| In recent years,the state attaches great importance to the control of rocky desertification,and has made many relevant rules and regulations on the control of rocky desertification on several occasions.The problem of rocky desertification has become the key and difficult problem of the ecological environment construction in the karst areas of China,which seriously restricts the healthy development of the local social and economic development,and Guangxi rocky desertification is also a serious problem in the country.One of the regions has become the research focus of many experts and scholars.Rocky desertification will cause a series of land problems,such as land productivity decline and the reduction of cultivated land resources,which restrict the development of rural economy and become one of the factors of poverty in rural mountainous areas.Therefore,the research on the development trend of the future rocky desertification has a certain reference value for Guangxi ecological restoration and rocky desertification control,which is to speed up the rocky desertification.We should speed up the pace of regional economic development,and make a modest contribution to the poverty alleviation efforts in poor areas of Guangxi.With the application of 3S technology in multidisciplinary field,this paper applies 3S technology to the study of rocky desertification.In view of this,under the support of GIS technology,by extracting phenological information of MODIS data,using object oriented classification method to extract rocky desertification information,and then analyze the spatial characteristics of rocky desertification through GWR model,and combine the particles swarm optimization algorithm to mine the transformation rules of cellular automata,and select the influence factors of rocky desertification through testing.In order to simulate the spatial and temporal variation of rocky desertification,the temporal and spatial variation of rocky desertification in the whole area is simulated.The main contents of this study include the following points:(1)Phenological information extraction and optimal data combination: first,compare the original NDVI values in 2010 and 2015,and select the data with the largest difference among different periods.Then use Timesat to extract their S-G filtering and denoising respectively to extract 11 phenological indicators of 2010 and 2015 and establish phenological curves,select suitable phenology and original NDVI to combine,and obtain the best data combination through separability analysis.(2)Object oriented classification: first,e Cognition selects the appropriate segmentation scale to divide and classify the combined image data.Then,the classification results are processed and the accuracy is evaluated.Finally,the area of rocky desertification was statistically analyzed.(3)The analysis of spatial characteristics of rocky desertification: using GWR4 software to analyze the geographical weighted regression analysis of rocky desertification in 2015,and the different positive and negative correlation and influence degree in each county and rocky desertification through the test.(4)The spatio-temporal evolution simulation of rocky desertification: Based on the GIS technology,the particles swarm intelligence optimization algorithm is introduced into the rule mining,and the theoretical basis of cellular automata is combined with the social factors,natural factors and space distance factors of rocky desertification to establish the rocky desertification evolution simulation.Finally,on the basis of the model theory,using Matlab2014 to implement the rule mining and the cellular automata simulation part,the spatio-temporal evolution of Guangxi rocky desertification in 2020 is simulated and compared.(5)Verification of simulation results: In this paper,the accuracy of the traditional Logistic-CA model and the PSO-CA model will be compared and analyzed,and the Kappa coefficient will be used to evaluate the results.The experimental results show that:(1)Using the phenological indicators and MODIS data combination to extract the total accuracy of rocky desertification in 2010 and 2015 reached 80.12% and 82.48% respectively,and the Kappa coefficient reached 0.777 and 0.778.The extracted rocky desertification information was more objectively reflected.The actual situation;(2)The impact of 13 rocky desertification factors on rocky desertification and their spatial distribution were analyzed.In addition to the fail to pass the test of the soil sand clay percentage,the other factors were all tested and R2 reached 0.672,and the number of regression lines of all the influencing factors all showed progressive relationship in different directions.Among them,the evaluation of human impact factors should be evaluated solely by social factors.Combined with the Geographically weighted regression model,the correlation and influence degree of the influence factors and the positive and negative rocky desertification are analyzed,and the reference and treatment direction for the treatment of rocky desertification is provided.(3)The particles swarm optimization algorithm(PSO)was used to conduct rule mining of land use change data from 2010 to 2015.After experimental comparison,the simulation results with 100 iterations were best.At the same time,combining the traditional Logistic-CA algorithm model and the PSO-CA algorithm model to simulate the 2015 results through comparative analysis of accuracy,we found that the overall accuracy and Kappa coefficient of the PSO-CA algorithm and the traditional Logistic-CA model algorithm respectively reach: 96.05%,0.71;95.26%,0.64.In view of this,based on the existing data,this paper uses the PSO-CA algorithm model to simulate the spatiotemporal evolution of rock desertification in Guangxi in 2020. |
PDF Full Text Request