| As an important part of geographical environment ,crops is a kind of special vegetation types,with a certain degree of the climate,topography,soil conditions correspond,controlled by a variety of factors,has more dependent on the geographical environment,crops are also very sensitive to the changes in other factors in response.Crop type recognition is a fundamental priority of the estimated area of crop, timely monitoring of crops are growing and as well as the early make forecast to crop production, is the basis for remote sensing of agricultural conditions,it can provide relevant information for food production and agricultural development strategy,identify types of crops for policy formulation and to maintain sustainable development, is very important.Normalized Difference Vegetation Index (NDVI) is a parameters that describes the quantity and quality of vegetation, which use of green leaves in the red and near-infrared spectral contrast to design a combination of vegetation index,and there is no dimension. NDVI can monitor the activities of seasonal vegetation growth with the annual and seasonal changes,NDVI is the best instruction factor of the vegetation growth status, NDVI time-series curves can reflect the seasonal changes and human activities.Therefore, NDVI is considered to be an effective ecological indicators of regional or global monitoring vegetation and environmental change.In this paper, Hebei Province as the study area, total of 248 MODIS remote sensing image from October 1, 2008 to September 30, 2009 as the main source of information, following these pretreatment process that are geometric correction, atmospheric radiation correction, cloud detection, NDVI is calculated and synthetized of ten days, remove noise of NDVI time series,and combined of the visual interpretation of Landsat TM images in 2005 land-use types of data, statistical data of Hebei Province as a reference, with RS and GIS for technical support,the seasonal rhythm of crops and biological characteristics of the mechanism is the research foundation, use the level-partition layered approach, carried out a major crop species identification studies.After determining the scope of arable land used land use data,distinguish crops between one year's autumn season and two crops a yearcropping patterns by monitoring the peak frequency of NDVI time series curve. On this basis, according to the different environments on crop growth and yield characteristics of that to be identified crops within the region, further partition, combined with crop phenology characteristics, correspond with the NDVI curve, used ISODATA unsupervised classification methods, finally extracted out of winter wheat and summer maize, cotton, spring maize and naked oats. From a province-wide, Cotton and oat errors are only 3.55% and 0.18%, Classification accuracy is higher, Spring maize error is 17.80%, Winter wheat - summer maize have relatively large errors, but the final classification accuracy is 87%, this result is acceptable for large-scale remote sensing monitoring survey,and mapping the spatial distribution of the major crops.The reshechshown that MODIS data have other advantages such as multi-band, high time, spatial resolution compared with the NOAA / AVHRR data, and the data is relatively easy to obtain, lower cost, with the development of remote sensing technology, MODIS used widely in terrestrial ecosystems, provides a new means of remote sensing for the agricultural condition monitoring, it has great potential for application. NDVI time series data can describeo the crop growth dynamics trajectories vividly and accurately, can demonstrate the process of the rhythm of seasonal changes in crop. Different types of crops, farming calendar is not the same, has a different phenological phases, NDVI curve has different laws, Thus, according to crop planting structure, growth patterns and phenology calendar features, analyze the characteristics and trends of NDVI curve, can determine the type of crops.
|
PDF Full Text Request