Sabrina Karim

Date of Award

January 2025

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Geography & Geographic Information Science

First Advisor

Bradley Rundquist

Abstract

Climate variability and weather patterns play a critical role in shaping agricultural land use and land cover (LULC), particularly in states like North Dakota, where agriculture dominates the landscape. This study explores the impacts of climatic factors—specifically rainfall, temperature, and Growing Degree Days (GDD)—on agricultural land cover (ALC) changes in North Dakota from 1997 to 2023, with projections for 2033. By combining historical analysis and predictive modeling, the research examines how climate variability has influenced crop dynamics and provides insights into future scenarios. The research uses advanced geospatial tools, including Google Earth Engine, and machine learning techniques such as Random Forest (RF), Autoregressive Integrated Moving Average and Long Short-Term Memory networks. Data mainly from the U.S. Department of Agriculture Cropland Data Layer and the North Dakota Agricultural Weather Network form the foundation for this analysis, linking climate variables such as rainfall, air temperature, GDD and bare soil temperature to ALC patterns. The findings reveal a significant shift toward monoculture practices, particularly the increasing dominance of spring wheat and corn, at the expense of sunflower, soybeans and other. While this shift enhances short-term agricultural productivity, it poses challenges to crop diversity, soil fertility, and long-term ecological resilience. Spatial analyses highlight strong clustering in the distributions of spring wheat and corn, whereas sunflower demonstrates fluctuating spatial dynamics and a declining trend. Projections for 2033 also indicate intensified hotspots of crop cultivation in southeastern and northeastern regions, driven by rising GDD and temperature. These trends underscore the potential risks to food security and the vulnerability of monoculture systems to climatic stressors. Using RF for ALC predictions, the study achieved an overall accuracy of 97 percent and a Kappa coefficient of 0.95 for crop prediction model, demonstrating the robustness of machine learning techniques in agricultural modeling. These findings underscore the need for adaptive management strategies to mitigate risks associated with climate variability and offer valuable contributions to sustainable agricultural planning and resilience.

Download

Share

COinS