Development of Remote Sensing Technologies for Monitoring Rangeland Community Health/Change Using Multispectral and Hyperspectral Satellite Data.

The Alberta Terrestrial Imaging Center (ATIC) was contracted by the Agriculture and Food Council to develop remote sensing technologies for monitoring rangeland community health and change using multispectral and hyperspectral image data. Support for this work was provided by the Advancing Canadian Agriculture and Agri-Food Program, and collaborating government (Agriculture and Agri-Food Canada (AAFC), Alberta Sustainable Resource Development (ASRD), Alberta Financial Services Corporation (AFSC), industry (Iunctus Geomatics) and educational (University of Lethbridge) partners.

Four main project objectives were identified:

(a) to develop methods for tracking rangeland composition and temporal change,
(b) to develop methods for mapping rangeland health indicators,
(c) to create software tools to implement the developed methods and
(d) to disseminate results.

To accomplish these project objectives, ground reference information (fractional cover, species composition, water and chlorophyll content) were collected throughout the two field seasons (2007, 2008) in support of coincident satellite and airborne image data acquisitions. Spectral characterization of dominant rangeland cover types were undertaken for creation of a spectral library and use in the developed technologies for mapping fractional cover of rangeland components. A method for detecting large and small scale changes in rangeland was developed with these methods combined into a complete modular processing chain. This chain contains four main modules (image pre-processing, rangeland delineation, fractional mapping and change detection) to produce product maps for monitoring rangeland health and change using optical satellite and airborne image technologies.

 

Flow diagram of the proposed analysis/processing chain for change detection and mapping fractional cover of dominant rangeland components.

 

Testing and validation of developed technologies show encouraging results. Rangeland extent can be delineated with accuracies from 83-96% using multispectral satellite image data. Fractional cover mapping was conducted using airborne hyperspectral casi data. Cover of shrub and soil corresponded well to field validation data and provided a quantitative evaluation of overall rangeland health in terms of shrub encroachment, soil exposure and erosional effects. Large scale (conversion to annual cropping) and small scale (oil and gas fragmentation) changes can be detected and quantified in terms of area affected using satellite data. These change detection methods provide a means of mapping extensive areas (Alberta native rangeland) in a cost-effective and timely manner.

Quantification of rangeland fragmentation caused by well sites and connecting road networks. Initial SPOT 5 panchromatic image (a) is classified (b) and soil/road classes (red) are selected. Fields are eliminated using shape characteristics for production of final map of well sites and connected road network (c).

 

Large scale rangeland change detection product over Lethbridge area from 1986 to 2009. Initial and final state imagery (top) were classified (middle) into range (yellow) and agriculture (red). Change detection identifies areas that were rangeland in 1986 and have been converted to annual cropping in 2009 (bottom, yellow).

 

The derived products and technologies from this project will facilitate the collection of important information for enhancement of Alberta Sustainable Resource Developments (ASRD) rangeland monitoring capacity. ASRD has expressed interest in implementation of several of the developed technologies to support their Grassland Vegetation Inventory and Land Use Framework Initiatives. Federal government stakeholders have also agreed to conduct a follow on study to research invasive species detection and improve litter content prediction.

In general, the outcome of this project will contribute to a better management of the rangeland ecosystem which in turn will assist in determining the impact of land use over time and the impact of climate change degradation over time by reviewing the archived satellite imagery. It benefits environmental values and supports regulators at all levels of governments in decision and policy making. ATIC will continue to utilize these methods and the framework developed in this project. Continued collaboration with AAFC stakeholders ensures further improvements will be made to the methods using better suited satellite sensors which will be available in the 2012-2015 timeframe. It is expected that these future sensor systems will provide hyperspectral information on an operational basis, and long term goals are to implement these monitoring tools using satellite hyperspectral data.