GRS has developed specialized algorithms and processes to enable the development of plant community and species/landscape feature-specific attributes from digital imagery. Our approach, called Discrete Classification, enables the use, evaluation, and identification of individual species and landscape characteristics found on the ground in the image data set. This methodology enables the mapping of species-specific cover, as well as the assignment of classification system names like the National Vegetation Classification System (NVCS) association levels names. GRS has developed Confusion and Fidelity Reports to identify sources of confusion, in terms of botanical/land cover characteristics and verify the fidelity of classification efforts before any actual image classification maps are ever developed. The ability to deal with individual training site data is a tremendous advantage over the more traditional clustered training data approach that is so often used. Training site data can come from a myriad of sources, old and new. GRS has also performed projects requiring the manual interpretation and extraction of GIS features from digital photographic images as well as high-resolution satellite images. GRS image classification projects have resulted in the creation of complex GIS layers with associated database tables consisting of detailed quantitative descriptions of classified features.
While Discrete Classification is typically viewed as a means of mapping plant communities and individual species/landscape feature attributes, it also enables the mapping of other pertinent features that support the use of these data sets by foresters, ecologists, fire scientists, wildlife biologists, resource planners, and other resource professionals. Examples of additional features that may be mapped or additional analyses and capabilities include:
1. Fire fuel counts by size and decay class may be mapped to generate Fuel Model names and estimates of biomass/tonnage of the different types of down woody material.
2. Species-specific cover estimates by canopy layer enable the development of type names using the National Vegetation Classification System, or other naming conventions by simply processing the species cover estimates using the appropriate type naming key. No crosswalking is necessary using this type of map data set.
3. Wildlife habitat suitability studies and analyses can be performed based on both type names/designations as well as the individual species-specific/landscape feature cover estimates.
4. Accuracy assessments can be performed at both the type and individual species levels by comparing actual cover estimates with mapped estimates.
5. Tree volume and biomass may be mapped. Stand level attributes can include species-specific estimates by both size class, canopy layer, and in total of stems per acre, quadratic mean diameter, volume, tonnage, and height height.
Recent GRS projects that have developed these integrated resource information map data sets include the Galena Forest Inventory and Planning Project, the BLM Tonsina Valley Forest Biomass and Mapping Project, the Redwood National and State Parks Vegetation Classification and Mapping Project, and the Lassen Volcanic National Park Comparative Mapping Project.
Using satellite imagery, GRS is capable of mapping anywhere in the world. Currently available images, from mid-range resolution (like Landsat) to high-resolution, enable GRS to develop accurate and reliable vector, raster, and attribute data. GRS utilizes satellite imagery for land cover mapping, change detection analysis, forest land inventory, topographic modeling, and as a baseline data for future evaluation.
GRS is a leader in the use of Image Classification techniques to produce vegetation and land cover maps. Using proprietary techniques for image processing, GRS has successfully mapped millions of acres to a highly accurate level of detail. GRS is able to map specific vegetation estimates for timber size, species composition, percent vegetation cover, and vegetation structure.
GRS has developed GRS_covmatrixsum and GRS_aggregate for support of our innovative image processing techniques. These two processes are instrumental in the processing of image class maps and the aggregation of large scale raster data into smaller scale vector-based polygon databases.
GRS provides end-to-end solutions for landcover mapping including all aspects of the process from planning to reporting.
- Project Planning and Needs Assessment
- Field Sample Design
- Field Sampling Logistics
- Field Data Collection
- Vegetation and Ecological Ordination
- Image Acquisition
- Image Compilation and Quality Assurance
- Image Correction/Terrain Normalization
- Image Classification
- Pixel Aggregation into Polygons
- Accuracy Assessment Sample Design
- Accuracy Assessment
- Map Production