A hybrid machine learning model to predict and visualize nitrate concentration throughout the Central Valley aquifer, California, USA

Science of the Total Environment (Elsevier) | June 9th, 2017

Intense demand for water in the Central Valley of California and related increases in groundwater nitrate concentration threaten the sustainability of the groundwater res

Intense demand for water in the Central Valley of California and related increases in groundwater nitrate concentration threaten the sustainability of the groundwater resource. To assess contamination risk in the region, we developed a hybrid, non-linear, machine learning model within a statistical learning framework to predict nitrate contamination of groundwater to depths of approximately 500m below ground surface. A database of 145 predictor variables representing well characteristics, historical and current field and landscape-scale nitrogen mass balances, historical and current land use, oxidation/reduction conditions, groundwater flow, climate, soil characteristics, depth to groundwater, and groundwater age were assigned to over 6000 private supply and public supply wells measured previously for nitrate and located throughout the study area. The boosted regression tree (BRT) method was used to screen and rank variables to predict nitrate concentration at the depths of domestic and public well supplies. The novel approach included as predictor variables outputs from existing physically based models of the Central Valley. The top five most important predictor variables included two oxidation/reduction variables (probability of manganese concentration to exceed 50 ppb and probability of dissolved oxygen concentration to be below 0.5 ppm), field-scale adjusted unsaturated zone nitrogen input for the 1975 time period, average difference between precipitation and evapotranspiration during the years 1971–2000, and 1992 total landscape nitrogen input. Twenty-five variables were selected for the final model for log-transformed nitrate. In general, increasing probability of anoxic conditions and increasing precipitation relative to potential evapotranspiration had a corresponding decrease in nitrate concentration predictions. Conversely, increasing 1975 unsaturated zone nitrogen leaching flux and 1992 total landscape nitrogen input had an increasing relative impact on nitrate predictions. Three-dimensional visualization indicates that nitrate predictions depend on the probability of anoxic conditions and other factors, and that nitrate predictions generally decreased with increasing groundwater age.

A hydrological simulation dataset of the Upper Colorado River Basin from 1983 to 2019

Nature Portfolio (Springer Nature) | January 20th, 2022

The Upper Colorado River Basin (UCRB) is one of the most over-allocated basins in the world. It provides water for 40 million people in Colorado and downstream states of

The Upper Colorado River Basin (UCRB) is one of the most over-allocated basins in the world. It provides water for 40 million people in Colorado and downstream states of Arizona, California, Nevada, and Utah. However, the UCRB water capacity is decreasing as a combined result of climate change and anthropogenic activities. Understanding the dynamic of the UCRB water cycle is crucial for long-term and short-term water resources management.

A multi-city COVID-19 forecasting model utilizing wastewater-based epidemiology

Science of the Total Environment (Elsevier) | January 15th, 2025

A multiscale model for El Niño complexity

Nature Portfolio (Springer Nature) | March 11th, 2022

El Niño-Southern Oscillation (ENSO) exhibits diverse characteristics in spatial pattern, peak intensity, and temporal evolution. Here we develop a three-region multiscal

El Niño-Southern Oscillation (ENSO) exhibits diverse characteristics in spatial pattern, peak intensity, and temporal evolution. Here we develop a three-region multiscale stochastic model to show that the observed ENSO complexity can be explained by combining intraseasonal, interannual, and decadal processes. The model starts with a deterministic three-region system for the interannual variabilities. Then two stochastic processes of the intraseasonal and decadal variation are incorporated. The model can reproduce not only the general properties of the observed ENSO events, but also the complexity in patterns (e.g., Central Pacific vs. Eastern Pacific events), intensity (e.g., 10–20 year reoccurrence of extreme El Niños), and temporal evolution (e.g., more multi-year La Niñas than multi-year El Niños). While conventional conceptual models were typically used to understand the dynamics behind the common properties of ENSO, this model offers a powerful tool to understand and predict ENSO complexity that challenges our understanding of the twenty-first century ENSO.

A New Technique for Modeling Land Subsidence Facilitates Better Groundwater Management

Stanford University | December 1st, 2019

Land subsidence – the sudden sinking or gradual settling of Earth’s surface – can occur naturally or be triggered by human activity. One of the most comm

Land subsidence – the sudden sinking or gradual settling of Earth’s surface – can occur naturally or be triggered by human activity. One of the most common human-affected factors is groundwater pumping, a practice that has been steadily increasing due to prolonged periods of drought. Groundwater pumping is so prevalent in some areas that it is severely compromising the quantity and quality of the groundwater as well as the physical structure of the land and aquifer capacity beneath it. Large-scale initiatives for putting water back into the ground, or groundwater recharge, are being implemented in some regions. However, groundwater resource managers still need better information to determine where the land is at greatest risk of subsidence as well as where recharge efforts would be the most effective. Integrating two reliable data gathering sources – helicopter-deployed systems (airborne electromagnetic or AEM technology) that measure electromagnetic fields and satellite-deployed systems (interferometric synthetic aperture radar or InSAR) to measure deformations – offers groundwater managers an improved method for more accurately modeling changes in the land surface related to the pumping and recharge of groundwater. Additionally, the ability to more precisely detect subsidence may offer an early warning system for declining groundwater levels.

