acase@tulane.edu

The principal objective of this dissertation is to develop Bayesian machine learning models for multiple reservoir, irrigation canal, and river system operation and management. These types of models are derived from the emerging area of machine learning theory; they are characterized by their ability to capture the underlying physics of the system simply by examination of the measured system inputs and outputs. They can be used to provide probabilistic predictions of system behavior using only historical data. The models were developed in the form of a multivariate relevance vector machine (MVRVM) that is based on a sparse Bayesian learning machine approach for regression. Using this Bayesian approach, a predictive confidence interval is obtained from the model that captures the uncertainty of both the model and the data. The models were applied to the multiple reservoir, canal and river system located in the regulated Lower Sevier River Basin in Utah. The models were developed to perform predictions of multi-time-ahead releases of multiple reservoirs, diversions of multiple canals, and streamflow and water loss/gain in a river system. This research represents the first attempt to use a multivariate Bayesian learning regression approach to develop simultaneous multi-step-ahead predictions with predictive confidence intervals for multiple outputs in a regulated river basin system. These predictions will be of potential value to reservoir and canal operators in identifying the best decisions for operation and management of irrigation water supply systems.

Although most of the earth is covered in water, a very limited amount of that water is fresh water, which is essential to our survival. Therefore, it is imperative that we do all that is possible to conserve and protect our extremely limited water resources, especially in arid regions such as the American West. While there are many ways and means to protecting and preserving our water resources, this thesis focuses on the strategy of rainwater harvesting (RWH) as it is done throughout the state of Utah. RWH is defined as taking the precipitation that falls on our built structures and putting it to good use when it would often otherwise end up in gutters, pipes, and storm drains to be processed and/or disposed of at a distant location. RWH systems consist of several different components including a catchment area (usually, but not always, a rooftop), gutters, a place for storage, and some way of future dispersal and use of the collected water. There are numerous documented benefits to RWH. Historically, RWH has not been allowed in states—such as Utah—that follow the doctrine of prior appropriation, which strongly defends the case for water rights and affirms that senior water rights should not be infringed upon. According to the law, when someone practiced RWH they were infringing on the water rights of others. However, in 2010, the Utah State Legislature modified these long-standing laws to allow residents of Utah to legally harvest up to 2,500 gallons at a time without fear of infringement on others’ water rights. Since then, many Utahans have adopted RWH. However, the number of Utah residents who are practicing RWH is still a tiny percentage of the entire Utah population. This research included surveying self-identified rainwater harvesters throughout Utah and sought to discover and understand their motivations for adopting the practice. This information has the potential to assist planners, water districts, water managers, cities, state agencies, and legislators in persuading others throughout the state to also adopt the practice. It was found that an emergency supply of water and concern for the environment are the most important motivators for Utah rainwater harvesters. Unsurprisingly, financial savings also had a significant influence on harvesters and their decision to practice RWH, although they spent considerably less on their RWH systems than is typically spent in other states and countries. A secondary aspect of the research was to examine Utah State Senate Bill 32, the current law in Utah concerning RWH. It was found that the current laws are written in a restrictive manner and should be changed and adjusted in order for a greater percentage of the Utah population to be motivated to adopt RWH.

This dissertation evaluates the water allocation challenges in the rural river basins of the developing world, where demands are growing and the supply is limited. While many of these basins have yet to reach the state of closure, their water users are already experiencing water shortages. Agricultural crop production in rural river basins of the developing world plays a major role in ensuring food security. However, irrigation as the major water consumer in these basins has low water use efficiency. As water scarcity grows, the need to maximize economic gains by reallocating water to more efficient uses becomes important. Water allocation decisions must be made considering the social economic and environmental conditions of the developing world. The purpose of this dissertation is to identify water allocation strategies that satisfy the above conditions, in the example of the Walawe River basin in Sri Lanka. In this dissertation three manuscripts are presented. The first manuscript takes a broad view of the current water allocation situation. The second manuscript develops a methodology to analyze water allocation under a priority-based approach with the use of network flow simulation techniques. The third manuscript analyzes the water supply-demand situation in the basin under future climatic conditions. The major findings of this study suggest that: (1) while up to 44% of water is still available for use, seasonality of inflows, poor water management, physical infrastructure deficiencies, and other socio-economic factors contribute to the irrigation deficits in the Walawe basin; (2) prioritizing irrigation over hydropower generation increases supply reliability by 21% in the Walawe irrigation system IRR 1. The corresponding annual loss in power output in less than 0.5%. Prioritizing the left bank irrigation area in system IRR 2 increases the economic gains from crop yields by US $1 million annually; (3) an increase of water use efficiency between 30-50% in agriculture can mitigate all water deficits in agriculture, urban water supply and industrial sectors; (4) the predicted 25% increase of rainfall over the Walawe basin in the 2050's allows for 43% increase in hydropower generation (with changes to power generation mode) and 3-16 % reduction in irrigation requirements; (5) network flow simulation techniques can be successfully used to evaluate different demand management strategies and improvements to the priority-based water allocation method.

