Infrastructure Engineering - Theses

Permanent URI for this collection

http://hdl.handle.net/11343/348

Filters

2022 1
1
Reset filters

Settings
Statistics
Citations
End date: 2022 × Start date: 2021 × Type: Masters Research thesis × Subject: Artificial Neural Networks, ×

Search Results

Now showing 1 - 1 of 1
  • Item
    Thumbnail Image
    A Hybrid Framework for Short-term Irrigation Demand Forecasting
    Forouhar, Leila ( 2022)
    Reliable short-term estimates of irrigation water demand can provide valuable information to help water supply system operators with day-to-day operating decisions. Modeling irrigation demand is a complex task due to the different natural (soil, water, crop, and climate interactions) and behavioral (farmers' decision-making) components of the irrigation process. So far, various approaches have been attempted to estimate irrigation water needs in different contexts. Early studies have used simplified physical models to determine irrigation water needs conceptually. However, many of recent studies have applied data-driven methods to map the relationship between the principal influential factors and the water demand. In this study, a generic hybrid framework has been developed to forecast irrigation water demand by integrating a conceptual model (estimating crop water needs based on existing knowledge of the physical system) and a data-driven model (capturing the remaining input-output relationships that cannot be picked up by the conceptual model). The performance of this hybrid framework is evaluated based on real-world system data in Victoria, Australia, and compared to a benchmarking data-driven model (developed using a similar data-driven approach as the hybrid model). It was found that the proposed hybrid framework is able to estimate, with reasonable accuracy, daily irrigation water demand values up to 7 days ahead of the case study system. The hybrid model performs better than the data-driven benchmarking model for most lead times, and particularly for the high-demand period. The results demonstrate that integrating system understanding with data-driven modeling can lead to improved estimates of irrigation water demand. In addition to better predictive performance, the proposed hybrid framework provides improved system understanding and thus increased capacity to support operational decisions.