From the Vienna University of Technology , what appears to be a negative drought feedback mechanism has been observed.
Summer rain more likely over drier soils
Summer rain is more likely over drier soil – this is the conclusion scientists have drawn from a detailed analysis of satellite data. State-of-the-art computer models predict the opposite effect; these models must now be reconsidered, says the study published in the journal “Nature“. Several international research groups were involved in the project: The Centre for Ecology and Hydrology (Wallingford, UK), the VU University Amsterdam, the Center of Meteorology CNRM in Toulouse, and the Vienna University of Technology.
Convective Showers: Hot Air Moves Up
Frontal rain systems, moving from the ocean across the land, can lead to rain over large areas. Summer showers, which frequently occur at the end of a hot day, are often restricted to a rather small region. This kind of rain is a completely different phenomenon. Instead of moving across the land, the air moves from the hot ground upwards, forming clouds high up in the air, and finally leading to rain. This is called “convective precipitation”.
Does Soil Moisture Lead to More Rain?
“It’s tempting to assume that moist soils lead to higher evaporation, which in turn stimulates more precipitation”, says Wouter Dorigo (Vienna University of Technology), one of the authors of the study. “This would imply that there is a positive feedback loop: moist soils lead to even more rain, whereas dry regions tend to remain dry.” But observations suggest otherwise: “We have analyzed data from different satellites measuring soil moisture and precipitation all over the globe, with a resolution of 50 to 100 kilometers. These data show that convective precipitation is more likely over drier soils”, says Wouter Dorigo.
The new data contradicts established computer models. A conclusive explanation for this effect has yet to be found. “The air over dry soils heats up more easily. This could lead to a more intense vertical draft”, Dorigo suspects. However, this cannot yet be described at a sufficient level of detail with today’s computer simulations.
Microwaves from Space
Soil moisture can be measured with satellites using microwave radiation. Unlike visible light, microwaves can penetrate clouds. Satellites can either measure the Earth’s natural microwave radiation to calculate the local soil moisture (passive measurement) or the satellite sends out microwave pulses and measures how strongly the pulse is reflected by the surface (active measurement). From this data, the soil moisture can be calculated.
here’s a second press release:
From the Centre for Ecology & Hydrology
Parched soils trigger more storms
Afternoon storms are more likely to develop when soils are parched, according to a new study published this week in Nature which examined hydrological processes across six continents.
The results have important implications for the future development of global weather and climate models which may currently be simulating an excessive number of droughts.
The research team included scientists from the UK, Holland, Austria and France and was led by Dr Chris Taylor from the NERC Centre for Ecology & Hydrology in the UK.
The scientists examined imagery from weather satellites which track the development of storm clouds across the globe. When they matched up where new storms appeared alongside images of how wet the ground was, they were somewhat surprised.
Dr Chris Taylor from NERC Centre for Ecology & Hydrology said, “We had been looking at storms in Africa and knew that rain clouds there tended to brew up in places where it hadn’t rained in the previous few days. We were surprised to see a similar pattern occurring in other regions of the world such as the US and continental Europe. In those less extreme climates, with more vegetation cover, we expected the soil wetness effect would be too weak to identify.”
The researchers compared their observations with six global weather and climate models used to simulate climate change. They found that the existing models do the wrong thing, triggering rain over wetter soils.
The implication is that existing climate models are more likely to go into a vicious circle whereby dry soils decrease rainfall, leading to even drier soil conditions. The paper concludes that fixing this problem is a priority for scientists developing the climate models.
Dr Taylor added, “Both heat and moisture are critical ingredients for rain clouds to build up during the afternoon. On sunny days the land heats the air, creating thermals which reach several kilometres up into the atmosphere. If the soil is dry, the thermals are stronger, and our new research shows that this makes rain more likely.”
Co-author Dr Françoise Guichard from CNRM-GAME (CNRS and Meteo-France) said, “We need to improve climate models so that we get a better idea of what global climate change will mean on smaller regional scales over land.”
The research team came from the NERC Centre for Ecology & Hydrology in the UK, CNRM-GAME (CNRS and Meteo-France) in France, Vrije Universiteit Amsterdam in the Netherlands, and the Vienna University of Technology in Austria.
The behaviour of water vapour (the source of some two thirds of the supposed CO2 warming) remains, as the IPCC have described it, “poorly understood”.
There is really nothing new about this. Here in the Valley of the Sun, it is common knowledge that if you get a good monsoon rain one night, it will be someone else’s turn the next.
