[asa] Climate Change and Hard Drive Controller Design

While I dabble in a whole bunch of different scientific endeavors what
I do for a living is design storage semiconductors. The first chip I
worked on in 1982 is in the Smithsonian. It was the first SCSI
controller and was done in four micron NMOS. Now I work in 40nm CMOS.
The challenge of doing a read channel for a hard drive controller is
immense. The dimensions on a drive have decreased and the speed coming
off the analog interface has increased -- both dramatically. What this
causes is a very, very noisy environment. Only a few companies in the
world have the knowledge to do read channels well. The drive
manufacturers are demanding better and better signal to noise ratios
(SNR) so they can continue to produce faster, smaller, and larger
capacity drives. There is an analogous problem in cell phones and
wireless communication (IEEE 802.11x).

One of the "tricks of the trade" is called a Viterbi algorithm (http://en.wikipedia.org/wiki/Viterbi_algorithm
). This computes a maximum likelihood path sometimes also called a
Viterbi path. Software and hardware Viterbi turbo coders are found
ubiquitously throughout the storage and communication spaces. Viterbi
algorithms are also standard fare in information theory. (One of my
pet peeves is how poorly the intelligent design movement understands
information theory. From what I heard from Randy the same is true with
the information theorists he knew when he worked for IBM.) If this is
found everywhere how do you get a competitive advantage? It turns out
if you "remember" more of the observations back in time you get better
SNR. Adding memory (in the right place) makes for a better read channel.

What does this have to do with climate? It's also a "noisy"
environment where climate is the signal while weather is the noise.
Randy recently asked me my opinion of a piece written by Roger Pielke
for Physics Today. One of the (many) flaws in the piece was how far
back in time he went to find trends. He only went back a few years.
Just like in my Viterbi coder example above you need to go back in
time far enough to tease out the climate trend, aka the "signal". The
use of a "too short" time frame is a common technique for all the
denialists. For example, they were all over the January 2008 numbers
because we were deep in La Nina period. Because the 2008 numbers
include January they tout this as "cooling off". (cf. Lynn's second
link in an unattributed quote of Janice (or vice versa) http://www.calvin.edu/archive/asa/200812/0272.html
, http://tinyurl.com/7kk6zp. In order to avoid treading on a forbidden
subject I will leave clicking through either of the links above as an
exercise for the reader. Look for claims of cooling in 2008 and look
at the time frames used.)

Climate needs to be tracked over the decadal time scale otherwise you
will be overwhelmed with "noise". Climate models make specific
predictions for these time scales not only at the surface but also at
different altitudes. For example, Jim Hansen's 1998 B scenario
predicts the following rate of increase for surface temperatures in
degrees C per decade: 1984-2008 0.19+/-0.05. It predicts roughly the
same rate of increase for the troposphere and a rate of decrease for
the stratosphere. This is because CO2 acts as a blanket and the warmth
is under the blanket. If the warming is cause by the Sun then the
stratosphere would also increase.

First up, the three major surface measurement organizations.

1. GISTEMP: 1978-2007 0.17+/-0.04ºC/dec;
2. NCDC: 1979-2008 0.16+/-0.04ºC/dec;
3. HadCRUT: 1979-2008 0.16+/-0.04ºC/dec;

So, the error bars overlap for the model versus the surface
observations. One of the complaints by the denialists is that the
satellite data doesn't match the surface data. Not true. Again, the
climate models predict roughly the same amount of increase for the
troposphere. Here's the 30-year trends for the lower troposphere:

1. UAH: 0.13 ºC/dec
2. RSS: 0.16 ºC/dec

The mid troposphere looks like this:

1. UAH: 0.11 ºC/dec
2. RSS: 0.16 ºC/dec
3. RATPAC (50-year trend): 0.15 ºC/dec

The stratosphere looks like this:

1. UAH: -0.67 ºC/dec
2. RSS: -0.66 ºC/dec

One thing that John Cristy doesn't mention is his mid troposphere data
needs to be corrected. The satellites are not precise enough and the
temperatures bleed between the stratosphere and the troposphere. He
still touts his bogus numbers even though the systemic error has been
known since 2004. See http://www.ncdc.noaa.gov/oa/climate/research/nature02524-UW-MSU.pdf
  for an explanation how the satellite error gets corrected.

Not only does the data confirm the climate models it also is only
consistent with anthropogenic global warming because of the
stratosphere cooling. One thing my little bit of analysis did not
cover was how global warming maps spatially. Another specific
prediction made by the climate models is that the warming is greater
in the polar regions than over the equatorial and temperate regions in
what is called polar amplification. That is the anomaly is greater at
the poles than like here in the lower 48. One complaint about the
measurement accuracy is the so-called urban heat island effect which
makes the surface measurements supposedly unreliable. First of all,
see how the radiosonde measurements match the surface network here: http://www.ncdc.noaa.gov/img/climate/research/2008/ann/ts-sfc-radiosonde-jan-nov-pg.gif
   Second of all, If that was significant the anomalies would be
greater in the temperate regions where the cities are rather than in
the Arctic. So, we can test this hypothesis. Here's the final world
map for 2008 by the NCDC: http://www.ncdc.noaa.gov/img/climate/research/2008/nov/map-blended-mntp-200801-200811-pg.gif

Note how much cooler the U.S. is compared with Northern Siberia. In
order for the denialists to hold on to their position they need to do
one of two things either not correct the errors in their data calling
themselves the mainstream and everyone else the outlier, or shorten
the timescale of their data in order to decrease the signal to noise
ratio or most frequently both. Good thing I am rather than they are
designing disk controllers.

Rich Blinne
Member ASA

To unsubscribe, send a message to majordomo@calvin.edu with
"unsubscribe asa" (no quotes) as the body of the message.
Received on Sat Dec 20 15:52:35 2008