The effects of roots on soil-moisture measurements taken with a neutron probe were studied. These effects were observed under three different soil-moisture conditions, with different sized roots, and with varying distances between the access tube and root. The moisture conditions used were air dry, field capacity, and saturated. Large roots can greatly affect neutron soil-moisture measurements when the access hole is drilled through the root. Positive errors as large as 52 percent were found when the soil moisture was at field capacity. With dry and saturated conditions positive errors of 43 percent and 38 percent respectively, were found. In most practical field situations, where the access hole is not drilled through large roots, root material appears to have very little effect, if any, on neutron soil-moisture measurements. The largest positive error found, when the access holes were not drilled through the roots, was only 8 percent. This occurred at a point where the access tube was in contact with the outside of a root

Water temperature is an important variable for aquatic ecosystems. Salmonid population numbers and distribution are heavily influenced by stream temperature, and there is growing concern about the health of salmonid populations with anticipated climate change. Managers are looking to efficiently evaluate options to maintain stream temperatures needed by salmonids. This study evaluated and compared stream temperature restoration alternatives in two streams with warm temperatures using stream temperature monitoring and modeling. The first study identified pockets of cold water that are important to native fish species in Nevada’s Walker River. Comparison of monitoring results with existing basin-scale model outputs identified two habitat features, beaver dams and irrigation return flow channels, that maximize stream temperature variability. Restoration should maintain and enhance these features, although different restoration approaches may be needed at different locations. This study may provide guidance for the interpretation of stream temperature results from other basin-scale models. The second study quantified stream temperature effects of wildfire and restoration plantings in Oregon’s Meadow Creek with current and projected mid-21st century climate. A stream temperature model developed and applied using Heat Source found restoration eliminated days above the lethal threshold (25°C) for salmonids and decreased the number of days exceeding spawning criteria during spawning periods. Days exceeding salmonid spawning (13°C) and rearing (18°C) thresholds were reduced by all vegetation restoration scenarios, but elimiated by none. Results highlights the importance of the length and location of restoration, which can maximize pockets of cold water for salmonids or alleviate the impact of warm water sections.

The water that is available for beneficial use in Utah is quickly approaching full appropriation; water that has been claimed is nearing the amount that is available for use. The Division of Water Resources of the State of Utah has organized a three-part plan to "Plan, Conserve, Develop and Protect Utah's Water Resources." One of these three elements has a focus to "provide comprehensive water planning." Such planning is best achieved when current and accurate data on the uses of the state's water are available. The primary purpose of this thesis was to provide an evaluation, from data collected on various case studies across the state, on the accuracy of water rights information. The studies were selected based on land that had recently been developed from agricultural usage to residential, commercial, or other municipal uses. After identifying the accuracy of the water rights information, observations to the methods of recording water right transfers were made. A template to summarize a municipality's water rights will also be made available.

This research seeks to understand how wetlands maintain a water supply in the Bear River Basin, where water is generally scarce. Research was conducted through semi-structured interviews with wetland and water experts in the basin and archival research of historical documents and water rights. The U. S. Fish and Wildlife Service manages three refuges on the Bear River, and has obtained water rights portfolios for each. Holding water rights does not ensure that there will be water available for refuge wetlands. Instead, position in relation to other powerful water users is the most important factor in determining the security of a refuge's water supply and the threats faced from drought. All refuges must manage their water because the human-hydrology of the river is complex and variable; this requires a combination of infrastructure and planning. Maintaining relationships with other water users is another important adaptation to the human-hydrology of the river, because all water users along the river are interconnected. Recognizing that they face the same threats to their water supply allows wetland managers and irrigators to cooperate in order to maintain the water supply for their region of the river and increases adaptability as the region faces climate change. The Bear River Migratory Bird Refuge is the oldest refuge on the river and has the least secure water supply, despite having the largest water rights portfolio. Because it is chronically short of water during the summer, refuge staff have developed an adaptive management strategy to effectively utilize the water they do receive. Management involves predicting water supplies each year, setting water level targets accordingly, actively diverting water to priority wetlands, and allowing non-priority wetland to dry. This is followed by extensive monitoring of habitat conditions and bird use, the results of which are shared in annual management plans. This strategy maintains the most wildlife habitat possible and offers important institutional adaptations. Most importantly, it demonstrates the refuge's water rights are being put to beneficial use. Sharing knowledge gained through management also builds trust and adaptive capacity among water users facing the complex human-hydrology at the end of the Bear River.

