Scale Detection & Extraction Tools
Level 1 Users are users with little GIS and remote sensing experience or access to specialized GIS or remote sensing software or hardware. Level 1 users will mostly interact with Toolbox-generated products and web applications and may work with level-2 or level-3 users to use other Toolbox tools for their management information needs.
Level 2 Users have an understanding of GIS and rangeland monitoring data and have experience with GIS software. Level 2 users can do some minor data creation (e.g., creating layers to define new management scenarios) or edit existing layers (e.g., correcting and updating polygons in the vegetation maps). They will use Toolbox tools and products to aid analyses for their management questions and may also run new scenarios based on existing state and transition models and vegetation maps. These users may provide outputs to level 1 users but may also require assistance from level 3 users to set up datasets for some tools.
Level 3 Users have a highly-developed skill set for manipulating and analyzing GIS, remote sensing, and assessment/monitoring data. They will have access to specialized software and hardware for creating new data layers (e.g., creating image objects from new imagery) and would set up the data layers that level 1 and 2 users would have access to. These users could also lead the process of creating new state and transition models or vegetation maps for new areas. Use of Landscape Toolbox tools in major planning efforts (e.g. RMP) would rely heavily on level 3 users.Software tools built on object-based image analysis for detecting and extracting ecologically-meaningful scale patterns.
WHAT ARE THE SCALE DETECTION AND EXTRACTION TOOLS?
One of the challenges to successfully using remotely-sensed data to answer rangeland management questions is separating patterns of interest from all of the other information contained in the image. The Landscape Toolbox Scale Detection and Extraction tools identify the different scales of patterns present in an image, and, through a new method called object-based image analysis, can extract these patterns for use in rangeland analysis.
HOW DO YOU GET/USE THEM?
The scale detection and extraction tools are intended for technical users who are comfortable with GIS and image analysis and who have access to ArcGIS 9.1 and Definiens Developer 7.0 or higher. For our case study areas in southern Idaho, the image objects resulting from running the scale detection and extraction tools can be downloaded from the Landscape Toolbox Data Center and used in other Toolbox applications.
The scale detection and extraction tools are a set of process steps and scripts that run in Definiens Developer and ArcGIS. The tools and technical documentation can be downloaded here.
HOW DO THEY WORK?
The scale detection and extraction tools are built on a relatively new technique called Object-based image analysis (OBIA). OBIA differs from traditional pixel-based remote sensing techniques in that pixels are first grouped together into “objects” according to how similar the image reflectance values of neighboring pixels are to each other through a process called segmentation. The objects then become the primary unit of analysis rather than the pixel. By adjusting the amount of variability allowed within an object, segmentation can produce small or large objects that correspond to different scales of patterns captured in an image.
Because the objects are defined based on the image data, they reflect differences in vegetation and landform on the ground. As the objects become larger and larger, their meaning can change from representing individual plants or small patches of vegetation to larger patches or mosaics of communities. The scale-detection tools look at how the variability of pixels within an object changes relative to the variability between objects to determine which sets of image objects are most likely to represent patterns and minimize unrelated information – or noise. Once the right set of image objects has been identified, the objects can be exported, the patterns interpreted according to ecological processes, and used in other applications like sample selection and design, vegetation mapping, or cumulative landscape assessment.