|Title||A study of regional crop type classification based on MODIS and Radarsat data|
|Author||Hong, G; Zhang, A; Zhou, F|
|Source||ASPRS 2008 Annual Conference, abstract volume; 2008 p. 1|
|Alt Series||Earth Sciences Sector, Contribution Series 20080481|
|Publisher||American Society for Photogrammetry and Remote Sensing|
|Meeting||ASPRS 2008 Annual Conference - American Society for Photogrammetry and Remote Sensing; Portland, OR; CA; April 28 - May 2, 2008|
|Subjects||geophysics; Agriculture; Nature and Environment; remote sensing; satellite imagery; land use; mapping techniques; methodology; crop yield estimation; cropland mapping; MODIS; Radarsat-1|
Geoscience, Climate Change Impacts and Adaptation for Key Economic and Natural Environment Sectors|
|Released||2008 04 01|
|Abstract||Crop land-use mapping is very important for crop acreage measurement and crop yield estimation. An effective regional crop land mapping by remote sensing requires large image coverage with sufficient
resolutions in spatial, spectral and temporal dimensions. But few earth observation sensors can meet all these requirements.|
The MODIS (Moderate Resolution Imaging Spectroradiometer) satellite imaging is one of the optimal data sources because of
its high spectral resolution, large coverage, daily revisit rate, and low cost. However, its spatial resolutions are too coarse to delineate crop field boundaries. The Canadian Radarsat-1 images are a good data source for obtaining spatial
information with high spatial resolutions (from 3 m to 100 m depending on the beam mode) at a frequent repeat rate, because of its all whether and all day collection capability, low cost, and large coverage (from 50kmx50km to 500kmx500km depending on
the beam mode); also it has a plenty of structural information, which makes it possible to delineate the field boundary. However, Radarsat images are noisy and have limited spectral information.
To find a cost-effective solution for frequent,
large coverage crop land-use mapping, this paper presents a study on combination of low spatial resolution MODIS multispectral images and high spatial resolution Radarsat amplitude images for crop classification at a regional level. An integrated
technique of Wavelet and IHS (Intensity, Hue and Saturation) was developed to fuse the multispectral information from MODIS and the spatial information from Radarsat-1, to produce an image which possesses the strengths of MODIS and Radarsat-1 images.
Object-based classification technique is then used to classify the MODIS-Radarsat fused images for obtaining crop land-use classes. Object-based classification of the fused images yields promising result for regional crop land-use mapping.