Encoded Canadian Coast Guard (CCG) Automatic Identification System (AIS) data (2019, 2020 and 2021) was decoded by Andrea Nesdoly using scripts developed in Python. This allows access to vessel information such as vessel positions, timestamp, vessels unique identifier or MMSIs (Maritime Mobile Service Identity), vessel speeds and course.

Afterwards, the AIS data was cleaned by removing entries that met the following criteria:

• duplicate AIS messages received within a 10 second timespan;

• stationary docked vessel in a harbour;

• invalid vessel positions with speed over ground greater than 70 knots; and

• invalid vessel positions with a sudden change of direction.

A list of commercial whale watching vessels based in British Columbia and Washington State and their corresponding MMSIs (Maritime Mobile Service Identity) was compiled from the whale watching companies and Marine Traffic (www.marinetraffic.com). This list was used to query the cleaned CCG AIS data to extract AIS positions corresponding to commercial whale watching vessels.

AIS data was then divided into three subregions: south coast Vancouver Island including Puget Sound, west coast of Vancouver Island, and north-east Vancouver Island. Datasets containing commercial whale watching trips and wildlife-viewing events were generated for each sub-region. A commercial whale watching trip was defined as a set of consecutive AIS points belonging to the same vessel departing and ending in one of the previously identified whale watching home ports. Each identified trip received a unique identifier.

A classification model (unsupervised Hidden Markov Model) using vessel speed as the main variable was developed to classify AIS vessel positions into wildlife-viewing and non wild-life viewing events. Each commercial whale watching trip was analyzed using the classification model to identify one or more wildlife-viewing events. Each wildlife-viewing event received a unique identifier.

Commercial whale watching trips in the south and north-east of Vancouver Island were limited to a duration of minimum 1 hour and maximum 3.5 hours. For trips in the west coast of Vancouver island the maximum duration was set to 6 hours. Wildlife-viewing events duration was set to minimum of 10 minutes to a maximum of 1 hour duration. These duration thresholds were applied after exploratory analysis were carried out in R and adjusted based on feedback from commercial whale watching operators. The same thresholds were applied for each year.

Cleaned AIS positions with whale watching trip information for each sub-region and year were imported to ArcGIS Pro to complete the following data processing:

1. Imported AIS vessel positions were saved as a point feature dataset.

2. Tracks were then generated using the tool ‘Point to Tracks Segments’.

3. Tracks from each sub-region and same year were combined using the tool ‘Merge’.

4. Tracks belonging to the summer period (June 1 to September 30) were saved as a separate layer.

5. Tracks were dissolved using the trip unique identifier to obtain one field entry for each whale watching trip.

6. Whale watching trip datasets for each summer period were intersected with a grid of 500 by 500 meters resolution.

7. Summary statistic tables were extracted for each year using the cell unique field and calculating the number of trips and distance travelled within the cell.

8. Output tables (one for each year) were then joined back to the grid dataset using the cell ID field, and then saved as a new dataset.

9. Undesired fields were deleted and remaining fields renamed.

AIS positions representing wildlife-viewing events for each sub-region and year were imported to ArcGIS Pro to conduct the following data processing;

1. Imported AIS vessel positions were saved as point feature dataset.

2. Positions found within the boundaries of harbours and marinas were eliminated to remove the likelihood of false positives.

3. Positions found in narrow passages or very shallow waters (i.e., Fraser Delta) were also removed to reduce the number of false positives.

4. From the remaining points, tracks were generated using the tool ‘Point to Tracks Segments’.

5. Tracks longer than 2000 meters were deleted as well as tracks with an estimated speed greater than 14.5 knots.

6. Wildlife-viewing tracks belonging to the summer period (June 1 to September 30) were saved as a separate layer.

7. Summer wildlife-viewing tracks were intersected with the same grid used in the whale watching trip analysis of 500 by 500 meters.

8. The duration of the wildlife-viewing events per grid cell was calculated using the estimated speed per cell (this is carried over from the ‘Points to Tracks Segments’ tool) and the distance travelled per cell (this is automatically calculated after the intersection).

9. Wildlife-viewing events from each sub-region and same year were combined using the ‘Merge ‘tool.

10. Summary statistic tables were calculated for each year using the cell unique field and calculating the number of wildlife-viewing events and event duration per cell.

11. Output tables (one for each year) were then joined back to the grid dataset using the cell ID field, and then saved as a new dataset.

12. Undesired fields were deleted and kept field renamed.