Progress 12/15/98 to 12/15/04
Outputs
The influence of pre-fire vegetation and topographic variables (slope, aspect, and elevation) on burn severity, as well as the effects of burn severity and topography on post-fire vegetation recovery was conducted in a 1986 burn that consumed roughly 17,000 ha. Vegetation was classified prior to the burn (1986) and at 16 years post-burn (2002), and a chronosequence of Landsat TM and ETM+ images was analyzed to estimate vegetation recovery. Burn severity, estimated using a remotely-sensed index, was strongly controlled by vegetation, with needleleaf forests experiencing much higher burn severity than broadleaf forests, shrubs, and woodlands. Elevation also had an influence on burn severity, presumably due to its control on vegetation composition. Vegetation recovery peaked 8-14 years post-fire, and recovery was highest for areas with the highest burn severity. Topography had a lesser influence on recovery than burn severity. Self-replacement of the existing vegetation
type was common at the burn and was generally highest in areas with moderate burn severity. Shifts between needleleaf and broadleaf forest were very rare, and in general the area shifted toward an open needleleaf woodland and shrubland, especially in areas experiencing high burn severity. Manuscript "Influence of Vegetation Type and Topography on Burn Severity in interior Alaska" submitted, revised and resubmitted to Canadian Journal of Forest Research
Impacts
The spatial and temporal variability of wildland fires in Alaska influences landscape dynamics and is critical toward understanding the mosaic nature of the interior region. Variability in burn severity creates differences in substrate and seedbed quality, which leads to the differential germination and survival of boreal forest species. This research demonstrated that areas vegetated with spruce forest experienced higher burn severity than broadleaf forests and unforested areas. Moreover, this pattern was density dependent, with increasing density resulting in increasing burn severity. The results supports the hypothesis that broadleaf patches can act as fire breaks on the landscape.
Publications
- Epting, J. F. 2004. Remote Sensing of Burn Severity and the Interactions Between Burn Severity, Topography, and Vegetation in interior Alaska M. S. Thesis, Dept. of Forest Sciences, University of Alaska Fairbanks.
- Verbyla, D. L. 2005. Assessment of the MODIS Leaf Area Index Product in Alaska. International Journal of Remote Sensing. 2005(1-9) in press
- A.D. McGuire, M. Apps, F.S. Chapin III, R. Dargaville, M.D. Flannigan, E. S. Kasischke, D. Kicklighter, J. Kimball, W. Kurz, D.J. McRae, K. McDonald, J. Melillo, R. Myneni, B.J. Stocks, D. L. Verbyla, and Q. Zhuang. 2004. Chapter 9. Land Cover Disturbances and Feedbacks to the Climate System in Canada and Alaska. Land Change Science: Observing, Monitoring, and Understanding Trajectories of Change on the Earths Surface. Series: Remote Sensing and Digital Image Processing, Vol. 6 Kluwer Academic Publishers. ISBN: 1-4020-2561-0
- Csiszar, I. C. O. Justine, A. D. McGuire, M. A. Cochrane, D. P. Roy, F. Brown, S. G. Conrad, P. G. H. Frost, L. Giglio, C. Elvidge, M. D. Flannigan, E. Kasischke, D. J. McRae, T. S. Rupp, B. J. Stocks, and D. L. Verbyla. 2004. Land use and fires. Chapter 19. Land Change Science: Observing, Monitoring, and Understanding Trajectories of Change on the Earths Surface. Series: Remote Sensing and Digital Image Processing, Vol. 6 Springer. Kluwer Academic Publishers. ISBN: 1-4020-2561-0
|
Progress 01/01/03 to 12/31/03
Outputs
Spruce stands become highly susceptible to wildfire following bark beetle infestation. Intensity of wildfire has immediate impact on carbon emission, and long term impact on carbon flux, post-fire regeneration, permafrost dynamics and associated subsurface hydrology. Therefore accurate remote sensing of fire severity is important for many applications. Remotely sensed algorithms such as vegetation indices, ratios and transformations has been developed in lower latitudes, but never tested in high latitude boreal forests to map burn severity classes. I compared these algorithms to map burn intensity at recent burns on a floodplain site and an upland site in interior Alaska. The floodplain site was the Survey Line site that burned during the summer of 2001, and the upland site was from the Yukon-Charlie National Reserve which burned during the summer of 1999. Thirteen Landsat TM and ETM+ algorithms were evaluated for these sites in interior Alaska. The algorithms
included single bands, band ratios, vegetation indices, and multivariate techniques, and each algorithm was evaluated as both a single date and bi-temporal model. Based on correlation analysis and field sampled verified assessment of classified burn severity maps, the Normalized Burn Ratio was consistently the most accurate algorithm for mapping burn intensity classes.
