The largest and to date the most comprehensive experiment to soak up greenhouse-gas emissions by artificially fertilising the oceans set sail from South Africa earlier this week.
more at:
http://www.newscientist.com/article/dn16390-climate-fix-ship-sets-sail-with-plan-to-dump-iron-.html

Wow, what a way to rush into the future. This has enormous potential for unintended consequences … perhaps we can sue them in the future.

Of course, I’ve spent as much time writing software bugs. One of these
years I’ll learn the job.

Ahhh, so you are the guy who creates those bugs I have to find?

One of the great benefits that computer simulations like SPICE brought to circuit design; especially analog circuit design, is that in the old days, we just used the rules that we learned from the textbooks, but the one thing they never ever taught in school about electronics, is that they aren’t any use to you until you turn them on; as in apply power to them.

So nobody ever thought much about what happens while the power supply is rising from zero Voltage up to some operating value; so a lot of circuits were designed that were perfectly capable of going into a latchup state at some supply voltage level, that they couldn’t get into at the correct Voltage, but couldn’t get out of one they were latched.

So throught the magic of transient analysis, we can watch the stability during the power up transient condition.

I gave up on writing code; as soon as I discovered “Smith’s First Law of Software Code.”

“When writing software code; no matter where you start, it is always necessary to do something else first !”

George

I’ve spent 40 of the last 50 years designing circuits that weren’t supposed to oscillate and did; or circuits which were supposed to oscillate and didn’t.

And they pay you to do that? :-)

Of course, I’ve spent as much time writing software bugs. One of these
years I’ll learn the job.

Those “registers” that you ’simply load a number into” consist of bistable memory elements (flip-flops) which at least in the silicon realm require positive feedback. Even in a dynamic memory cell (single transistor), the current state of the cell, which is just a charge on a gate capacitor, won’t be maintained except fo microseconds of time, unless it is constantly “refreshed” by a circuit, which senses the present charge state; decides whether to call it a one or a zero, and “reloads” that state into the cell which is a positive feedback enhancement of the charge condition on the capacitor.

So you “software” won’t work without the hardware to make those registers or memeory locations, and maintains the data in them which in dynamic memories requires constant pulsed positive feedback to goose the collapsing state back to where it belongs. Static memory cells which typically may require six transistors; are simple bistable multivibrators which most definitely store their state as a result of a positive feedback loop that drives them to saturation in one limit state or another.

I’ve spent 40 of the last 50 years designing circuits that weren’t supposed to oscillate and did; or circuits which were supposed to oscillate and didn’t.

I’ve been everywhere from “valves” (vacuum tubes to you) to bipolar transistors; tunnel diodes, PMOS NMOS CMOS, magnetic cores, and every other kind of possible oscillating or switching circuit or device. I’ve even built digital counters out of rotary solenoid driven switches (they don’t forget where the are when the power goes out).

I may only have a batchelors degree, but then i had five majors, including RadioPhysics in that lineup.

So people can’t fool me with hoky ideas of what constitutes feedback.

Don’t ask me about X-rays though; because although I know the basic Atomic Physics spectroscopy of X-rays, I’ve never actually worked in that region of the electromagnetic spectrum; but damn near everywhere else from DC to Cosmetic Rays.

George E. Smith (13:18:00) :

And if you note that the global warmers insist that small CO2 warmings are amplified by water vapor positive feedback into large warming effects; then I would say that the models (time static) proposed for CO2 warming with water vapor feedback almost certainly guartantee that they would oscillate; well they would if they were true. And because of the delays in climate systems, its a fairly safe bet that they must oscillate if they provide any warming amplification at all.

Going along with that train of thought, hypothetically, suppose the input level (CO2) is increasing and that the water vapor feedback is true, what would that imply about the frequency change of the oscillations over time, and what is the mechanism by which the oscillation Y axis value at a given point in time is driven down? “”

Michael, I am not sure I grasp what it is you are asking; but I’ll take a whack at it any way.

First off, when you have an oscillating system, you have a sytem that is inherently unstable. If the “loop gain” is less than one, then it isn’t an oscillator. Any transient input disturbance will eventually die out. If the loop gain is greater than 1 and given certain other conditions like phase shifts (time delays), the output would build up without limit if the system was linear; but systems never are, so eventually non-linearities will bring the loop gain at some output level down to one, and the amplitude of the oscillation would be limited at that point.

If you start of with extremely small values of CO2 (or water or other GHG), then each additional molecule of CO2 (or what have you) will absorb thermal radiation on its own, and the warming effect would be linear with GHG concentration. But the amount of emitted thermal radiation gets reduced by the lower layers of the GHG, thus reducing the input to the higher layers, which tend to reduce the same fraction of the input they see, as do the lower layers.

When the total concentration gets high enough the remaining amount of unabsorbed radiation is so small, that not much more warming can be obtained by adding more GHG, and this is where you enter the realm where the response to more GHG becomes logarithmic; which is tantamount to saying that each doubling of the GHG reduces the residual radiation by the same decremental amount, and if you presume that the warming is linear with captured energy, then the temperature increase would follow a logarithmic law too.
Now none of this has anything to do with what frequency the system should oscillate at. The frequency tends to be controlled by energy storage processes and energy transfer between such processes.
For example with a weight boucing up and down on a spring, the frequency depends of the mass of the “weight”, and the stiffness of the spring, and so long as those are constant, the frequency won’t change with amplitude of the oscillation.

A thermal oscillation such as could occur in a climate situation, would have a frequency that depends on things like the heat capacity or specific heat of heat storage materiasl such as rock or water, and also on the rate at which heat energy can move from one place to another.

Heating processes are generally slow compared to other physical phenomena, so the delay between applying some thermal energy, and the receiving body converting that to a temperature rise, is what would determine a period of oscillation.

Well climate thermal processes aren’t too linear, and there is a lot of chaos to contend with too; but I have never seen a time dependent feedback equation for any of these “climate sensitivity” models, that one could use to determine whether it should oscilalte.

One thing that seems to get lost in the shuffle, is that water vapor, and liquid water in clouds are quite absorbing in the same 5-50 micron wavelength range where the earth radiates; whereas CO2 absorbs in the range from about 13 to 16.5 microns centered at about 14.77 microns.

So water vapor is plenty capable of grabbing the very same thermal radiation that CO2 grabs, and that effect has to diminish the effectiveness of CO2. Moreover, there is no magic to CO2 as far as starting “positive feedback warming”. The water vapor is perfectly capable of doing that all by itself; but the difference is CO2 doesn’t form clouds, like water does, and the water clouds eventually produce a strong negative feedback, by preventing solar radiation from reaching the ground; which is one of the reasons that I believe it is surface temperatures that should be measured and not air temperatures.

Don’t forget, we can vote once every 24 hours: click
And, don’t forget to write down the time you voted, so you’ll know when you can vote again, and to help avoid “time creep” (before you know it, your vote time will be past your bedtime, and you can end up skipping a vote).
Looks like the spread between WUWT and Phary has settled in to right around 12%.
Comfortable, but keep the pressure on!

Apparently the climate models in use by the warmers consider increased water vapor in the air a positive feedback. So in their models a forced temperature increase from CO2 causes increased evaporation which adds water vapor to the atmosphere increasing temperature further increasing evaporation even more, etc. and the process goes into “runaway greenhouse”.

Actually (leaving Hansen’s latest claims of what might happen for large enough forcings aside), the climate models do not produce a runaway greenhouse. Rather, the water vapor feedback produces an amplification of the effect. As a simple example that is close to the actual estimated value for the feedback, if a 1 C rise in temperature due to increased CO2 (or whatever) causes additional evaporation that produces 0.5 C of additional warming then this additional warming will produce additional evaporation that causes 0.25 C of additional warming…which will then add an additional 0.125 C of warming…and so on. What you have is an infinite series but one that converges to the value of 2 (i.e., the water vapor feedback doubles the warming).

According to Roy Spencer, Ph. D. the observations are the opposite. Increased water vapor acts as a negative feedback. That seems to be the primary disconnect between the “climate” models from people such as Hansen and the observed reality.

(1) Actually, as I understand it, Spencer is claiming a negative feedback due to clouds, not water vapor. Of course, it is water vapor that condenses to form clouds, so the two are interrelated. Nonetheless, in climate science there has generally been a distinction made between the direct effect of water vapor and the effects of clouds (i.e., condensed water vapor in the atmosphere). I don’t think Spencer’s work argues against a positive water vapor feedback at all.

(2) It is Spencer’s analysis of the observations that shows the opposite. It is not the observations themselves. That is, there is a non-trivial amount of analysis involved to get this result from the observational data. Spencer’s claims are in contradiction with a lot of other peer-reviewed work and I believe that most of his work has not yet appeared in the peer-reviewed literature yet…Some of it may have very recently but other scientists have not had a chance to respond (although there have been some responses in the blogosphere to certain parts of his work, e.g., one over at RealClimate and another over at tamino’s blog). It is usually not wise to “hang your hat” on work that other scientists have not had a chance to evaluate, especially when it contradicts a lot of other work that has been in the peer-reviewed literature for a long time. As just one example of the many problems with Spencer’s work: It would seem to be very difficult to explain how the ice age – interglacial transitions occurred if the climate system really has such a strong negative feedback in it.

