WRLCM

Criterion

Explanation

General Description

A state-space life-cycle model for winter-run Chinook Salmon in the Sacramento River that has multiple stages, including eggs, fry, smolts, juveniles in the ocean, and mature adults in the spawning grounds. The model is spatially explicit and includes density-dependent movement among habitats during the fry rearing stage. Parameter estimates for the model were obtained from external analyses, expert opinion, and estimation by statistical fitting to observed data.

Model Domain

The WRLCM is structured spatially to include several habitats for each of the life history stages of spawning, rearing, smoltification, outmigration, and ocean residency. The model uses discrete geographic regions of Upper Sacramento River, Lower Sacramento River, Yolo Bypass Floodplain, Delta, Bay, and Ocean.

Developer

NMFS-SWFSC, NMFS-NWFSC, QEDA Consulting, UC-Santa Cruz

Hardware computing requirements

Not specified

Code language

R

Original application

Broad goal is to quantitatively evaluate how Federal Central Valley Project (CVP) and California State Water Project (SWP) management actions affect Central Valley Chinook salmon populations. Examples of management actions include changes in water project operations, addition or removal of barriers, and a variety of habitat restoration initiatives.

Public/proprietary and cost

N/A

Physically or empirically based

Mechanistic and empirical

Mathematical methods used

Population dynamics are governed by Beverton-Holt transition equations. Several parameters are functions of temperature, flow, and gate position; fitted to empirical data with linear and logistic regression. Calibration was performed with Expectation-Maximization algorithm (gradient search). Monte Carlo simulations are used to evaluate management actions.

Input data requirements

Requires running of numerous models to generate input data: CALSIM II, HEC-RAS, SRWQM, DSM2 HYDRO, Habitat Capacity Models, and ePTM. Also requires estimates of impact rates on vulnerable age class of Chinook Salmon from the Pacific Fisheries Management Council annual forecast.

Outputs

The model operates at a monthly time step in the freshwater stages and at an annual time step in the ocean stages. The model can output fish abundance by life stage and region for each for each time step.

Pre-processing and post-processing tools

No processing tools provided

Representation of uncertainty

For eight parameters estimated in calibration, Monte Carlo parameter values were drawn from multivariate normal distribution centered on the maximum likelihood estimates (MLE) and using the covariance matrix estimated from the Hessian obtained at the MLE. The draws from the multivariate normal distribution incorporated the relative uncertainty in the estimated parameters and preserved the strong correlation among several of the life cycle model parameters that were identified in the correlation matrix of the parameter estimates. For the random effects, iid normal N(0,1) random variables were drawn to reflect the annual random effects in the process noise. All other parameters were set to their fixed values.

Prevalence

High profile modeling effort involving collaboration of government agencies, universities, and consultants. No peer-reviewed publications. Used in WaterFix.

Ease of use for public entities

Running the full model is not currently an option for use by public entities. There was previously a public web application that allowed users to explore model output over a subset of the parameter space, but that web application is no longer publicly available. Because the WRLCM is integrated with several other models, running the model requires broad expertise in hydrologic, hydraulic, and fisheries modeling.

Ease of obtaining information and availability of technical support

Source code availability

N/A

Status of model development

Model is under active development. Future work will refine existing model and extend model to include other runs of Chinook Salmon in Central Valley. NMFS and USBR host workshops to address questions about the model structure and capabilities.

Challenges for integration

Model is already highly integrated. The analytical framework consists of linking and applying hydrological, hydraulic, water quality, and salmon population models. The primary challenge for integration with other models is that the WRLCM is under active development and not generally accessible to modelers outside of the core development team.