Does the Ivanpah solar facility toast 642 or 28,000 birds a year? Solar #24

 

(Photo by James Ulvog)

Officials from Ivanpah Solar Electric Generating System report there were 321 dead birds found on the site in the first six months of operation. Annualizing that would imply 642 birds will die each year at the facility shown above.

Is that the total of birds killed at the site? Let me ask you a few questions.

• Is the total number of drunk drivers on the road determined by looking at the official FBI statistics on DUI arrests?
• Is the extent of insider trading equal to the number of SEC enforcement actions filed?
• Is the total amount of criminal behavior leading up to the 2008 great recession equal to the number of criminal indictments issued by the Justice Department (which is zero if you didn’t know)?

If you answered yes to all of those three questions then you will certainly agree that the grand total of birds killed at Ivanpah is limited to the reported amount of 321 in six months, or around 600 a year.

On the other hand, if you think the number of drunk drivers is greater than the number arrested or if you believe there is actually more insider trading going on than the SEC prosecutes, then the question is to what extent are the toasted wings undercounted at Ivanpah.

Causes of undercount

There are at least three reasons why the official tally is a severe undercount of bird fatalities.

1. Large birds that are mortally wounded can have enough strength to glide outside the facility’s perimeter or fly for a few more miles before dying. No effort is made to find bird carcasses outside the facility’s perimeter. Federal observers have observed large birds flying through the solar flux, emerging with a trail of smoke, and falling to the ground. Those are called “streamers.” Some streamers glide to outside the perimeter of the facility. Such fatalities would not count since they are outside the perimeter. Federal observers noticed a streamer every minute or two, according to an AP article.
2. Small birds that die might land more than a few hundred feet from the solar collector. The staff only searches for bird carcasses out to a certain distance from the solar collector because most of the birds land very close to the tower. The implied assumption is that zero birds land more than a certain distance from the tower.
3. Predators such as kit foxes, ravens, and coyotes carry off dead or dying birds before the staff can count them. Two things to consider. First, I have read a few of the reports from the facility and noticed that they only survey the site every three or four weeks. That leaves a huge amount of time for predators to dine. Second, cause of death for somewhere around half of the found birds cannot be determined because they are too decomposed. Estimated time since death ranges from a few hours up to several weeks.

In the accounting world, we would call this auditing with your eyes closed. Don’t look at the sample items very closely and you don’t find any exceptions during testing. Each of the three factors above are guaranteed to produce an undercount. The only question is…

How severe is the undercount?

An estimate of the total bird casualties from Ivanpah was provided in testimony to the California Energy Commission by ecologist K. Shawn Smallwood from the Center for Biological Diversity, which the AP article above describes as an environmental group. Mr. Smallwood estimates there are 28,000 bird fatalities at the Ivanpah facility.

How did he get that number?

Chris Clarke, writing at ReWire helps out by Deconstructing the Dispute Over Ivanpah Bird Kill Numbers.

Starting point is the reported count of dead birds in April and May 2014 of 183. A count later in the first six months is more accurate because the facility has been gearing up most of that time.

“Scavenger bias” is the portion of birds that don’t get eaten or taken by ravens, foxes, and other predators. I’m guessing coyotes are in the ‘other’ category. The researcher estimates a scavenger bias of 20%, which means five times as many birds are killed as are found by surveyors.

That gets the ecologist to 473 per month.

I get a slightly different number: 183 / 61 days in 2 months = 3.0 per day x 30 days = 90 per 30-day month x 5 for scavenger bias = 450.

A 31 day month would be 465.  Small difference. Let’s proceed.

The staff at Ivanpah only survey one-fifth of the site. Assuming the birds don’t just drop near the tower would mean we need to multiply that tally by five again. Observations by federal employees that streamers fall outside the perimeter of the facility means the underlying assumption is false and the reported count must be increased by some factor.

So the 473 per month falling inside the tiny survey area multiplied by 5 is 2,365 per month for the full facility.

Multiply that by 12 again suggests that around 28,380 birds meet their end at the Ivanpah facility a year. Let’s round that to around 28,000.

Keep in mind an unknown portion of those birds are protected under the migratory bird act. By the way, each such death of a protected bird is a federal crime.

Also keep in mind the proposed Palen facility will have three towers, instead of two at Ivanpah. Each of the towers will be taller. I recall a comment somewhere that the solar flux will hit 1,000 degrees at Palen instead of 800 degrees at Ivanpah. If it matters to readers, I’ll find the source.

Resource articles:

• 8/23 – Wall Street Journal – Spontaneous Solar Combustion – The WSJ weighs in on the 28,000 count and shows a hint of schadenfreude from environmentalists fighting with environmentalists.
• 8/18 – AP – Emerging Solar Plants Scorch Birds in Mid-Air – the article that kicked off the attention this week.
• 8/21 – ReWire – Deconstructing the Dispute Over Ivanpah Bird Kill Numbers – As usual, Chris Clarke explains the details so non-scientists can understand. He has been covering this story for a long time.

What do you think?

What scavenger bias rate would you use if you think 20% is too high?

Would you use a higher factor knowing the site is only surveyed every three or four weeks and that each survey takes several days, giving those cute little foxes time to get some dinner each evening during the survey?

What adjustment factor would you suggest for the intentionally unsurveyed area?

How would you increase the estimate for streamers landing yards or miles outside the perimeter?

How would you change the estimate for seasonal factors?