I believe his paper shows his figure to be something like a +0.5C change from doubling CO2. And one must remember that response to CO2 increase is logarithmic so one must greatly more than double CO2 again to get another 0.5C increase.

Your last statement here is simply wrong. If a quantity A has a logarithmic dependence on a quantity B then each successive doubling of B produces the same increment of change in quantity A. This is opposed to a linear dependence which would mean that each fixed increment of change in B (e.g., an increase of 100 ppm) produces the same increment of change in quantity A.

Scientists understand that the dependence of temperature on CO2 concentration is expected to be approximately logarithmic which is why they talk about the amount of temperature rise from a doubling of the concentration. (If they expected a linear rise, they would instead talk about, say, the amount of temperature rise from a 100ppm increase in concentration.)

“Have you found http://www.giss.nasa.gov/tools/modelE/modelE.html ?”

Sadly yes. As you discovered on your own, who knows what the various subroutines are really doing, given all the obvious hacks, lack of comments, etc.. And not a differential equation in sight…what a mess!

“A minuscule amount of money from the financial impact of ModelE would do wonders.”

This really isn’t about money (they could easily fund a documentation/validaion effort if they wanted) – it comes down to *** priorities ***, as defined by the principal investigators and project managers. Clearly, they feel that documenting what they do with model E (and gistemp for that matter) is not a priority. Instead, blogging, speaking out against coal-fired power plants and defending vandalism in the name of AGW are apparently more urgent…

Thank you for the link! When I have some time (we will be in an ice age by then) I would really like to begin digging into some of this code. Its been a while since I have worked with Fortran, but I am sure I can tear in to it and get up to speed pretty quickly.

As for your comments regarding the coding quality and such, I have to agree with you completely, as I described in my prior post.

I have been around the software industry for a time, working for the DoD, Navy, DHS, for various software companies, manufacturing companies and the like. With exception to my government work, all of these have operated on relatively shoestring budgets, yet in each instance (except gov.) I have always been required to produce maintainable, quality code using proven coding practices. I have spent many years developing source control management processes, release management and coding standards. When I was working for the government, I brought source control management, release management and coding standards to them because “I” required it and because they lacked any trace of it. Out of all of the places I have worked, the government was, by far, the most lackadaisical about source control, release management and coding practices. The sources presented by your links illustrate this problem pretty well.

Perhaps I am just getting old and have become nit picky in my ways, but I can make some very strong cases as to why proper development practices are critical in any software development, no matter the budgets or time constraints. In fact, the smaller the budget and tighter the time, the more important the practices become. I believe the whole reason why this is not done in most government projects, is simply because they don’t care. They are working with YOUR money and nobody holds anyone accountable for anything. This is why private industry can produce so much better than government can.

And when you are talking about these climate models, for all the money surrounding the AGW garbage, for all the money that is being slated to be spent on radically changing our lives, the software better work and it better be balls on accurate! This notion from Hansen that “our program blows up”, for me, has to take the cake. Any software that “blows up” is “half baked”. Put it back in, its not done yet!

Simply put, this is all a joke, and I completely dismiss any results produced by this kind of crap. Just doesn’t fly with me. If I can’t trust your code, you expect me to trust your results?

My 2 cents anyway.

And, Happy New Year to all!! I truly hope this year is a great for all!

WUWT: Congratulations on the WebLog Awards Finalist, and I am voting for you every day! In my book, this is bar none the greatest climate and science blog on the planet. Great bunch of people here!

I’m not trying to bash the language itself. I believe in using the right tools for the right job, and certainly Fortran has its place. I was simply surprised as I was expecting something more like “SIM Climate” and instead it looks more like “Pong” with a missing paddle. Just wasn’t what I had expected.

I personally cannot excuse the poorly written, poorly structured coding practices when considering that billions of our tax dollars go into creating this stuff, and after considering the tillions of dollars consequences that are hanging on the results, I just gotta step back as it truly takes my breath away.

> (Finally, a chance to use the way it wants to be used!)

I forgot to change the spaces to " "s. Silly me. Sorry about the formatting....

Ric Werme (07:09:12) :

“It’s actually better than I expected. It’s not a commercial product, it’s research.”

I’m sorry but this doesn’t fly with me. While research codes are often written with expediency in mind (versus careful documentation of procedures), in the case of Model E, the results are being used to actively promote a specific view of climate change.

