Stock assessment and data management

From Gcube Wiki
Jump to: navigation, search


In this page a number of models and tools are reported, which have been developed for the gCube platform and that are currently running in the D4Science e-Infrastructure. These methods have been developed in the course of several European projects. They are available through the i-Marine project web portal or through the D4Science web portal, offered by the StatMan and the TabMan facilities.


Catch, effort and abundance data

Size/length composition data

Biological data

Other data


Catch only methods


Biomass dynamics models

Delay-difference models

Age-structured production models

Virtual population Analysis

Statistical catch-at-age methods

Integrated analyses

Stock Assessment

An algorithm to estimate the Maximum Sustainable Yield (MSY) from catch statistics and qualitative estimates of the resilience and the carrying capacity of the stock. The algorithm uses Bayesian methods as well as Monte Carlo estimates. If also a Biomass trend or a CPUE trend is provided, MSY is estimated with higher precision. The method was described, tested and verified in the ICES WKLife IV meeting. Its performance is documented in the ICES WKLife IV report, which indicates CMSY as the currently best performing model for Stock Assessment, especially with limited-data scenarios. The algorithm was developed in the R language by R. Froese, G. Coro and H. Winker in 2014.

Example of input files for the process are available here and here.

  • Name of the algorithm on StatMan: Cmsy

Length-Weight relations

An algorithm to estimate Length-Weight relations for marine species using Bayesian methods. The algorithm is based on an R procedure by Froese et al. 2013 relying on the Cube-law theory. The algorithm is executed in parallel processing fashion, using a number of machines that drastically reduce its computational time. For further details about this parallelization see Coro et al. 2014.

A summary output for more than 11,000 species is available here. One example of input provided to the algorithm, which contains the list of species and families is here.

  • Name of the algorithm on StatMan: Lwr

Vessels Transmitted Information

Vessel Transmitted Information (VTI) is a Virtual Research Environment that allows importing, "curate" and aggregate data that contain latitude and longitude information. The VRE offers procedures to enhance the information contained in a given dataset with bathymetry and other data. For each point along the trajectory, processing algorithms can estimate fishing activity based on the speed of the vessel and the bathymetry of the location.

Each transmitted record can also be enriched with information about the FAO Area it belongs to, according to five types of subdivisions. The VRE also contains procedures to calculate the monthly fishing effort per area, according to the time interval involved into the dataset. Two GIS Maps can be produced, visualized and overlaid, which contain (i) the vessel trajectories with classification, for each point, of the vessel activity (fishing, steaming, dodging, etc.); (ii) the estimated monthly fishing effort with half degree resolution. Furthermore, the VRE is equipped with generic functionalities for Time Series management in a distributed infrastructure like (i) data aggregation, union, filtering, etc.; (ii) charts production; (iii) data analysis with the R language; (iv) resources sharing with other VRE users.

Details and usages of this technique are available here.

  • Access to the VTI environment here.

Vessels Trajectories Interpolation

An interpolation method relying on the implementation by the authoritative Study Group on VMS (SGVMS). The method uses two interpolation approached to simulate vessels points at a certain temporal resolution. The algorithm processes up to 10000 vessels trajectory points and interpolates the trajectories according to a user's defined temporal resolution. The estimation of trawling tracks use cubic Hermite spline interpolation of position registration data.

The input to the algorithm is a file in TACSAT, format uploaded on the Statistical Manager. The output is a TACSAT file containing interpolated points. The underlying R code has been extracted from the SGVM VMSTools framework. Further details on using this tool are available here.

  • Name of the algorithm on StatMan: Sgvm Interpolation

Seasonality and periodicity detection

An algorithms applying signal processing to a time series of catch statistics. The process uniformly samples the series, then extracts hidden periodicities and signal properties. The sampling period is taken as the shortest time difference between two points. Finally, by using Caterpillar-SSA the algorithm forecasts the Time Series. The output shows the detected periodicity, the forecasted signal and the spectrogram. One experiment and example using this technique to predict fishing activity in the Indian Ocean is available here.

  • Name of the algorithm on StatMan: Time Series Analysis

Other Links