Northern Utah Valley is one of the fastest growing areas of Utah and has increasing needs for domestic, industrial, and agricultural water. To meet these needs, groundwater and surface water systems must be understood to maximize their use. Chemical studies of the sediment mineralogy and related water-chemistry give insight to the movement of the water. There are three major aquifers present in the valley: shallow Pleistocene; deep Pleistocene; and Tertiary. They are composed of sands and gravels and are separated by confining layers (aquitards) composed mostly of clay. Along the flanks of the bordering mountains there are undifferentiated aquifers which act as conduits supplying water for aquifers in the valley. Sediment samples from aquifers and confining layers were obtained by rotary and cable-tool drilling. X-ray diffraction analyses showed that the aquifers are mainly composed of quartz, calcite, and dolomite, whereas the con fining layers contain illite and montmorillonite with some kaolinite, quart z and calcite. One hundred nine water samples were collected in this study from s ur face water, spring water, undifferentiated aquifer water, shallow Pleistocene aquifer water, deep Pleistocene aquifer water and Tertiary aquifer water. Results show that the ground water system has several geochemical cells in each aquifer, due to diverse areas of recharge. Three major water types can be identified in different areas of the shallow Pleistocene aquifer, three in different areas of the Tertiary aquifer, and four in different areas of the deep Pleistocene aquifer. The differences in these water types are related to the composition of the mountain recharge areas and positions of faults within the valleys. The aquifer composition exerts relatively little influence on the chemistry of the ground water. Mountains of predominantly carbonate rocks produce recharge waters rich in calcium and bicarbonate. Mountains of predominantly granitic rocks produce recharge water low in mineral content. Valley sediments near major faults produce highly mineralized waters.

River barriers, such as dams, culverts and diversions are important for water conveyance, but disrupt river ecosystems and hydrologic processes. River barrier removal is increasingly used to restore and improve river habitat and connectivity. Most past barrier removal projects prioritized individual barriers using score-and-rank techniques, neglecting the spatial structure and cumulative change from multiple barrier removals. Similarly, most water demand models satisfy human water uses or, only prioritize aquatic habitat, failing to include both human and environmental water use benefits. In this study, a dual objective optimization model identified in-stream barriers that impede quality-weighted aquatic habitat connectivity for Bonneville cutthroat trout. Monthly streamflow, stream temperature, channel gradient and geomorphic condition were indicators of aquatic habitat suitability. Solutions to the dual objective problem quantify and graphically present tradeoffs between quality-weighted habitat connectivity and economic water demands. The optimization model is generalizable to other watersheds, but it was applied as a case study in Utah’s Weber Basin to prioritize removal of environmentally-harmful barriers, while maintaining human water uses. Modeled results suggest tradeoffs between economic costs of removing barriers and quality-weighted habitat gains. Removing 54 in-stream barriers increases quality-weighted habitat by about 160 km and costs approximately $10M, after which point the cost effectiveness of removing barriers to connect river habitat slows. In other words, there is decreasing benefit of removing barriers, so that after removing the first 54 barriers, it costs more to connect more high-quality habitat. Removing reservoirs or diversions that result in large economic losses did not substantially increase habitat. This suggests that removing numerous small barriers results in greater increases in habitat for the same removal costs, without significant water scarcity losses. The set of barriers prioritized for removal varied monthly depending on limiting habitat conditions for Bonneville cutthroat trout. The common barriers removed in the model were identified to communicate the most environmentally harmful barriers to local stakeholders and inform decision-making. Additionally, limiting the budget or number of barrier removal projects resulted in a different set of barriers removed. This research helps prioritize barrier removals and future restoration decisions in the Weber Basin although the model formulation is generalizable to other watersheds. Available data and a simplified approach limit the scope of this model. The modeling approach expands current barrier removal optimization methods by explicitly including economic and environmental water uses.