Description Usage Arguments Value Author(s)
This tool allows for the visualization and analysis of the harvest output from an Atlantis model. This tool includes analysis of catch, bycatch, and discarding either at the level of the entire population or by age class. In addition to graphical output, it is also possible to perform a skill assessment of the model performance (currently only for the time series of catch and bycatch). The quantitative metrics used to analyze the performance of the model is based on the approach described by Olsen et al. (2016) and Stow et al. (2009) using: the correlation coefficient, root mean squared error, reliability index, average error, average absolute error and the modeling efficiency routines.
1 |
grp.csv |
Character string with the path to the Groups *.csv file (Atlantis input file). |
fish.csv |
Character string with the path to the fisheries *.csv file (Atlantis input file) |
catch.nc |
Character string with the path to the catch netcdf output file from Atlantis. Usually this file has the name of [Your_Model]CATCH.nc, where [Your_Model] is the name of your Atlantis model |
ext.catch.f |
(Default = NULL) Character string with the path to the external file with the observed catches and discards by year. This helps to calibrate the harvest section of Atlantis and it is required to perform the skill assessment of the model. |
This function provides 3 different sets of analyzes of the catches in the following tabs:
Biomass: This function helps to analyze change through time of all the variables (i.e. catch and discard) for the selected Fishery. This information can be displayed by each recording time step (toutfinc parameter) or by year.
Numbers: This function allows the user to analyze the change by age and through time of the variable catch and discards for the selected functional group
Compare: This function allows for several skill assessment of the model (based on the analysis of the simulated and observed time series of catch and bycatch) to be performed. The analysis is based on the approach described by Olsen et al. (2016) and Stow et al. 2009, which is composed of the following quantitative metrics:
Correlation coefficient (r): measures the tendency of the predicted P and observed O values to vary together. The values of correlation can range from -1 to 1, with negative values of correlation for time series that vary inversely.
r = \frac{\displaystyle∑_{i=1}^{n} (O_{i} - \bar{O})(P_{i} - \bar{P})}{√{\displaystyle∑_{i=1}^{n} (O_{i} - \bar{O})^2 \displaystyle∑_{i=1}^{n} (P_{i} - \bar{P})^2}}
Average Error (AE; or model bias): The average error is a measure of aggregated model bias.
AE = \frac{\displaystyle∑_{i=1}^{n} (P_{i} - O_{i})}{n} = \bar{P} - \bar{O}
average absolute error (AAE) and root mean squared error (RMSE) : Both equations calculate the bias of the model, considering the magnitude rather than the direction of each discrepancy. That is because values near zero in the AE can be misleading, as negative and positive discrepancies can cancel each other.
AAE = \frac{\displaystyle∑_{i=1}^{n} |P_{i} - O_{i}|}{n}
RMSE = √{\frac{\displaystyle∑_{i=1}^{n} (P_{i} - O_{i})}{n}}
Reliability index RI: This quantifies the average factor by which the predicted values differ from observations. If the model predictions do not differ too much from the observed then the value of RI should be close to 1. But if the value of RI is 2 it means that a model predicts the observations within a multiplicative factor of two, on average.
RI = exp√{\frac{1}{n} \displaystyle∑_{i=1}^{n} (log\frac{O_{i}}{P_{i}})^2}
Modeling efficiency MEF: The modeling efficiency measures how well a model predicts relative to the average of the observations. A value on MEF close to 1 indicates that the model matches the observations. A value of MEF less than zero means that the average of the observations is a better predictor than the model.
MEF = \frac{\displaystyle∑_{i=1}^{n} (O_{i} - \bar{O})^2 - \displaystyle∑_{i=1}^{n} (P_{i} - O_{i})^2}{ \displaystyle∑_{i=1}^{n} (O_{i} - \bar{O})^2 }
where n is the number of observations; O_{i} the ith of n observations; P_{i} ith of n predictions, and \bar{O} and \bar{P} are the observation and prediction averaged, respectively.
Demiurgo
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.