Font Size: a A A

Study On Ecological Landscape Patterns Change Based On Multi-layer Perceptions Neural Network Model

Posted on:2013-02-18Degree:MasterType:Thesis
Country:ChinaCandidate:J C LiuFull Text:PDF
GTID:2230330371471178Subject:Physical geography
Abstract/Summary:PDF Full Text Request
The land use landscape pattern is the performance of landscape heterogeneity and the result of ecological processes, which reflects the research area ecological stability degree. The land use landscape pattern changes will change the flow of energy and material between different landscapes, and then cause the change of ecological system structure and function. The study of the landscape pattern and its potential changes has important significance to the regional ecological process, and also lays the foundation for landscape pattern change prediction and landscape optimization.This study is based on landscape ecology and neural network theory; taken Yongchuan as the study area. In order to get the land use data of2000,2005and2010, TM satellite remote sensing image o these three periods are interpreted with the RS and GIS technology. On this basis, the land use change in space and time scale is analyzed with land use change model and land use type of transfer matrix analysis model. The landscape pattern change in space and time scale is analyzed with the max Figure spot index and Spread degrees. Lastly, elevation and accessibility factor is set as the landscape transfer model variable. Through analyze landscape patterns potential with IDRISI, the potential distribution of2000-2005and2005-2010periods is got.On the basis,we can get the simulation of landuse layout in Yongchuan area of2020.Through the analysis of three periods’landuse change, we can get the following conclusion:the structure of land use in Yongchuan has changed greatly from2000to2010; the area of construction land, forest land and water area has increased orderly, but the area of cultivated land and open land has decreased. Construction land has the biggest growth, its growth rate is59.58%; Forest land has the second growth, its growth rate is114.37%; water area has a stable growth of11.54%.Open land has the biggest reduce, its reduce rate is69.62%; Cultivated land has a fluctuant reduce of1.55%in the study period. The analyses of landuse conversion in space and time scale with Transfer matrix analysis model show that:each landscape in Yongchuan has a complex transformation in different periods. Construction land’s increase mainly comes from cultivated land, and its contribution rate is59.88%; forest land’s increase mainly comes from open land, and its contribution rate is78.73%; water land’s increase mainly comes from open land, and its contribution rate is73.62%; cultivated land’s increase mainly comes from open land, and its contribution rate is99.01%. Open land has the biggest reduce all of the landscapes; its main change direction is cultivated land and forestland, which account for73.99%and20.10%of the total amount transferred. Cultivated land has changed to open land and construction land from2005to2010, and its contribution rates are61.02%and19.63%respectively; construction land’s increase mainly comes from cultivated land and forestland, and its contribution rate are57.14%and36.18%; forestland’s increase mainly comes from open land, and its contribution rate is95.60%; open land mainly changes to forestland, which account for99.40%of the total amount transferred.The analysis of landscape pattern sequential change with plaques density, the max Figure spot index and the spread of the landscape pattern index shows that:the ecological diversity level has increased, but reunion degree of landscape has reduced with an increased fragmentation degree. Cultivated land has been maintain maximum patches index, and its fragmentation and complexity degree increase reflected from the border density and the max Figure spot indexes its stability trend reduced.Construction land’s area increased sharply, the shape index changed rapid in the early time and slowly later. Its biggest patches index stayed more stable, and its main part has been not affected by human activities for many years, so its advantage degree increased. Open landscape’s shape index changes show that the area reduced and broken down; water landscape area has increased gradually, and its complicated degree increased also.The contrast analysis of potential changes of part of the landscape’s transfer shows that:the conversion potential of cultivated land to construction land increased from0.74to0.77, and high potential range expanded; the conversion potential of forestland to construction land reduced from0.87to0.42; the conversion potential of open land to construction land rise slightly, and its range expanded; the conversion potential of open land to forestland has been in high probability level, and its scope mainly concentrated in the high altitude which is the transitional zone of two landscape. Markov transition probability matrix, which comes from the data of land use change from2000to2010, and the main factor metastatic potential distribution, was combined with2010as the beginning to simulate the change of landuse, the results show that:the area of cultivated land, forestland and open land present decreasing trend, and construction land presents the fast increase trend with a growth rate of45.21%, water presents slightly increase trend with a growth rate of6.55%from2010to2020.
Keywords/Search Tags:Landscape pattern, Neural network, Land-use change, Yongchuan
PDF Full Text Request
Related items