HR: 1340h
AN: A13B-0101    [Abstracts]
TI: Testing MODIS Capability to Map Peatlands
AU: Pflugmacher, D
EM: Dirk.Pflugmacher@oregonstate.edu
AF: University Of Applied Sciences, Friedrich-Ebert-Str. 28, Eberswalde, 16225 Germany
AU: * Krankina, O
EM: krankinao@fsl.orst.edu
AF: Oregon State University, Richardson Hall 321, Corvallis, OR 97331 United States
AU: Cohen, W
EM: warren.cohen@oregonstate.edu
AF: USDA Forest Service, 3200 SW Jefferson Way, Corvallis, OR 97331 United States
AB: Although peatlands cover only about 3 percent of our planet's land and freshwater surface, they play a major role in the global carbon cycle. Despite their importance, peatlands are largely overlooked in current large-scale vegetation mapping efforts. In this study, we investigated the potential of the MODIS sensor to capture extent and distribution of peatlands in the St. Petersburg region of Russia. We used Forest Inventory and Peat Deposit Inventory data to extract characteristics of 50 peatlands and map them across the study area. The mapped peatlands vary in size from 0.6 ha to 7710 ha and represent three major nutritional types (oligotrophic, mesotrophic, eutrophic). In addition, parts of 6 peatlands have been used for peat-mining and these were mapped separately. MODIS Nadir BRDF-Adjusted Reflectance data from 2002 at a spatial resolution of 1-km was used to empirically model continuous estimates of percent peatland and percent mined peatland. We extracted spectral information from MODIS pixels that were completely covered by our mapped areas. In total we sampled 1105 pixels, where peatlands occupied from 0 to 100 per cent; 50 per cent of these pixels were used for regression analysis and the other half for model validation. We employed a canonical correlation analysis on the MODIS reflectance bands, producing a multi-spectral percent Peatland Cover Index (PCI). The PCI was then used as the predictor in a Reduced Major Axis (RMA) regression model. Our results suggest high potential for peatland mapping with MODIS. The RMA regression models explained much of the variance in the PCI (r2 = 0.86 for mined and r2 = 0.81 for unmined peatlands). The validation showed a high correlation between observed vs. predicted peatland cover (mined: r = 0.87; unmined: r = 0.92). We will produce a peatland map for the study area from the developed PCI models and compare it with MODIS land cover products.
DE: 1694 Instruments and techniques
DE: 1640 Remote sensing
DE: 0400 Biogeosciences
SC: Atmospheric Sciences [A]
MN: 2004 AGU Fall Meeting