I have no disagreement with that! Some random research hack has managed to morph into something that is influencing major policies throughout the world without anyone vetting, cleaning, documenting, and likely fixing the beast. I’m merely saying I’m not surprised that people managed to forget that “little” step. I am surprised that in my perusal of the code I didn’t see much in the way of diagnostic hooks, unit tests, or much of anything I’d put in to monitor data munging where it’s hard to tell if the output is right or not. It’s usually pretty easy to tell if a compiler is generating good code or if a file system is holding data you wrote. As long as the model output has believable temperature and doesn’t create more water than is already on the planet, how can you tell if it represents the intention of the programmers?

A minuscule amount of money from the financial impact of ModelE would do wonders.

Ah well, somehow Bernie Madoff’s 50 G$ Ponzi scheme got missed in the SEC’s review too.

Ric – can you tell me from perusing the code what equations are being solved and the subroutines where they numerical algorithms for these equations are implemented? There are some comments in parts of the code and virtually none in others. And it doesn’t matter to me that the code is written in Fortran – I would expect that a scientific code written in C++, Java, Python, … would have proper commenting and pointers to single, comprehensive documentation manual (rather than a collection of papers which say almost nothing about how the models are implemented in the source code).

Have you found http://www.giss.nasa.gov/tools/modelE/modelE.html ?

I poked around a bit. Having virtually no knowledge of the equations behind a lot of the modeled physics meant all I could do is pattern recognition. In particular, the simulations and graphics hacks I’ve written tend to have a few lines of code doing the math and huge amounts of infrastructure, user interface, etc. Not much user interface in ModelE, though. For example, in the DYNAM subroutine, this code:

DO J=J_0S,J_1S ! eastward transports
DO L=1,LM
I=IM
DO IP1=1,IM
AIJ(I,J,IJ_FGZU)=AIJ(I,J,IJ_FGZU)+
& (PHI(I,J,L)+PHI(IP1,J,L))*PU(I,J,L)*DT ! use DT=DTLF/2
I=IP1
END DO
END DO
END DO


(Finally, a chance to use <code> the way it wants to be used!) This code is clearly updating some grid. The code browser says this is in ATMDYN.F so it’s probably something to do with atmospheric dynamics. http://www.giss.nasa.gov/tools/modelE/modelE.html says “ATMDYN.f, ATMDYN_COM.f: Code for the calculation of atmospheric pressure gradients, and dynamic time stepping.” The only other reference to AIJ is some external reference. DT is probably delta-T, the time step size, I can’t guess what DTLF is. So, an apparently meaningless array name, a too-brief comment, I have no idea what it’s computing. Earlier I was thinking it would be interesting to include bibliographic references to each loop like this so one could go off and read up on the physics.

Some code has comments that looks like it was written straight from a recipe:

c**** 2.2 First test: valid dt/dz ?
…
c**** 2.3 dtdz is valid > something in German
c**** —————————————-
c**** 1.lineare in p^kappa (= Dieters neue Methode)

Others mean either I don’t understand the context or worry me:

C**** Restart after 8 steps due to divergence of solutions

though that appears to be referenced in http://www.giss.nasa.gov/tools/modelE/modelE.html#part3_1

If I needed to learn about this code, I wouldn’t start with the source. I’d pick some facet, say warm air advection or convection, avoid complexities like condensation, review the physics and then look in the code for some of the math. I’d get frustrated, go off for a few days learning about the important arrays and what’s in them, then go back to the code and it would start making a little more sense.

Oh – in answer to your question – No.

And if you note that the global warmers insist that small CO2 warmings are amplified by water vapor positive feedback into large warming effects; then I would say that the models (time static) proposed for CO2 warming with water vapor feedback almost certainly guartantee that they would oscillate; well they would if they were true. And because of the delays in climate systems, its a fairly safe bet that they must oscillate if they provide any warming amplification at all.

Going along with that train of thought, hypothetically, suppose the input level (CO2) is increasing and that the water vapor feedback is true, what would that imply about the frequency change of the oscillations over time, and what is the mechanism by which the oscillation Y axis value at a given point in time is driven down?

But these days you don’t have as much need for conventional oscillators if you have adequate cash to substitute. You can run RF directly into a DSP chip and do all your demodulation with software. At least that was the state of the art in the late 1980’s. I haven’t worked with such things in years, though.

Except there’s a catch – the D in DSP stand for digital logic (okay, really it’s binary) and digital logic uses lotsa flipflops and flipflops use positive feedback to remember their state.

(You folks who have no idea what the difference is between a D flipflop and a JK flipflop can flip a light switch. The ones with a good snap to them are mecahnical flipflops.) And no, the JK flipflop is not named for John Kerry. (USA joke – John Kerry ran for president and roundly criticized for changing his mind. At least Al G is consistant!)