Using CRiSP Harvest

CRiSP Harvest version 3.0.6 is intended to be a demonstration of a user-friendly salmon harvest management simulation model. CRISP HARVEST IS NOT THE SAME MODEL CURRENTLY BEING USED BY THE PACIFIC SALMON COMMISSION CHINOOK TECHNICAL COMMITTEE TO ANALYZE CHINOOK SALMON HARVEST STRATEGIES. Use CRiSP Harvest to learn about general model behavior and functionality, but DO NOT USE CRISP HARVEST TO EVALUATE CURRENT MANAGEMENT PROPOSALS.

Functionality

CRiSP Harvest version 3.0.6 duplicates the forecasting portion of the fall 1995 version of the PSC Chinook Model. The Monte Carlo and In-River Management options in CRiSP Harvest version 3.0.6 are additional features added by the University of Washington School of Fisheries staff and are not contained in the PSC Chinook Model. These new features have not been formally peer-reviewed by the PSC Chinook Technical Committee. Monte Carlo and In-River Management model behaviors should be considered preliminary.

Calibration

CRiSP Harvest version 3.0.6 uses input data files from the 9525 calibration of the PSC Chinook Model.

Default Management Strategy

The default management strategy for the simulation period (1995 - 2017) defined by the input files accompanying CRiSP Harvest represents a modified version of one strategy being analyzed by the PSC Chinook Technical Committee during 1995. In simple terms, this default strategy simulates reduced harvests for the years 1995 - 1997 (by reducing harvest rates to about 75% of the base period rates) and simulates average 1991 - 1994 catch ceilings and harvest rates for years beyond 1997.

Default Brood Year Survival Rates

The model results during the simulation period are very sensitive to the estimates of the stock specific brood year survival rates (roughly the survival rate from egg through age one; also called EV, or Environmental Variability, scalars). The simulation period EV Scalar values used in CRiSP Harvest are estimated from the calibration period values under the assumption that the EV Scalars are independent and log-normally distributed. When using Monte Carlo mode, the model randomly selects EV scalar values from the estimated log-normal distribution associated with each stock. The Monte Carlo values are selected independently for each stock, implying that there is no correlation between brood year survival rates among stocks.

Documentation

The online documentation in this program has not yet been fully implemented. Many of the dialog screens have no context sensitive help.

We have included the CRiSP Harvest Manual in the Program Files/CBR/CRiSP Harvest directory.

How to Get a Copy

The latest copy of CRiSP Harvest v3.xx can be obtained at the following address: www.cbr.washington.edu/analysis/archive/harvest/crispharvest