Uncertainty Analysis for Assessment of Conversion from Flood to Sprinkler Irrigation in Montana

This project provides files used in the uncertainty analysis of polygon areas resulting from overlaying/clipping/erasing GIS operations that map the conversion from mid-21st century flood (and sprinkler irrigation) to sprinkler irrigation (center-pivot and other sprinkler), and other land types (fallow, crop, and flood remaining flood) in Montana, by 2019.

This project is part of a larger effort that maps the conversion from mid-20th century flood (and sprinkler irrigation) to sprinkler irrigation (center-pivot and other sprinkler), and other land types (fallow, crop, and flood remaining flood) in Montana, by 2019.
This file contains results of mapping the conversion from mid-20th century flood (and sprinkler irrigation) to sprinkler irrigation (center-pivot and other sprinkler), and other land types (to cropland—C, hayland--H, fallow –FA, and sprinkler remaining sprinkler) in Montana, by 2019. As part of the Montana Water Center’s effort to better understand the effects of increased irrigation efficiency in Montana (Lonsdale et al. 2020), historic conversion from flood to sprinkler irrigation was analyzed using available agricultural statistics, maps from state and federal sources, and an independent Geographic Information Systems (GIS) analysis. This project presents the GIS analysis and maps the amount and spatial distribution of conversion from flood to sprinkler irrigation, between the mid-20th century and 2019. Historic mid-20th century irrigation was mapped in detail from 1943-1965 by the State Engineer’s Office and from 1966-1971 by the Montana Water Resources Board—the predecessor of the Montana Department of Natural Resources and Conservation (DNRC). A scanned and georeferenced version of the Water Resources Surveys (WRS) was compared with maps of contemporary irrigated land (Montana Department of Revenue’s 2019 Final Land Unit Classification—DORFLU2019) to estimate the area of land converted from flood to sprinkler irrigation. Prior to GIS analysis, both datasets were edited to ensure valid comparison between irrigated field mapping conducted at the two points in time. To estimate the amount of conversion from flood to sprinkler irrigation, and other uses, the GIS layers (WRS flood and sprinkler 1946-1971 and DOR-FLU 2019) were overlain in ArcGIS; then the clipping erase functions were used to select the WRS flood and sprinkler parcels that were shown as sprinkler irrigated in 2019. Additional conversion classes were also mapped that represent the changes from WRS flood and sprinkler to cropland, hayland and fallow, and WRS sprinkler remaining sprinkler.

There are several sources of uncertainty in the conversion mapping results. The first is that the analysis only accounts for changes that occurred between the WRS 1946-71 and DORFLU2019; it is possible that additional flood irrigation developed between the two points in time may have also been converted to sprinkler. Lacking statewide mapping of irrigation for intervening years, it was not possible to evaluate this. In addition, WRS were not available for several counties, and the amount of conversion could not be estimated. Although several of the counties are in eastern Montana and have little irrigation, Beaverhead and Yellowstone Counties have significant irrigation and could have significant conversion--therefore the statewide estimate of conversion should be considered a minimum value.

Another source of uncertainty is due to GIS processing and overlay/clip/erase functions that create “sliver” polygons of apparent change due to misalignment of the WRS 1946-71 and DORFLU2019 layers (i.e. co-registration error). This was evaluated using the spatially distributed probabilistic (SDP) method of Leonard and others (2020) and found to be small—generally less than one percent of the area of conversion polygons. Digitizing error was evaluated indirectly and found to be about ±12 percent of the reported area values. The values sum in quadrature to provide an overall estimate of error in polygon area of 12%.

Conversion from flood to sprinkler polygon areas presented in the report, and associated error statistics, apply to the whole dataset at the statewide scale. For use at the basin scale (for example, HUC4 Upper Yellowstone, the end user should review the uncertainty estimate for specific conversion polygons and refine if necessary.

Please see Appendix D_Uncertainty analysis.pdf for details of the analysis: https://www.hydroshare.org/resource/51957cd254b54891ba2e239428bd132d/data/contents/Appendix_D__Uncertainty_Analysis.pdf