[obol] Thrush migration maps "cool" but way too smooth
- From: Joel Geier <joel.geier@xxxxxxxx>
- To: obol@xxxxxxxxxxxxx
- Date: Tue, 29 Apr 2014 07:49:39 -0700
Hi all,
There's an old saying among exploration geophysicists:
"If the noise level is low, you probably forgot to connect a wire."
The animated maps on the ABA website that Tom Crabtree mentioned are
indeed fun to watch, and perhaps useful in terms of gaining a very
broad-brush, continental-scale sense of thrush migration.
http://blog.allaboutbirds.org/2014/04/23/spring-thrush-animated-migration-map-identification/
However, the depiction of a broad migration front sweeping smoothly
northward is deceiving. It likely results from excessive smoothing of
the ebird data by whoever produced these maps.
There are some really crazy things that make this obvious. For example,
watch what happens along the North Dakota border during May. Apparently
huge masses of Hermit Thrushes, then Swainson's Thrushes, suddenly show
up in Saskatchewan and Manitoba, then spread backward into the northern
prairie states, before bouncing back into Canada as if propelled by
gigantic rubber bands.
Those must be some very confused thrushes!!
Of course it doesn't really happen that way. However you can create maps
like this by inappropriate use of geospatial/temporal smoothing
algorithms, This happens easily when your observations are strongly
clustered (as is usually the case for ebird data), and especially if you
impose artificial/political borders (such as the Canadian and Mexican
borders, in these maps) on your smoothing algorithm.
I may have looked more closely than others at the Upper Midwest part of
the map, because I was curious (following a note from Mike Clarke, who
mentioned that Swainson's Thrushes commonly sing in migration in
Indiana) why I never noticed Swainson's Thrushes singing in Renville
County, Minnesota, where I lived & watched birds for most of my teenage
years.
A quick side note for folks who are not familiar with the details of
Minnesota geography: Renville County is one of the biggest counties in
southern Minnesota, roughly the size of Marion County. It was
historically prairie, but its NE corner is only about 10 miles SW of the
historic edge of the "Big Woods" eastern hardwood forest ecosystem. The
Minnesota River runs diagonally along its SW boundary, in a deep river
valley with forested bottomlands (and incidentally, exposures of some of
the oldest rocks in North America, the Morton Gneiss which at 3.5
billion years old is nearly ten times as old as the oldest rocks in
Oregon). Ft. Ridgely State Park (one of the few places with SWTH
observations) is also interesting/infamous as the epicenter of the 1862
Dakota War.
It turned out, when I took a quick look into the ebird database, that
there are NO records for Swainson's Thrush from Renville County away
from the wooded Minnesota River Valley, at least for the past 10 years.
There are a handful of records from the neighboring towns of Willmar and
Hutchinson (both proudly signed as "Tree City USA," so effectively
outposts of the eastern forests). By far the majority of reports from
southern Minnesota come from east of the historic prairie/big woods
boundary.
So, I guess it's not surprising that I never heard, nor saw, a
Swainson's Thrush on our farm, for all of the hours that I spent combing
the mature hardwood grove that served as our windbreak.
I suspect that a more carefully drawn map would show a much more
irregular front for Swainson's Thrushes moving northward, funneling
mainly through the historic big-woods region of eastern Minnesota, with
a few angling NW up the Minnesota River Valley, and/or using it as a
refueling stop before making another long-haul flight across the
northern prairies to reach the boreal forests of Alberta and
Saskatchewan.
I'm sure the Minnesota/Dakota region is not the only place on the
continent where excessive smoothing makes these maps misleading. It's
just a place for which I happened to be curious enough to look into the
data. If you pick your own favorite patch and drill down to the data,
you could well find similar problems.
Bottom line, these maps -- like many of the maps that emerge from ebird
these days -- can be fun to look at. However, there's a certain amount
of mathematical fiction that goes into producing any maps that come out
looking as tidy as these ones do. It's better to view them as artistic
creations rather than as scientific plots.
We should be very wary of relying on these maps to judge the validity of
individual reports of migrant Catharus thrushes. To be fair, the author
of this blog posting doesn't go that far: She only suggests that they be
used to get a general sense of migration timing, as one of four basic
clues for sorting out the identity of individual birds.
Good birding,
Joel
On Tue, 2014-04-29 at 01:07 -0400, obol@xxxxxxxxxxxxx wrote:
> Date: Mon, 28 Apr 2014 11:12:53 -0700
> Subject: [obol] Re: Spring Arrival Dates: Long and likely annoying for
> some
> From: Wayne Hoffman <whoffman@xxxxxxxx>
>
> Thanks, Tom -
> Cool maps!
>
> The picture for these birds is more complicated locally than this
> article
> shows, because in the west we have multiple subspecies that behave
> somewhat
> differently.
...
--
Joel Geier
Camp Adair area north of Corvallis
OBOL archives: www.freelists.org/archive/obol
Manage your account or unsubscribe: //www.freelists.org/list/obol
Contact moderators: obol-moderators@xxxxxxxxxxxxx
Other related posts:
