ARS | Publication request: Remote Sensing Classification of Grass Seed Cropping Practices in Western Oregon

Submitted to: International Journal of Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: December 2, 2009
Publication Date: April 29, 2011
Citation: Mueller Warrant, G.W., Whittaker, G.W., Griffith, S.M., Banowetz, G.M., Dugger, B.D., Garcia, T.S., Giannico, G., Boyers, K.L., Mccomb, B.C. 2011. Remote Sensing Classification of Grass Seed Cropping Practices in Western Oregon. International Journal of Remote Sensing. 32:2451-2480.

Interpretive Summary: Remote sensing classification methods are capable of extending our knowledge of crop production management practices across wider areas than it is possible to directly survey. Because such knowledge is crucial for understanding the impact of agriculture on the environment, accuracy of remote sensing classifications is of critical importance. In order to achieve high accuracy in identifying grass seed crops with very similar growth and development patterns, it was necessary to combine satellite images from multiple dates and multiple sensor platforms along with rules describing allowed and prohibited behaviors of crop rotation patterns over time within fields. Our remote sensing classifications of grass seed and rotational crop in western Oregon were used to produce images of field disturbance patterns over multiple years, and to optimize the location of field studies evaluating water quality and the abundance and reproductive success of indicator species, including amphibians, birds, and fish.

Technical Abstract: Multiband Landsat images and multi-temporal MODIS 16-day composite NDVI were classified into 16 categories representing the primary crop rotation options and stand establishment conditions present in western Oregon grass seed fields. Mismatch in resolution between MODIS and Landsat data was resolved by edging of training and test validation areas using 3 by 3 neighborhood tests for class uniformity, resampling of MODIS data to 50-m resolution followed by 3 by 3 neighbourhood smoothing to artificially enhance resolution, and resampling to 30-m for stacking data in groups of up to 64, 55, and 81 bands in 2004-2005, 2005-2006, and 2006-2007. Imposing several object-based rules raised final classification accuracies to 84.7, 77.1, and 87.6%. Total grass seed area was under-predicted by 3.9, 5.4, and 1.8% compared to yearly Cooperative Extension Service estimates, with Italian ryegrass overestimated an average of 8.4% and perennial ryegrass, orchardgrass, and tall fescue underestimated 10.4, 3.3, and 2.1%.