| Title | Arable Land Information Extraction Using MODIS Multi-temporal Data |
| Abstract | Information about cultivated land area, crop phenology and vegetation change intensity were got from MODIS NDVI time series .Data of crop growing area, cultivated land ratio, topography were analyzed using dynamic clustering method so as to make district planning of cultivating phenology of Zhejiang province.In this way ,the neural network training samples were selected based on the districting result above. Zhejiang northern plain district was finally chosen as research region and the NDVI time series data of the research region was got.MODIS product deriving and data preprocessing were introduced .Causation of MODIS image noise and distortion was pointed out and the corresponding rectification algorithms，including the vegetation index algorithms ,were appraised .The TM data was processed as follows: radiometric calibration ,geometric calibration and finally maximum-likelihood classification in order to get the cultivated land validation data.The BP and RBF neural network were both used as a method to extract cultivated land area. The training function of BP neural network and spread value of RBF neural network were experimented to valid the influence on the performance of neural network. The final extraction of neural network was analyzed using grid-analysis with the help of TM data. The extraction result got from RBF neural network was better than that from BP neural network in binary output mode ,while the opposite result was got in the multi-valued output mode. The highest correlation coefficient between cultivated land area extracted from MODIS data and TM data was 0.92.The NDVI time series were treated using discrete fourier transformation, and the maximum value of time series was used to analyze the crop phenonology consistency.The result shows there is a good consistency of rice phenology,while the wheat and rape has low consisitency toward the maximum point of NDVI time series ,which may result from the existence of rape florescence and the unconsistency of wheat and rape growing periods in study area.Discrete fourier transformation was implementing to make the analysis about intensity of vegetation cover change. The result shows human activities and anthoropogenic planting phenologies change have greatly influences the intensity of vegetation cover change. |
| Category | Internet |
| Keywords | change vector analysis, disctrete fourier transformation, land cover, neural network, phenology information, |
| FileType | |
| Pages | 135 |
| Price | US$60.00 |
| Buy Now | |
| Download | |
| Contact |
E-Mail:itpaper@hotmail.com TEL:1-888-786-998A |
| FAQ |
How to get this paper's electronic documents? 1, Click the "Buy Now" button to complete the online payment 2, Download the paper's electronic document from the successful payment return page/Or the system will send this paper's electronic document to your E-Mail within 24 hours |
| Favorite | ADD TO FAVORITE |
| Version | zh-cn |
Arable Land Information Extraction Using MODIS Multi-temporal Data
Billy HU, Ph.D
E-Mail:itpaper@hotmail.com
TEL:1-888-786-998A