Impacts
Accurate estimates of burn severity across large spatial and temporal scales are important as inputs for estimating carbon emissions from wildfire and for modeling and prediction of post-fire plant succession, permafrost dynamics, and carbon flux. This research demonstrated the Normalized Burn Ratio can be reliably applied in high latitude boreal forests of Alaska to map burn severity classes. Results from this research will be used in a workshop on geostatistical methods in GIS to be taught in Fairbanks and Anchorage next winter. A manuscript based on this research will soon be submitted to the Canadian Journal of Remote Sensing.
Publications
- Dissing, D. and D. L. Verbyla. 2003. Spatial patterns of lightning strikes in interior Alaska and their relations to elevation and vegetation. Canadian Journal of Forest Research. 33:770-782.
- Stow, D. A., Hope, A., McGuire, D., Verbyla, D., Gamon, J., Huemmrich, F., Houston, S., Racine, C., Sturm, M., Tape, K., Hinzman, L., Yoshikawa, K., Tweedie, C., Noyle, B., Silapaswan, C., Douglas, D., Griffith, B., Gensuo, J., Epstein, H., Walker, D., Daeschner, S., Peterson, A., Zhou, L. and R. Myeni. 2003.Remote sensing of vegetation and land-cover change in arctic tundra ecosystems. Remote Sensing of Environment. 89(3):281-308.
|
Progress 01/01/02 to 12/31/02
Outputs
Progress: A new remote sensing product, the MODIS Leaf Area Index 8-day product was evaluated for three tiles in Alaska. 8-day leaf area index values were acquired in hdf format, projected to an Alaska map projection, and merged into grids covering most of the state at 1-km grain size. The positional accuracy of the product was assessed by comparing the locations of recent wildfire burns with the leaf area images. The temporal pattern of spring greenup to fall senescence was tested along latitudinal and elevation transects and was reasonable. However, the leaf area index estimates in general were too high. Also unreasonable estimates were seen when a maximum seasonal leaf area was mapped across the state of Alaska. The reason for these high estimates are due to the algorithm being dependent on biome land cover maps which are not accurate in Alaska and also due to high near infrared reflectance from broadleaf shrub and forest communities in the boreal forest and tundra
of Alaska. A manuscript "Verbyla, D. L. Assessment of the MODIS LAI Product in Alaska" was submitted to the International Journal of Remote Sensing. Results from this research will be presented in a users workshop at the American Society of Photogrammetry and Remote Sensing Annual Conference in May 2003.
Impacts
Accurate estimates of forest leaf area are important for assessing forest damage by insects, wildfire danger, and forest productivity and carbon dynamics. This research demonstrated that new estimates of leaf area from the new MODIS sensor are not reliable in Alaska. The use of Alaska MODIS leaf area estimates could lead to false predictions by simulation models of forest productivity and regional carbon budgets.
