Publication Date

12-2014

Date of Final Oral Examination (Defense)

10-15-2014

Type of Culminating Activity

Thesis

Degree Title

Master of Science in Hydrologic Sciences

Department

Geosciences

Supervisory Committee Chair

Alejandro N. Flores, Ph.D.

Supervisory Committee Member

Kelly Cobourn, Ph.D.

Supervisory Committee Member

Jennifer Pierce, Ph.D.

Abstract

Climate change has raised concerns about the interplay between agricultural productivity, water demand, and water availability in semi-arid to arid regions of the world. As these regions cover nearly 41% of the Earth’s surface and are home to more than 38% of the total global population of 6.5 billion, it is important to understand the implications of changes to water use and water availability on the civilizations and industries that rely upon already scarce water resources. Currently, irrigated agriculture is the dominant water user in these regions and is estimated to consume approximately 80% of the world’s diverted freshwater resources. Future climate change is anticipated to produce increased variability in precipitation, including a reduction in winter snowfall in areas such as the Snake River Basin of Southern Idaho. It is therefore important to discern how irrigated land-use (water use) is changing on an annual basis to improve water management practices and to be able to deduce factors (both natural and social) leading to changes in practices. Current methods for mapping irrigated land-use either lack sufficient accuracy or are time intensive, costly practices. This study aims to create an improved irrigated land-use mapping technique using remote sensing observations, which could not only reduce data processing time and cost, but also increase the temporal resolution at which irrigated areas are monitored. Using USDA Census of Agriculture county-scale irrigated area data from 2002 and 2007 as validation, our model was able to produce area-weighted average percent errors for the study region of 2.73% and 6.29%, respectively. When considering classification error at a regional scale, this is an improvement on the results from the widely accepted method of using single date imagery to classify irrigated land-use, which produced area-weighted average percent errors for the study region of 33.46% and 36.71%, respectively. Individual county correctness varied on an annual basis, with the accuracy being a direct correlation to the quantity and accuracy of observation locations chosen. Increasing the quantity of observation locations within each county should reduce the effect of observation point classification uncertainty on model accuracy, hopefully leading to improvements in water use accounting and helping advance understanding on the impacts of changes to irrigated land-use towards food security, economic effects, and environmental impacts.

Download

Share

COinS