Determines the importance of observation choice and parameter values on aquifer management objectives when controlling for model error, and compares how the relative performance of management alternatives varies when exposed to endogenous uncertainties. [Github]
Investigates how model projection errors influence assessments of water availability and financial stability for irrigation districts in California.
Framework for dynamic adaptation of water resources systems under climate and land use uncertainty by designing and testing policies that combine relevant indicators, actions, and thresholds in a flexible structure. [Github].
Investigates the relationship between scenario properties (annual runoff, snowpack, and baseline regret) and the ability of policies to generalize out of sample. [Github].
CALFEWS simulation model connecting statewide inter-basin transfers with conjunctive water management strategies at the scale of irrigation and water storage districts. [Github].
Dynamical systems model of asymmetric resource access in resource-based communities linking industrial resource degradation, community well-being, and migration in response to economic and resource conditions. [Github].
Automated generation of model structure to describe human behavior in water systems using symbolic regression. Evaluation along performance-complexity tradeoff, and sensitivity analysis of resulting model structures. [Github].
Multivariate calibration of SWAT models for the Tulare Basin, California to evaluate changes in runoff timing due to current and future snowpack decline.
Adaptations by parameterizing the structure of existing policies using a dynamic flood control rule curve and revised snowpack-to-streamflow forecasting methods. [Github].
Extensions of bagging and cross-validation techniques for policy search problems using streamflow resamples from the paleo record.
Dynamic classification of the occurrence of water supply vulnerability to climate change, represented by the long-term moving average of reliability falling below a threshold. [Github].
Comparing managed recharge alternatives to reduce the effect of historical overdraft using a physically-based groundwater model.
Review article framing dynamic adaptation to climate change as an optimal control problem. Discussion of challenges that arise due to multiple sources of uncertainty.
Modifying process representation in the coupled WRF/Noah-MP model to improve hydrologic simulation upstream of California reservoirs.
Optimization-based design of satellite constellations for near-continuous global coverage.
Sensitivity analysis of a global groundwater model using Method of Morris.
Analysis of perennial crop data and implications for water supply derived from pesticide permits over the period 1974-2016. [Github].
High-resolution soil dataset with reduced uncertainty compared to standard prior distributions used in land surface model calibration.
A statistical testing approach for threshold-based classifiers of vulnerable scenarios in long-term streamflow projections. [Github].
Clustering California reservoirs based on typical behavioral profiles.
Extension of policy tree optimization method using synthetic ensembles of streamflow forecasts for reservoir control, including groundwater banking.
Short communication describing open-source version of the CALVIN model with LP solver performance benchmarks. [Github].
Adaptive levee planning with nonstationary flood risk, using stochastic dynamic programming.
Impact of preference constraints (i.e., optional constraints on objective values) on the performance of multi-objective evolutionary algorithms.
Policy search for multi-reservoir systems using a hierarchical parallelization scheme, using both deterministic and stochastic hydrology.
Binary tree policy search for control of dynamic systems. Uses a heuristic simulation-optimization approach. [Github].
An approach for reservoir operation to incorporate downstream biological objectives without sacrificing water supply reliability. The method is applied for fall-run Chinook salmon in the Lower American River.
An editorial based on the results of practitioner interviews and analyzing course syllabi from the systems analysis field.
Optimization of decision alternatives across an ensemble of deeply uncertain scenarios, rather than optimizing to historical observations and then simulating the ensemble.
Implementations of the Sobol, Morris, FAST, and Delta methods, among others. [Github]
Estimating the mean and variance of global parameter values using extra-trees regression, after first reducing the dimension of the parameter space using sensitivity analysis.
Multi-objective optimization for urban water utilities combining long-term infrastructure development with short-term conservation options, where decisions are triggered by thresholds of a risk metric.
Modifies a synthetic streamflow generation technique to allow user-defined frequency and severity of droughts while preserving autocorrelation and spatial correlation. [Github].
The OpenMORDM library for decision making under deep uncertainty. [Github]
Large-scale ensemble evaluation of VIC model parameters on a global grid. Constraining ensemble members with observed data and calculating parameter sensitivity using the CDF distance metric.
Analyzing and optimizing coverage of satellite constellations for rainfall observation.
A review of frameworks for decision making under deep uncertainty, where the probabilities of future scenarios are difficult to quantify.
Given a set of multi-objective solutions optimized under historical conditions, this study subjects them to a range of hydrologic and economic uncertainties to identify the most robust solutions.
Multi-objective portfolios of water conservation and transfer options for urban water utilities to maintain supply reliability and reduce financial impacts of drought.
Identifying an existing reservoir operating policy from historical data using a network of radial basis functions, and re-optimizing the policies for multiple objectives across an ensemble of synthetic streamflows.
Parameter sensitivity computed at a three-hour timestep and compared to aggregate event-scale sensitivity to improve process understanding of a spatially distributed hydrologic model.
For a spatially distributed watershed model, we show that the Morris method can produce parameter sensitivity values and rankings similar to the Sobol method, with significantly reduced computation time.
Sobol sensitivity analysis of three hydrologic models (Hymod, HBV, SAC-SMA) at monthly and annual timesteps, performed across a hydroclimatic gradient of 12 watersheds in the US.
A comprehensive diagnostic comparison of state-of-the-art optimization methods for multi-objective problems, using three real-world case studies: hydrologic model calibration, water supply planning, and groundwater monitoring. Uses MOEAFramework and Borg MOEA.