Publications
- No publications reported this period
|
Progress 01/01/01 to 12/31/01
Outputs
Optical remote sensing of insect damage at high latitudes suffers from the following problems: 1) Significant cloud cover due to long daily periods of evapotranspiration. 2) Even more significant cloud shadow due to relatively low solar elevation. 3) Variable dust and smoke due to glacial and wildfire disturbances. Radar remote sensing does not utilize solar radiation and therefore does not suffer from these problems. In theory, short band (C-band and shorter) may be useful to detect changes in canopy structure due to defoliation and changes in water status due to bark beetle infestation. To test the potential of radar remote sensing, RADARSAT and ERS-1 radar images were acquired for the periods of pre and post spring leaf flush. Despite its theoretical advantages, there was no consistent trend in backscatter in any radar sensor or forest type (spruce or broadleaf forest). Radar backscatter increased, decreased, or remained constant for various stands of spruce and
broadleafs. Potential confounding factors that may influence backscatter include freeze/thaw conditions and standing water/ice within stands. Because of confounding factors, I concluded that ERS-1 and RADARSAT satellite data are not useful for monitoring insect infestations within the boreal forest.
Impacts
Although short-band radar remote sensing from aircraft may be useful for mapping and monitoring insect infestations in the boreal forest, current space borne satellites such as RADARSAT and ERS-1 are not useful for mapping of insect infestations because the resolution is too coarse and confounding factors that influence radar backscatter such as freeze/thaw and water conditions may mask any backscatter signal from insect infestations.
Publications
- Verbyla, D. L. 2001. A test of detecting spring leaf flush within the Alaskan boreal forest using ERS-2 and Radarsat SAR data. International Journal of Remote Sensing. 22:1159-1165.
|
Progress 01/01/00 to 12/31/00
Outputs
Most satellite change detection techniques for estimating insect infestations rely on multi-temporal images...typically images taken before and after a heavy infestation. These multi-temporal images are spatially co-registered as a pre-requesite before any change detection analysis is started. A critical assumption is therefore no positional error in the co-registration of multi- temporal images. This is not possible because there is always residual error in image co-registration models. The assumption is usually that sub-pixel co-registration error is not important. Under some conditions, this assumption is not true. Using Landsat Thematic Mapper and Advanced Very High Resolution Radiometer imagery, I found that small co-registration error in change detection analysis can be very important as a bias source in estimating change. This false change is especially large in change estimates from heterogeneous images. I tested a statistcal resampling techinque, the
bootstrap, to estimate the bias associated with positional error in multi-temporal change detection studies.
Impacts
Any study that uses historic remotely sensed imagery to estimate area that has changed or defoliated should be aware of the significant potential to overestimate area of defoliation due to positional error inherent in multi-temporal images.
Publications
- Verbyla, D. L. and S. H. Boles. 2000. Bias in land cover change estimates due to misregistration. International Journal of Remote Sensing. (in press)
|
Progress 01/01/99 to 12/31/99
Outputs
In this study, before and after-defoliation Landsat TM imagery were acquired along with ground plot information from the National Park Service for the Copper River Valley, Wrangell St. Elias National Park. The data were radiometrically rectify to control for spectral variation due to atmospheric and viewing geometry differences. The images were then co-registered based on at least 15 control pixels with a linear affine transformation model error of less than 0.5 pixels. Bilinear interpolation was be use for pixel resampling during the image rectification. Only spruce-dominated stands will be sampled and mapped in this study. To eliminate non-spruce pixels, all non-spruce pixels from the 1986 image have been masked out. This mask will be applied to all images so that all subsequent analysis will be exclusive with spruce pixels. A simulation study was completed to analyze the impact of co-registration positional error and to investigate the precision of a bootstrap
approach of estimating this false change. This study is published on the web at www.lter.alaska.edu/dverbyla/change_detection and submitted as a manuscript.
Impacts
Satellite remote sensing may provide a more detailed landscape-level map of these infestations at relatively small grain size (30 meters) on medium-scale maps. In the next five years, new satellite technology such as imaging radar satellites, Landsat-7 Enhanced Thematic Mapper (1999), the hyperspectral ASTER sensor (2000) may be useful and as a relatively inexpensive improvement over existing systems.
Publications
- Verbyla, David. 1999. Estimating Bias in Change Detection Using Bootstrap Resampling www.lter.uaf.edu/~dverbyla/change_detection/
|
|