首页>
外文会议>IAC;International Astronautical Congress
>ASSESSING CROP WATER DEMANDS FROM SPACE: CLASSIFICATION OF IRRIGATION SYSTEMS IN ARID CENTRAL ASIA USING LATEST OPTICAL REMOTE SENSING SYSTEMS
【24h】
ASSESSING CROP WATER DEMANDS FROM SPACE: CLASSIFICATION OF IRRIGATION SYSTEMS IN ARID CENTRAL ASIA USING LATEST OPTICAL REMOTE SENSING SYSTEMS
For sustainable water management in irrigation-dominated river basins information about the crop distribution isnecessary to estimate the crop water demands. Remote sensing techniques offer the opportunity to determine cropdistribution in large study areas within well known error boundaries. This research aimed on the identification of thecrop acreage based on a high resolution land use classification of multi-temporal remote sensing data. A first step ofthe classification process was the automatic segmentation of RapidEye Images (6.5 m) to extract the agriculturalfields on which the classification is based. To assess the accuracy of the field-boundary delineation a comparison ofthe area and shape of the segmentation results with randomly chosen manually digitized reference polygons wasapplied. Subsequently, a per-field classification method was used to separate the agricultural fields from other landuseclasses. To validate the classification results the user's, producer's and overall accuracy was calculated.Afterwards, a classification using a randomForest classifier was used to detect the different field crops such ascotton, wheat, maize, rice and fruit trees. In this process a cross-validation was applied to calculate the accuracy ofthe crop classification. Subsequently, the information on crop type and acreage can be used to determine the cropwaterrequirement; for example by using the FAO CropWat model. Therefore, this methodology allows forsupporting the implementation of water allocation plans, assessing the productivity of irrigation systems and thus forincreasing the efficiency of water use in all irrigation systems worldwide.
展开▼