Assessing climate change impacts on hydrology and agriculture in a boreal watershed: a combined hydrological modeling, machine learning, and efficiency analysis

Loading...
Thumbnail Image

Keywords

hydrological modeling, climate change impacts, machine learning, boreal watershed, agriculture

Degree Level

doctoral

Degree Name

Ph. D.

Volume

Issue

Publisher

Memorial University of Newfoundland

Abstract

Climate change poses significant challenges to water resources and agriculture sustainability, particularly in boreal watersheds. Understanding hydrological dynamics and their response to climate variability is essential for effective water management and sustainable farming practices. This dissertation uses various models and a transdisciplinary approach to investigate the hydrological dynamics, climate change impacts, and agricultural sustainability in the Upper Humber River Watershed (UHRW) of western Newfoundland, Canada. Using the soil and water assessment tool (SWAT) model, the study evaluates watershed hydrology and identifies key parameters influencing streamflow, achieving favorable performance metrics during calibration and evaluation. Seasonal and monthly flow patterns, water balance components, and flow duration analyses validate the model's suitability for sustainable water management. Similarly, a Long Short-Term Memory (LSTM) machine learning model was developed for UHRW. The comparative analysis of the SWAT and LSTM models for streamflow prediction highlights the superior accuracy in capturing streamflow prediction. The LSTM model’s integration of real-time data demonstrates its potential for effective water management in cold climates. Furthermore, this dissertation explores climate change impacts on soil water availability (SWA) in UHRW using Coupled Model Intercomparison Project Phase 5 (CMIP5) projections and Representative Concentration Pathways (RCP) scenarios. Results reveal up to an 11% decline in SWA under RCP 8.5, driven by increased evapotranspiration and streamflow, despite rising precipitation. Lastly, agricultural sustainability is assessed through Data Envelopment Analysis (DEA), identifying high technical efficiency yet notable disparities in allocative, cost, scale, and environmental efficiencies. Sustainable practices, such as permaculture, no-dig farming, and resource optimization, enhance productivity while minimizing environmental impacts. This dissertation offers insights for water management, climate adaptation, and sustainable farming in boreal ecosystems.

Collections