A Rapid Assessment Method to Identify Potential Groundwater Flooding Hotspots as Sea Levels Rise in Coastal Cities

Water, Multidisciplinary Digital Publishing Institute (MDPI) | October 25th, 2019

Sea level rise (SLR) will cause shallow unconfined coastal aquifers to rise. Rising groundwater can emerge as surface flooding and impact buried infrastructure, soil beh

Sea level rise (SLR) will cause shallow unconfined coastal aquifers to rise. Rising groundwater can emerge as surface flooding and impact buried infrastructure, soil behavior, human health, and nearshore ecosystems. Higher groundwater can also reduce infiltration rates for stormwater, adding to surface flooding problems. Levees and seawalls may not prevent these impacts. Pumping may accelerate land subsidence rates, thereby exacerbating flooding problems associated with SLR. Public agencies at all jurisdiction levels will need information regarding where groundwater impacts are likely to occur for development and infrastructure planning, as extreme precipitation events combine with SLR to drive more frequent flooding. We used empirical depth-to-water data and a digital elevation model of the San Francisco Bay Area to construct an interpolated surface of estimated minimum depth-to-water for 489 square kilometers along the San Francisco Bay shoreline. This rapid assessment approach identified key locations where more rigorous data collection and dynamic modeling is needed to identify risks and prevent impacts to health, buildings, and infrastructure, and develop adaptation strategies for SLR.

A real-time groundwater management model using data assimilation

American Geophysical Union (AGU) | June 30th, 2011

We apply the proposed management model to the Aquifer Storage and Recovery (ASR) Project of the Las Posas Groundwater Basin in southern California. The adapted nudging al

We apply the proposed management model to the Aquifer Storage and Recovery (ASR) Project of the Las Posas Groundwater Basin in southern California. The adapted nudging algorithm is used to estimate the missing pumping rates of private wells. We then use the estimated pumping rates in a real-time groundwater management model to improve management decisions of the ASR well fields.

A Simulation-Optimization Model for Water-Resources Management, Santa Barbara, California

U.S. Geological Survey (USGS) | September 1st, 1998

A simulation-optimization model has been developed for the optimal management of the city of Santa Barbara's water resources during a drought. The model, which links grou

A simulation-optimization model has been developed for the optimal management of the city of Santa Barbara's water resources during a drought. The model, which links ground-water simulation with linear programming, has a planning horizon of 5 years. The objective is to minimize the cost of water supply subject to: water demand constraints, hydraulic head constraints to control seawater intrusion, and water capacity constraints. The decision variables are monthly water deliveries from surface water and ground water. The state variables are hydraulic heads. The drought of 1947–51 is the city's worst drought on record, and simulated surface-water supplies for this period were used as a basis for testing optimal management of current water resources under drought conditions. The simulation-optimization model was applied using three reservoir operation rules. In addition, the model's sensitivity to demand, carryover [the storage of water in one year for use in a later year(s)], head constraints, and capacity constraints was tested.

A Spatial Hedonic Analysis of the Housing Market around a Large, Failing Desert Lake: The Case of the Salton Sea in California

Journal of Environmental Planning and Management (Taylor & Francis) | March 2nd, 2018

Many lakes around the world exhibit acute environmental stress due to water transfers, persistent droughts, and polluted runoff. In addition, falling water levels worsen

Many lakes around the world exhibit acute environmental stress due to water transfers, persistent droughts, and polluted runoff. In addition, falling water levels worsen air quality by exposing desiccated shores. To our knowledge, however, no published hedonic study has analyzed the costs of deteriorating water quality jointly with the air quality impacts of falling water levels for a large inland water body. We conduct such an analysis for the Salton Sea, the largest lake in California. Our spatial autoregressive models estimated on single-family properties located within 10 miles (16.1 km) of the Sea show that a 1 km reduction in distance to the Sea results in a $595 decrease in the price of a single-family residence. In addition, a 1% increase in annual particulate matter concentration reduces the value of the average family residence by $1,140. These results highlight the vulnerability of poor rural communities to deteriorating environmental conditions.

A spatially comprehensive, hydrometeorological data set for Mexico, the U.S., and Southern Canada 1950–2013

Nature Portfolio (Springer Nature) | August 18th, 2015

A data set of observed daily precipitation, maximum and minimum temperature, gridded to a 1/16° (~6 km) resolution, is described that spans the entire country

A data set of observed daily precipitation, maximum and minimum temperature, gridded to a 1/16° (~6 km) resolution, is described that spans the entire country of Mexico, the conterminous U.S. (CONUS), and regions of Canada south of 53° N for the period 1950–2013. The dataset improves previous products in spatial extent, orographic precipitation adjustment over Mexico and parts of Canada, and reduction of transboundary discontinuities. The impacts of adjusting gridded precipitation for orographic effects are quantified by scaling precipitation to an elevation-aware 1981–2010 precipitation climatology in Mexico and Canada. Differences are evaluated in terms of total precipitation as well as by hydrologic quantities simulated with a land surface model. Overall, orographic correction impacts total precipitation by up to 50% in mountainous regions outside CONUS. Hydrologic fluxes show sensitivities of similar magnitude, with discharge more sensitive than evapotranspiration and soil moisture. Because of the consistent gridding methodology, the current product reduces transboundary discontinuities as compared with a commonly used reanalysis product, making it suitable for estimating large-scale hydrometeorologic phenomena.