COS 21-1 - Image stratification as a means to guide field data collection efforts and develop comprehensive field data sets

Monday, August 7, 2017: 1:30 PM
B114, Oregon Convention Center
Kenneth A. Stumpf, Geographic Resource Solutions, Arcata, CA

The development of comprehensive field data sets that represent the diversity of ecological types found with large project areas that may range in size from thousands of acres to millions of acres can be problematic. Such large area efforts necessitate sample design considerations that will ensure that field sampling efforts are thorough, complete, and economical. Such field sampling efforts should be representative of the project area in terms of their distribution throughout the project area, as well as across the variety of ecosystems or “types” to be sampled. Frequently occurring “types” should not be oversampled and rare or infrequently occurring “types” should not be undersampled or missed. Field samples should not be overly concentrated in small portions of the project area or only in areas of easy access. Field sampling plans should be developed that result in the efficient and economical collection of field data.

Image stratification was implemented during vegetation classification projects at both Lassen Volcanic National Park and Redwood National and State Parks to guide field data collection efforts.


Image stratification enabled the development of a “candidate site” database in which between 150-200 “types” were typically recognized. Each stratification effort yield information describing the location and relative abundance of each “type.” Such information readily identified the rare, infrequently occurring “types” from the abundant “types.” Spatial information also enabled the identification of the larger contiguous areas of each different “type” and its proximity to points of access. Daily and weekly field sampling plans were based on sampling the rarely occurring areas and filling out the day’s plan with nearby frequently occurring “types.” Sampling plans were developed with the use of a GIS and target locations were identified and loaded into GPS units to guide field crews to the proper location of the sample areas. The “types” were sufficiently sampled across the landscape and in each park some new “types,” not found during prior sampling efforts, were found and sampled. Sampling efforts were completed within the forecast time and budget estimates.

Image stratification proved to be a valuable sample design tool to guide field data collection efforts for large area projects of this nature. Examples from these projects will be presented to demonstrate these concepts.