A cross combined land use classification procedurefor remote sensing detection of irrigated crops

摘要

In the irrigation planning, the calculation of crops water requirements(CWR) is a key element. The available algorithms and calculation models need of verydetailed land use data, considering that CWR can change between different varieties ofthe same cultivation. Furthermore, due to the rotation/crop shifts, the irrigated land usesare significantly variable from year to year, especially in horticulture areas.To respond to this need a methodology and a classification systembased on the combineduse of airborne hyper-spectral and high resolution scanner (CASI) and LANDSAT TMhasbeen developed and applied on a test site in Sardinia (Italy). The methodology is based onthe definition of clusters of crops with similar WRand spectral firm, and the combination ofautomatic image classification and video-interpretation techniques.The use of CASI (Compact Airborne Spectrographic Imager) allowed to classify cropswith a combined cluster analysis (crops with similar CWRand spectral signature) to derivea classification applicable to commercial mediumresolution satellites image processing foragricultural water use monitoring.The complete procedure is based on the use of CASI data only sporadically, for cropthematic and geometric classification, cluster definition, and validation purposes, andLANDSAT TM images yearly or seasonally for cluster detection.