| In this study ecological range unit (ERU) delineations combined with multispectral satellite data were examined to address the need for consistent, spatially accurate, and temporally current methods to inventory rangeland and estimate relative biomass productivity in the context of ecologically sensitive site parameters. ERUs, Landsat 7 ETM+ combined band values, and vegetation index data from 13 scenes acquired from June 2000 to August 2002 were used as predictive variables in linear regression estimates of total biomass using field data collected from 263 locations within 24 ecological range sites on 5 Montana ranches. GIS spatial data analysis techniques were applied to certified soils data themes and published landscape level ecological units to produce the ERU categories used to stratify the field data collection and image analysis, and as a method to test the use of an independent data set for addressing the known influence of soil and site variability on the spectral response of vegetation. ERU categories, in combination with the near and mid-infrared bands (Band 4, 0.75–0.90 μm; Band 7 2.09–2.35 μm), were significant independent variables, and in linear regression predictions collectively explained 66% of the variability in total biomass (p -value < 0.001), as compared to 52% explained by the combined bands alone, suggesting that ERU categories might be accounting for a component of soil variability. This report also introduces an efficient, remote sensing directed method for preliminary identification of locations within ERUs where indicators of soil and site stability or biotic integrity might be outside the established means. A comparison between sites with spectrally anomalous brightness, greenness, and wetness Tasseled Cap indices and selected measurements of spectrally sensitive rangeland ecological health indicators were used to develop a classification method for locating and screening rangeland categories. Pixels where site productivity and exposed soil percentages were within average field ranges (non-anomalous) were identified with an overall accuracy level of 98% (Khat = 0.96). The results of this study support the use of moderate resolution spectral imagery combined with ecological site delineations to enhance efficient rangeland inventory, the effectiveness of rangeland monitoring, and ecologically sustainable management. |
