Phenologic Assessment of Western South Dakota Rangelands

Assessing the health of rangeland ecosystems based solely on the annual biomass production does not fully describe plant community condition, as biodiversity and species composition are also critical components in sustaining rangeland systems. The phenology of production is of particular concern in assessing rangeland health due to inferences that can be made on species composition. In this study a variety of phenological indicators were derived from examination of the phenological profiles of sample sites using 2000 to 2008 Moderate Resolution Imaging Spectrometer (MODIS) imagery at 250 m resolution in the Bad River watershed of western South Dakota. Analysis revealed that biomass production decreased with precipitation from east to west across the study area. Alternatively, cool season percentage of total production increased from east to west. Cool season biomass production averaged 76.8% of total which compared favorably to previously reported values for the study area (Tieszen et al. 1997; Foody and Dash 2007). This study, however, provided a much higher spatial resolution than previous work. Additionally, this study reports the spatial, temporal, and spatial-temporal phenologic patterns, each of which offers a unique perspective. Accurate maps of biomass production and dominant photosynthetic pathways are useful to land managers. These maps increase the efficiency of management by denoting areas were degrading practices were occurring. Validation of remotely sensed observations with field data is critically important to any remote sensing study. Remotely sensed phenology and biomass data for western South Dakota rangelands was determined using MODIS normalized difference vegetation index (NDVI) imagery. Validation of remotely sensed data was conducted through a variety of field data sets 1) pasture scale long term plant community and biomass data, 2) gross photosynthesis data from a carbon flux tower, 3) visual obstruction (Robel pole) measurements, 4) Daubenmire vegetation cover data, 5) Soil Survey Geographic (SSURGO) range production values, and 6) State Soil Geographic (STATSGO) C3 percentage. Varying degrees of success were achieved in the six validation methods. Vegetation cover by Daubenmire frame provided a poor relationship with remotely sensed data. Robel pole and pasture-scale validation proved to be moderately related to remotely sensed data. Remotely sensed and flux tower metrics yielded similar results. Remotely sensed time-integrated NDVI (TIN) was strongly related (R2 = 0.87) to SSURGO range production data. Finally, cool season percentage had a moderately robust relationship (R2 = 0.45) with STATSGO C3 percent reference data. These comparisons demonstrated the proficiency of medium resolution remotely sensed data to accurately reflect in situ and reference data