Research On Extraction Model And Spatial Distribution Driving Model Of Rubber Forest In Five Provinces Of Northern Laos Based On Multi-source Remote Sensing

Natural rubber is an important strategic material.The natural conditions in the five northern provinces of Laos make it a new natural rubber growing area.While,the late effects of environmental changes such as the expansion of rubber forests will affect the overall environment of Laos for many years.Therefore,the spatial distribution of rubber forest and its driving mechanism and dynamic monitoring have become a hot issue in the study.It is of great significance to grasp the spatial distribution and spatial and temporal variation of rubber forest in time to plan the rubber planting,government decision support and regional ecological environment protection.(1)A new method which is rapid and repeatable to extract rubber forest spatial distribution was realized by using multi-source remote sensing data combined with rubber phenology.At first,we got the key phenology characteristic time window of rubber(Second half of mid-March to mid-March),followed by differentiating vegetation and non-vegetation,high-density vegetation and low-density vegetation,on the basis,we got the decision tree to extract rubber forest,and have a better accuracy of 80.79%.(2)The rubber-forest extraction results validation based on GF-1.Based on the GF-1 WFV image,we extracted the rubber-forest in Luang Namtha of Laos by the object-oriented method.The rubber-forest arroud of Namtha was extracted by the object-oriented and human-computer interaction method based on GF-1 PMS image,the aim to verify the accuracy of MODIS data extraction.The consequence show that the results of all kinds of verification are consistent with the results of confounding matrix verification in Chapter 3.Finally we extracted the rubber forest of five provinces in northern Laos in 2015 and 2010 by the verified method of MODIS image synchronicity.(3)Dynamic Monitoring of Rubber-forest Spatial and Temporal Expansion.The results of the dynamic expansion of the rubber-forest in the northern part of Laos from 2010 to 2015 show that most of the rubber young forests in 2010 become rubber forest in 2015.The rubber forests are mainly distributed in the Luang Namtha,the western part of Phongsaly and the northern part of Luang Prabang province,with relatively small sporadic distributions in the west of Bokeo and Oudomxay.And in the five years,the largest rate of change in the area of rubber forest is Oudomxay,followed by the Bokeo,Phongsaly had the slowest expansion,the rubber forests in Luang Namtha and(4)Luang Prabang maintain stable growth.The distribution of rubber forest showed a rising trend at the elevation,and it was sporadic distribution over 1500m.The expansion on the slope showed a rising trend of slope,and 35 0 is the maximum slope,while the aspect had no obvious regularity.(5)Relationship between spatial distribution of rubber forest and natural factors based on GAM.The dynamic relationship between the natural conditions of rubber plantation distribution and the spatial distribution of rubber forest is explored by using GAM.The results show that the spatial distribution of rubber forest is the optimal model for the spatial distribution of rubber forest,and the cumulative interpretation is 20.1%.The interpretation of the slope is the highest,and followed by the annual precipitation,the last one is elevation.the spatial distribution of rubber forest is consistent with the five factors.So,the GAM reveals the relationship between the spatial distribution of rubber forest and natural factors.(6)The driving model of Rubber forest spatial distribution and natural factors.Altitude,slope,precipitation factors and rubber forest spatial distribution area ratio of exponential regression simulation relationship,while aspect and temperature had no obvious relationship.The slope regression model which had the best fitting effect can be used to predict the rubber forest spatial distribution area under a certain slope in the Five province of northern Laos.The results of GAM model analysis also show that the slope,precipitation and altitude are the main natural driving factors of the spatial distribution of rubber forest,so we obtained the natural factors driving model of rubber forest spatial.