Multiple Species Conservation Program (MSCP) Habitat Monitoring: Remote Sensing Research at San Diego State University
|
Multiple Species Conservation Program (MSCP) Habitat Monitoring: Remote Sensing Research at San Diego State University |
Initial Results
The team at SDSU performs geometric processing with the multi-temporal imagery for the purpose of precisely registering and aligning the data prior to performing change detection. Initial conclusions from the first year of the project are:
The accuracy of spatial registration for
multi-temporal ADAR imagery (1 m resolution) varies between 1 to 5 meters
and is dependent on topographic variability.
Such registration accuracy limits the detection of land cover change to features that are larger than a single pixel (i.e., 1 meter). If pixel-level change detection is desired then interactive visual approaches should be used and automated procedures avoided.
SDSU also is evaluating methods of radiometric processing for year-to-year normalization of imagery, correction of anisotropic reflectance within airborne image data, and normalization of phenology related variations in brightness between multi-temporal imagery. The team has found that:
Matching the histograms of multi-temporal
imagery by matching the mean and standard deviations of the individual
wavebands is an efficient method of normalization which reduces the
influences of phenology related variations in brightness but enables
detection of local land cover changes.
A model developed by SDSU and implemented
in the Spatial Modeler of ERDAS Imagine effectively corrects within-scene
variations in brightness associated with anisotropic reflectance.
However, the process is computationally and time intensive.
Differences in time of day and time of year in imaging as well as year-to-year variations in precipitation impact the ability to detect and identify changes of interest to habitat reserve managers.
The project has evaluated many methods of change detection including: spectral waveband differencing, vegetation index differencing, fraction image differencing, and change vector analysis. Results are consistent between the various methods and highlight land cover changes associated with: loss of vegetation due to clearing, trampling, and fire; regrowth of vegetation; and differences in phenological state. Repeat observation and change detection to be completed in the future will enable assessment of the potential for detecting long-term changes related to habitat degradation.
The project seeks to develop relationships between image-derived products and ground variables associated with habitat condition and quality. Ground data was collected at two sites imaged during Spring 2000. Preliminary results indicate that:
Percent bare ground cover can be reliably estimated and mapped from spectral indices derived from ADAR imagery collected at 1 meter resolution.
Identification of unauthorized trails is central to the management of reserves as it allows managers to enforce access restrictions and reduce impacts to protected areas. The project is using image processing techniques to enhance and extract unauthorized trail locations from 1 meter ADAR 5500 imagery.
SDSU is evaluating the utility of 4 meter resolution multispectral imagery for detecting region-wide change. The project is utilizing simulated 4 meter data derived from 2 meter ADAR 5500 airborne data until IKONOS imagery collected during Spring 2000 is acquired and processed. Initial change detection results from 4 m multispectral data suggest that:
Landscape-level changes in land use and land cover should be effectively monitored with IKONOS multispectral (4 meter resolution) imagery; sub-pixel disturbance features are likely to be detectable by spectral mixture analysis.
Brief overviews of some of the required image processing and ground sampling procedures outlined above can be found in the linked pages below. Additionally, links to two online GIS demonstration pages can be found on the change detection page below. These two GIS demos utilize data derived from multi-temporal ADAR imagery to highlight changes detected at the Mission Trails Regional Park (MTRP) study site.