Monitoring And Simulation Of Land Cover Change In Mianning County

Land cover change plays an important role in the study of global change and sustainable development,and regional land cover change research is an extension of large-scale research.Mianning County is located at the gateway of the Panxi region.The ecological environment is fragile,geological disasters occur frequently,and man-made transformation leads to diverse types of land cover and drastic changes.The current research on land cover is mainly concentrated in large-scale plain areas.Due to the difficulty of remote sensing interpretation at the county scale in complex terrain areas,the factors affecting the complexity of land cover changes and the difficulties in field investigations have led to relatively few studies.With the advancement of national land spatial planning,it is especially necessary to master the efficient method of remote sensing interpretation of land cover changes at the county scale,the evolution of space-time pattern and future development trends.This paper takes Mianning County in Sichuan Province as a research area to study the process,mechanism and pattern of land cover change.(1)The object-oriented image classification method applied to county-level land cover classification has a significant effect.The ESP2 best segmentation scale evaluation tool automatically selects the best segmentation scale after debugging the initial value,which replaces the tedious traversal attempts and improves the scientific nature.The best segmentation scale for Landsat5 TM / Landsat8 OLI image is 82,of which the shape factor weight It is 0.1,and the weight of the compactness factor is 0.5.The contour boundary of the ground class is clear at this segmentation scale,which is better than the excessive segmentation at a small scale and a large amount of calculation.The CART decision tree algorithm combined with visual interpretation to classify the segmented images to verify the 2018 land cover interpretation results.The overall classification accuracy is 0.9284,and the Kappa coefficient is 0.8866,which exceeds the evaluation criteria,confirming that this classification method is accurate and efficient.(2)The land cover change in Mianning County changed significantly in the past30 years from 1990 to 2018.From 1990 to 2018,cultivated land,grassland and other land use continued to decrease,artificial surface and forest land continued to increase,and water bodies experienced staged changes;from 1990 to 1999,from 1999 to 2010,and from 2010 to 2018,cultivated land moved to artificial surface and grassland The transfer to forest land is more active;from 1990 to 2018,the spatio-temporal differentiation of the artificial surface is the most significant;various land types have been relocated in the three phases of 1990-1999,1999-2010,and 2010-2018.The change trend of the cover is consistent.The migration distance is within 4km except for the water body exceeding 8km from 1999 to 2010.(3)The three factors of topography,socioeconomics,and distance have significant effects on land cover,and the spatial scales are different.Each driving factor has different effects on land cover changes.After selecting the driving factors that affect the change of land cover and collinearly removing it,8 factors significantly affecting the land cover were obtained.Seven different scales were designed to perform multi-scale regression analysis on land types and driving factors,and it was found that the land cover was in different spaces.There is a scale conversion effect on the scale.Starting from the actual situation and applicability of the study area,at a spatial scale of 30 m × 30 m,the maps are integrated and the contours of the land types are clear,which is used as the spatial scale for simulation prediction.(4)The Logistic-CA-Markov coupling model has a good effect on the prediction of land cover changes.The simulation of human intervention scenarios is more reasonable than the future development trends presented by the simulation of naturally occurring scenarios.The Logistic-CA-Markov coupling model is used to simulate two scenarios of natural occurrence and human intervention in Mianning County in 2027 and 2036.The simulation under human intervention is superior to the simulation under natural occurrence scenarios.The development trends of the two models are the same,but the direction and magnitude of change are different.Under the simulation of human intervention scenarios,compared with the status of land cover in 2018,the cultivated land first decreased and then increased,and there was a significant increase by 2036;the forest land first increased and then decreased,with little change;the grassland continued to decrease,with a large change;Due to the expansion of the artificial water surface,the water body has a larger increase;the artificial surface continues to grow,which is nearly three times that of 2018 by 2036;other land use decreases first and then increases slightly,with little change.(5)Put forward specific suggestions on the control and development of nationalspace based on the protection of basic farmland,taking into account the ecological and economic benefits.Taking the protection of basic farmland as the bottom,focusing on the rational allocation of space for urban expansion,the protection and development of forest land and grassland is based on the local natural and human environment,giving full play to the advantages of the local area,and the protection of the water area is the Daqiao Reservoir and Yele,which integrate economic and ecological functions.Reservoirs are the core,while paying attention to the protection of river basins,and other unused lands are mainly developed into barriers for ecological protection.