COVID-19 Bird Strike Update #3
It’s time for another update on the impact of COVID-19 traffic downturns on wildlife strikes. My first (and updated) post was looking at the initial months of the pandemic and the second looked at the northern summer period but it has been a big three months since I’ve looked at these numbers.
Generally speaking, we’ve had the holiday season but with lockdowns, border closures, more record-breaking days, things have been anything other than festive. Due to the lag in US traffic data, the analysis below only goes up November 2020.
Doesn’t that feel like a lifetime ago?
New Data!
In this update post, I have a few new pieces of information including some analysis from the FAA which isn’t exactly complementary to my earlier posts.
But we’ll come back to that…
Same Old Analysis First
In order to establish the new data within the context of my previous posts, I’ll quickly update my original graphs using my very rough-and-ready technique.
As a reminder, I use the raw number of strikes reported each month as found in the FAA Wildlife Strike Database and for traffic data, I continue to download data from the US Bureau of Transport Statistics.
Let’s have a look at what we’ve got before I go over the limitations again.
Wildlife Strikes
I’ve retained the “potential late reports” line as I continue to see movement in the report numbers from the middle of 2020. Despite the traffic drop, the raw strike numbers seem to be converging with the 5-year average. But as the northern winter approaches, strike numbers tend to drop anyway. So, any proportional difference will see an absolute reduction too.
Wildlife Strike Rates
The lower traffic numbers of the post-summer and pre-Thanksgiving travel periods, have the potential to power-up the impact on any late reports but time will tell on that front too. Again, the trend is looking like a return to (dare I say it?) normal.
What’s Wrong with this Picture?
I’ve been honest since the first update that there is a potential disconnect between the wildlife strike reports and the traffic statistics. The wildlife strike database includes reports from every type of airport in the US involving every type of operator. There are reports from large international airports of strikes with transport category aircraft and reports from small airstrips involving private planes.
On the other hand, the traffic statistics are drawn from a very specific sub-set of airline and cargo operators. On this list are only those domestic operators with more than $20M in annual revenue and foreign operators with more than 10K passenger per month to and from the US. Lucky for me, they provide this list in a, relatively, simple format for me to copy into a spreadsheet.
Contradictory Analysis
While I “promised” to look into these limitations three months ago, I have to admit that I was particularly motivated to do it better by some contradictory analysis coming out of the FAA. I haven’t yet claimed any statistical significance in the trends I’ve identified above but I was a little taken aback when the FAA analysis showed no identifiable deviation of previous numbers.
You can download the whole presentation from the World Birdstrike Association website but here is the particular slide (reproduced without any permission but in the interests of education):
Now, you may not notice it, but that name at the bottom of the chart, that’s a pretty intimidating name. Dr Dolbeer prepares the book, every year, on wildlife strike trends in the US and a bunch more very useful information in lots of other places.
In the interests of science, I thought I would finally do that more robust analysis and try to replicate the FAA analysis.
Time to Speak Parseltongue
In addition to the wildlife strike data from the FAA and the traffic data from the US BTS, I downloaded the current list of certificated US airports, also from the FAA, and the list of operators included in the US BTS data. Using a combination of the wildlife strike data and the operator list, I created a map of the operator names used in the former file that corresponded to those used to compile the traffic statistics.
I used these two new files to filter out wildlife strike reports that were not related to airports with a Wildlife Hazard Management Plan (WHMP) and not included in the traffic statistics. This means that my wildlife strike rates should be much more accurate for that sector of the aviation industry.
This time, I ran everything through Python and I am happy to share the code with anyone interested.
So, do my results match the FAA?
Ah…
Wildlife Strikes
I think these numbers are pretty similar. I guess the difference will be in the list of air carrier and air taxi aircraft that Dr Dolbeer uses. When the FAA analysis is fully published, I will be very interested in seeing this data.
But wildlife strike rate is the more important statistic to look at.
Wildlife Strike Rate
The gap between the five year average and the 2020 numbers has definitely closed. I would be more reluctant to claim the difference is statistically significant now.
The even closer relationship shown in the FAA presentation is partially due to the traffic data they use and to the historical data being limited 2019 only. With wildlife strike rates trending up over the last five years, 2019 is, itself, above the average mark and I have included it here in grey as well.
More Work to be Done
My next challenge is to test the impact of COVID-19 on wildlife strike rates using a robust statistical method. I think I now have more appropriate raw data but it is taking some time identifying the right testing protocol. I’m getting closer but I’m just not quite there.
But in the meantime, stay safe and healthy.
