Research On Periodicity Analysis And Crop Planting Classification Based On NDVI In The Heihe River Basin

Green biomass covers nearly three-fourths of the Earth’s land surface and play a keyrole in global energy, hydrological and biogeochemical cycles. Satellite remote sensingprovides spatially and temporally continuous measures of the dynamics change ofvegetation coves. Vegetation indices calculated from satellite data is useful formonitoring global climate changes. The Heihe River Basin is typical and the secondinland river basin in the arid zone of Northwestern China. The natural condition is harshand the ecological environment is fragile. Therefore, vegetation cover which is the mostsensitive factor to environment plays an important role in the Heihe River Basin.The time series NDVI (Normalized Difference Vegetation Index) data in the HeiheRiver Basin were analyzed and applied in two aspects.1. Periodicity analysis of time series NDVI and its relationship with climatic factorsin the Heihe River Basin.The VEGETATION NDVI during the period from2007to2009was converted tothe GIMMS NDVI for the same period. Then the SINDVI (Seasonally Integrated NDVI)was calculated based on the extended GIMMS NDVI data set. The periodicity analysisbased the SINDVI and meteorological data during1982-2009was performed in theHeihe River Basin by using a novel multiple time-scale analysis method, EMD(Empirical Mode Decomposition). At the same time, the relationship among SINDVI, airtemperature and precipitation was analyzed. The results indicate that SINDVI,temperature and precipitation have the similar3and10year quasiperiodics in the upperreaches of the Heihe River Basin. SINDVI and temperature have the similar3,10yearquasiperiodics, SINDVI and precipitation have the similar3,6,8and15yearquasiperiodics in the middle reaches of the Heihe River Basin. At the meantime, in thelower reaches of the Heihe River, SINDVI and temperature have the similar3and10year quasiperiodics, SINDVI and precipitation have the similar3and6yearquasiperiodics. It is indicated that the temperature and precipitation are both the driving factor affecting the vegetation cover in the Heihe River Basin. In addition, the EMDmethod can be effectively used to analyze the relationship between time-series data withthe meteorological data.2. Crop planting structure classification based on time series NDVI in the middlepart of the Heihe River Basin.The time series NDVI data set with three-day temporal resolution was composedbased on MODIS09250m reflectance products from September2010to October2011.Then140scenes of composed images were integrated into an image data cube of timeseries NDVI, which was used to extract the crop planting structure for the middle ofHeihe River Basin. The typical spectral curves of the vegetation features were generatedfor corn, wheat, rapeseed, vegetables and other crops based on in-suit measurements andexpert knowledge of visual interpretation. The major crop types were classified by usingthe Adaboost (Adaptive Boosting) and SVM (Support Vector Machine) algorithms. Theresults showed that the classification accuracy of Adaboost was87.06%while the SVMwas74.48%, and the Kappa coefficient of Adaboost is0.8477while the SVM was0.6988.The classification results could express the spatial distribution of the main crops as wellwhile the mixed pixel decomposition can provide more accurate information. Theaccurate crop planting structure information is of great significance to improve andoptimize the land surface models and other applications.