USGS - science for a changing world

Northern Prairie Wildlife Research Center

  Home About NPWRC Our Science Staff Employment Contacts Common Questions About the Site

Effects of Climate on Numbers of Northern Prairie Wetlands

Results: Testing and Validation

The models were robust with respect to random and systematic deletion of data. Random deletion of 10% of the data resulted in Pearson product moment correlations between observed and predicted percent wet basins of 0.81, 0.79, and 0.80 for parkland, Canadian grassland, and United States grassland, respectively. Similarly, average correlations after deletion of one transect were 0.81 (range 0.80 - 0.81), 0.80 (0.79-0.80), and 0.80 (0.79-0.80) for parkland, Canadian grassland, and United States grassland, respectively.

Strata 34 and 35 contain a total of 11 transects, all classified as parkland. Climate data for these transects were used to obtain predicted wet basin counts for 1968 through 1990 using the parkland model. Predicted and actual wet basin counts were correlated well (r = 0.77).

Because different variables were important in parkland and grassland, it was necessary to determine the degree to which the models differed. To examine this question, I ran the parkland model on Canadian grassland data, and the Canadian grassland model on parkland data. For grassland data in the parkland model, the correlation coefficient between observed and predicted wet basins was 0.70, compared with 0.80 for grassland data in the grassland model. For parkland data in the grassland model, r = 0.73, compared with 0.81 for parkland data in the parkland model.

Previous Section -- Results: Models
Return to Contents
Next Section -- Results: Effects of Changing Temperature and Precipitation

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo logo U.S. Department of the Interior | U.S. Geological Survey
Page Contact Information: Webmaster
Page Last Modified: Saturday, 02-Feb-2013 06:58:07 EST
Sioux Falls, SD [sdww55]