Date of Award

2023-12-01

Degree Name

Master of Science

Department

Environmental Sciences

Advisor(s)

Marguerite M. Mauritz

Abstract

Evapotranspiration (ET) is a critical component of the hydrologic cycle, encompassing both evaporative water loss from surfaces and transpiration through plant stomata. The environmental factors influencing ET include water and energy availability, atmospheric capacity for water uptake, and various meteorological variables. ET serves as a unique climate variable linking water, energy, and carbon cycles. In agroecosystems, accurate ET quantification is vital for optimizing water use efficiency, irrigation management, and crop yield. Traditional methods for ET estimation involve direct measurements and indirect models, with both presenting limitations.

Recent years have witnessed the integration of remote sensing and machine learning (ML) algorithms for ET estimation. While ML models offer flexibility, they may lack physical constraints. Hybrid ML models, combining theoretical foundations and ML predictive power, aim to address this limitation. This study evaluates the applicability of a globally trained hybrid ML model for drylands, specifically natural and agricultural sites. Challenges in quantifying ET in drylands stem from dynamic structures, intermittent water availability, and limited long-term measurements. The evaluation reveals that the hybrid model's performance is deficient for dryland sites, with significant discrepancies in predicted and true ET values. The study underscores the need for further refinement and assessment to develop more physically realistic models for dryland ecosystems.

Language

en

Provenance

Recieved from ProQuest

File Size

54 p.

File Format

application/pdf

Rights Holder

Katya Esquivel Herrera

Download

Share

COinS