Estimation of lost fishing gear based on observed vessel activity
Note: the main branch does not include results!
Please checkout populated to find output files in the output folder-- or run the code according to the instructions below to regenerate them from inputs.
The goal of this project is to estimate the flow of fishing gear into the sea that results from normal industrial fishing activity. We attempt to model the relationship between effort and catch by linking reports of landed fishery catch with observations of fishing effort using telemetry data from Global Fishing Watch. We then use the resulting relationship to estimate the dissipation of fishing gear using stochastic simulation.
The code for organizing catch and effort data was written in R. There are three main steps, comprised of the three files in scripts/analysis:
cleans_fao_datarepository.RmdCompute allocation factors to each logical fishery based on the Perez Roda database.match_catch_effort.RmdAssign catch to fisheries and match with GFW effort data. Used to generate most of the output data used in the simulation and the paper.quantile_reg.RmdCompute quantile regression results for effort intensity, used to estimate total fishing effort.
Thereafter, the effort intensity models are linked to gear intensity and dissipation models to predict gear losses. This step happens in python using the unit gears framework (JIE article). The code for this step is found in the fishery_model python package.
This section is executed in R. We recommend Rstudio. This code requires the data present in the data directory and populates CSV files in output. Note: for access to the pre-computed outputs, checkout the populated branch.
- Open the R project file in the present directory.
- Source "R/set_up.R" Required packages should be installed automatically.
- In sequence, run the scripts in the
scripts/analysisdirectory:- cleans_fao_datarepository.Rmd
- match_catch_effort.Rmd
- quantile_reg.Rmd
- If desired, review scripts in
scripts/figures_and_tablesfor visualization options.
This section is executed in python.
- Create an appropriate environment for the project (python 3.6 or higher).
- In a shell, run
pip install -r requirements.txt - Execute
python run_simulation.pyto execute a full simulation. Results are stored inoutput/simulationdirectory.
- Use the switch
-nto change the number of Monte Carlo iterations (default 1000) - Use the switch
-fto generate figures equivalent to the publication figures.
For interactive or custom use, please review the two notebooks in the jupyter directory, as well as the documentation for unit_gears.
The study object can also be used for many other queries. For instance, a table of gear use and dissipation per
logical fishery can be produced by indexing across the result sets differently, as shown in
fishery_model.per_fishery_table():
>>> import fishery_model
>>> st = fishery_model.run_study(n=55, year='2018')
>>> fishery_model.per_fishery_table(st, 'trawlers', '0.9')
Country FAO Year Gear tau unit_median unit_05 unit_95 oper_median oper_05 oper_95 diss_median diss_05 diss_95
0 AGO 47 2018 trawlers 0.9 5.64 1.75 20.5 5.2e+05 1.61e+05 1.89e+06 1.66e+04 4.02e+03 7.76e+04
1 ARG 41 2018 trawlers 0.9 4.07 1.51 17.9 4.96e+06 1.84e+06 2.18e+07 1.75e+05 6.19e+04 6.7e+05
2 AUS 71 2018 trawlers 0.9 8.58 2.56 28 7.88e+05 2.35e+05 2.57e+06 2.39e+04 8.9e+03 8.89e+04
3 AUS 81 2018 trawlers 0.9 7.86 2.17 19.6 1.07e+05 2.95e+04 2.67e+05 3.53e+03 1.19e+03 8e+03
4 BEL 27 2018 trawlers 0.9 6.17 2.42 42.7 2.55e+05 1e+05 1.76e+06 9.19e+03 3.37e+03 5.05e+04
.. ... .. ... ... ... ... ... ... ... ... ... ... ... ...
108 USA 77 2018 trawlers 0.9 7.15 1.94 34.6 2.9e+04 7.85e+03 1.4e+05 1.02e+03 269 5.6e+03
109 VNM 71 2018 trawlers 0.9 7.13 2.35 27.1 6.94e+06 2.29e+06 2.64e+07 2.35e+05 7.74e+04 1.13e+06
110 ZAF 47 2018 trawlers 0.9 5.57 1.32 21.6 8.74e+05 2.07e+05 3.39e+06 2.61e+04 7.57e+03 1.13e+05
111 ZAF 51 2018 trawlers 0.9 7.33 2.11 17.7 3.29e+03 948 7.95e+03 87.8 30.1 278
112 NA 27 2018 trawlers 0.9 9.27 1.91 28.5 2.63e+03 543 8.09e+03 89.8 18.7 295
>>>
-- data
|__fao_landings
|__gfw
|__perez-roda_discards
|__watson
-- doc
-- mapping_tables
-- output
|__figures
|__simulation
-- R
-- scripts
|__analysis
|__figures_and_table
|__mapping_tables
-- .gitignore
-- .Rhistory
-- README.md
- data: contains all the raw data used in this part of the project organized by data source. The four main data source are:
- fao_landings: FAO publicly available global production
- gfw: Global Fishing Watch vessel activity data. This data is not publicly available.
- perez-roda_discards: publicly available data used in the FAO Third Assessment of Global discards
- watson: Reginald Watson (2019) Global Fisheries Landings V4.0, Australian Ocean Data Network
For more detail on each data source please see Wiki
-
doc: folder containing some documents related to the project.
-
mapping_tables: all files used to re-conciliate countries, species, gear and FAO area names between data sources. This files were created for this project. Find more information about the content of these tables here
- intermediate_files: intermediate files used to create the countries mapping table. Files in this folder were altered manually.
-
output: contains all files created through the scripts of this project. All files can be reproduced by running one of the scripts and are part of the core results of the project, or by checking out the
populatedbranch. -
R: R scripts that contain all functions needed for the project. Scripts are organized by data source.
-
scripts: core data processing of the project.
- analysis: Three main steps to reconcile catch and effort data: classify species into a logical fishery, group landings by fisheries gear type and match to effort data, analyse relationship by running multiple quantile regressions.
- figures_and_tables: scripts that create figures used in this project.
- mapping_table: scripts used to create all the mapping tables used in this project. Each script documents the process and steps to create the final outcome.