This repo handles the results post-processing for VAST SDM projections. This is also the home of the Shiny app development and testing code.
When species distribution models have been run, their output projections are put into a designated lab folder for the project on Box
These raw model outputs can be found here:
Mills Lab/Projects/COCA19_Projections/.
Each model follows this naming convention:
{Species}_full_CMIP6_{SSP Scenario}_mean.rds
ex.
AtlanticMackerel_full_CMIP6_SSP1_26_mean.rds
This script is responsible for gathering all the different species and scenario outputs and processing them in the following ways:
-
Density projections are pulled from the larger outputs
-
Seasonal density estimates are smoothed over time using a 5-year rolling average, performed at each Lat/Lon node
-
A dataframe of the rolling-average density projections are exported locally for further pre-processing for the app:
VAST_all_densities_all_species.csv -
Node locations are given unique ID's, and these locations are then saved as a standalone file:
Data/spatial/unique_location_coords.csv. These can be used to more efficiently flag nodes that fall within a given shapefile etc. -
Baseline period average densities are calculated, and exported locally at:
Data/projections/VAST_baseline_2010to2019_densities_all_species.csv
Everything in this script is then passed for further processing.
This script begins at the processing starting point or the 5-year
rolling average density estimates for all the species:
VAST_all_densities_all_species.csv.
It then processes the average biomass density for each species & scenario at specific points in the future. These periods are determined outside this repo, but correspond with when both the SSP1 and SSP5 scenarios cross important temperature milestones above our baseline period climate of 2010-2019 for our study area.
We end with two files:\
- The average annual densities at these horizons:
Data/projections/Cmilestones_all_species_test.csv\ - The average seasonal densities at these horizons:
Data/projections/Cmilestones_all_seasons_test.csv
These are used to power the maps projecting spatial distributions at these points in the future.
This script begins at the processing starting point or the 5-year
rolling average density estimates for all the species:
VAST_all_densities_all_species.csv.
The purpose at this step is to produce timeseries specific to a number of areas of interest:
-
Ecological Production Units
-
Gulf of Maine
-
Georges Bank
-
Mid-Atlantic Bight
-
-
Local Regions based on Statistical Areas:
-
Downeast Maine
-
Midcoast Maine
-
Southern Maine
-
Northern MA
-
Cape Cod
-
Southern MA / RI
-
NY / NJ * NC / VA / DE
-
-
Port Community Footprints - from Vessel Trip Reports
This script produces the following files:\
Regional Timeseries:
Data/projections/annual_regional_species_projections.csv
VTR Footprint Timeseries:
Data/projections/annual_community_footprint_projections.csv
VTR Footprint Baselines:
Data/projections/baseline_community_footprint_projections.csv
For a small number of communities, the fishing gootprint represented by the VTR data is quite small. For these communities we decided that a larger region might better reflect the projected climate and fisheries outcomes. This script details the rule based approach used to create new footprints for small communities that uses the statistical areas for fisheries management.
The following maps display how the rule based areas are constructed. VTR footprints in orange are overlaid against statistical areas outlined in black. If there is sufficient overlap the statistical areas will be used. Areas in green are statistical zones that satisfy this minimum overlap rule.
Timeseries for these communities are then generated here.
This script produces the following files:
Data/projections/annual_small_footprint_szone_overlap_projections.csv
Data/projections/baseline_small_footprint_szone_overlap_projections.csv
The shiny app and supporting code can be found in the
Coca_SDM_app_dev/ directory.
Data needed for the app to run is processed in the
R/app_ready_data_prep/ directory.