2008 Joint Annual Meeting (5-9 Oct. 2008): Salinity Assessment of Large Spatial Extents Using MODIS Imagery and Electromagnetic Induction Directed Soil Sampling.

The 2.3 million ha Red River Valley (RRV) straddling eastern North Dakota and western Minnesota and extending into Canada is perceived to be increasing in soil salinity due to changing weather patterns and land management, which in turn affect landscape-scale hydrologic processes resulting in increased root zone salinity levels. Assessing the spatial distribution of salinity in the RRV is an essential tool for NRCS’s soil survey program, not only in the RRV but other areas of the USA. To assess salinity over large spatial extents (i.e., hundreds of thousands of ha), satellite imagery (i.e., MODIS) and electromagnetic induction (EMI) were used. The approach was evaluated in the RRV’s Kittson County, MN. Enhanced vegetation index (EVI) values from MODIS imagery collected from 2000-2006 were averaged over the summer growing season for all years. The average summer EVI for the 7-yr period (EVIAV) was used as a stratification variable. The spatial EVIAV readings were separated into 20 distinct classes. Fields were randomly chosen from within each class and their mean ground-truth salinity levels (MGTSLs) were determined using EMI to direct soil sampling. We hypothesize that if a strong correlation can be established between MGTSLs and corresponding EVIAV values, then it should be possible to use the EVIAV data to predict salinity levels across Kittson County (and hopefully the larger Red River Valley). Results show that there is an acceptable correlation between MGTSLs and EVIAV for Kittson County, except in those areas where landuse is the Conservation Reserve Program (CPR). Ostensibly, the approach is not able to assess salinity levels on CPR land correctly because of the natural vegetation diversity, which is distinct from the predominantly monoculture found in Kittson County (i.e., spring wheat). Future work will evaluate the approach for 2 additional counties in the RRV to establish its broader application over areas with significantly different soil properties.