Remote sensing classification of grass seed cropping practices in western Oregon

摘要: Our primary objective was extending knowledge of major crop rotations and stand establishment conditions present in 4800 grass seed fields surveyed over three years western Oregon to the entire Willamette Valley through classification multiband Landsat images multi-temporal Moderate Resolution Imaging Spectroradiometer (MODIS) 16-day composite Normalized Difference Vegetation Index (NDVI). Mismatch resolution between MODIS data resolved by edging training test validation areas using 3 neighbourhood tests for class uniformity, resampling 50-m followed smoothing artificially enhance resolution, 30 m stacking groups up 64, 55 81 bands 2004-2005, 2005-2006 2006-2007. Imposing several object-based rules raised final accuracies 84.7, 77.1 87.6% 16 categories cropping practices 2005, 2006 2007. Total area under-predicted 3.9, 5.4 1.8% compared yearly Cooperative Extension Service estimates, with Italian ryegrass overestimated an average 8.4% perennial ryegrass, orchardgrass tall fescue underestimated 10.4, 3.3 2.1%. Knowledge field disturbance patterns will be crucial future landscape-level analyses relationships among ecosystem services.