Alberta bird models using ABMI+BAM+BBS data

Verion 2020

The code in this repo originates from the https://github.com/psolymos/abmianalytics project.

What you can find here

This repository contains code to reproduce the 2020 version of Alberta bird models presented on the ABMI data and mapping portals.

The numbered R scripts are the distinct steps in the process (functions that all steps might need, data processing, data packaging, model summaries, prediction/mapping).

The cc folder contains code to run the models on WestGrid computing cluster.

The sector-effects folder contains code for calculating sector effects with partial backfilling.

The workflow

1. Combine all the data inputs into a common format (01-data-processing.R):

• species data: BAM and BBS species code can be found in the borealbirds organization, ABMI data processing is documented in the ABbiodiversity/abmidata repo
• Veg/soil/HF info: ABbiodiversity/veg-hf-soil-summaries repo
• QPAD offsets: borealbirds/qpad-offsets repo
• Lookup tables, etc.: ABbiodiversity/abmianalytics, inside the lookup folder

2. Create a data package that can be copied over to other machines to run the models (north, south separately) (02-data-packaging.R):

• Various ID, methods related variables (ARU, ROAD, date/time etc), spatial predictors, veg/soil/HF info at various scales
• Species matrix
• Offsets matrix
• Models list
• Matrix with bootstrap indices

3. Run models and save the output for each species (scripts in the cc folder: these will run locally or on a cumpute cluster)
4. Check results (03-model-summaries.R)
5. Prediction maps, sector effects, figures, etc. are described in the pipeline/2020 folder of the ABbiodiversity/abmianalytics repo.
6. Final coefficients are wrapped in the ABbiodiversity/allinone and ABbiodiversity/allinone-coefsrepos for easy distribution and reproducibility.