the overall vegetation abundance change dynamics. The study indicates that the reservoir inflow water and groundwater depth correlate well with the dynamic succession of aquatic plants and terrestrial plants, respectively. The average annual precipitation and the average mean temperature are statistically significant at0.05-level with the probability of change from terrestrial plants to aquatic plants. The study also indicates that the reservoir inflow and outflow water, as well as the groundwater depth are more strongly correlated with the succession of terrestrial plants when using a two-year transition probability, which may perhaps due to a mismatch in the one-year temporal scale between the environmental factors and the type of changes. The fractional values of terrestrial plants increase and then decrease while the values of aquatic plants decrease dramatically as the distance from water boundary increases. |
PDF Full Text Request