That corresponds to David Stockwell’s findings of increasing precipitation in Australia when the CSIRO’s models predicted increasing drought.
Tests of Regional Climate Model Validity in the Drought Exceptional Circumstances Report
But isn’t “negative feedback” secret code for “oil money”?
My, my calibrating models with empirical data and actual observations. How novel. Anyone who lives in a semiarid climate with a mountain range next door know all about his effect. I guess it was simply to obvious to write it down before.
FYI Other Publications of Wouter Dorigo, & Scholar
How novel! [sarc-off]
When I was growing up in Texas during the drought years of the early fifties, we were taught that our most common source of rain was the “air mass thunderstorm”. After the sun had heated the earth most of the day, the heat caused thermals to rise, condensing water to form clouds, and if we were lucky, to condense enough to bring rain before the cooling of the evening dissipated those thunderclouds.
These storms were always local, scattered about. I never associated their locations as eschewing damp ground, but it seems logical; I’ve never seen a dust devil (the archetypical local flat land thermal) where it wasn’t hot and dry.
I suppose I aught to cut these European researchers some slack. I doubt any of them have much experience with non- frontal storms.
cheers,
gary
Keep in mind that in arid climates rain storm development, does not necessarily mean that the rain actually reaches the ground.
In very dry climates you can have rain clouds every day with nary a drop reaching the ground (virga). The modelers should also consider the local humidity and height of the cloud base in their calculations to account for those situations where all the rain that develops in the cloud simply gets recycled back into cooler air of higher humidity, but little if any moisture reaches the surface.
(another minor monkey wrench in the models I suspect)
Larry
And this is a surprise to dry land farmers how?
I love the “models wrong” in the title.
Nothing like settled science.
But, why would we have, and keep collecting reams of weather data, other than its variability ?
Not a clear cycle to be found yet.
In any time frame.
State of the art climatcastrology models. Does anyone know of a naturally occurring positive feedback mechanism?
Skeptical Science will be reporting on this any minute now, yep, any minute now…
(/sarc)
“We need to improve climate models so that we get a better idea of what global climate change will mean on smaller regional scales over land.”
Notice the mandatory little genuflection of obeisance towards the high priests of AGW.
SHOCKING!!!! not really.
[/sarc] Actual research contraindicates the assumptions used for computer models? I never would have imagined that. [/sarc]
Any other positive feedback, run away catastrophes missing in action? Anybody? Hansen? Is your hand up? no, your scratching your head, time for more tweakage I guess.
Dryland farmers strip field. What that means is that they divide their fields into strips, Every other strip is then used year one, and the other strips used in year two. Why? So that soil moisture can be replenished. How? Sequential thunderstorms. It is how we irrigate on dry land. It works quite well.
I take it most climate researchers have never asked farmers one single question.
This is a 24carat crap: ”moist soils lead to even more rain, whereas dry regions tend to remain dry.” But observations suggest otherwise: “We have analyzed data from different satellites measuring soil moisture and precipitation all over the globe,” OTHERWISE?!?!
No, you have NOT!!! Where is topsoil moisture, water storages and ”water vapor” in the air -> attracts rain-clouds from the sea like a magnet . Where is dry – clouds go around the land as cars around a traffic island!!! Australia is a perfect example, so is Sahara, so is Brazil. Looks like those people cannot tell the truth, to save their lives… This post is a jewel, for when the whole truth is known – for when the time comes – for them to be put on the witness stand, under oath.
Sorry but this paper contravenes all accepted climate science practice and procedures.
If real data contradicts a computer model, that doesn’t prove that the theory behind the model is dubious, it proves that there’s some unknown factor causing the data to be distorted. More research grants are necessary in order for the correctness of the theory to be properly vindicated.
(Shouldn’t this have a “/sarc” tag? – mod.)
The atmosphere does not simply depend on what the local soil moisture content is at any given spot when it comes to how much precip will fall. All things being equal, the atmosphere is dynamic and not only convects but advects and large synoptic and meso-scale patterns determine where the rain will set up. The inner-workings of weather are absolutely chaotic and soil moisture is just one small part.
I live in Oklahoma where it has been dry the past two summers. The sun quickly heats the ground, but I’ve failed to see those thunderstorms and rain. Recently, I was in eastern North Carolina which is quite humid this year and it rained about every other day. I guess the model applies somewhere else.
This is of course a heat conduit between the warm earth and the lower atmosphere which is carried to the high atmosphere via conduction, convection, and radiation. When the water vapor condenses at altitude the heat released has a shorter trip to space. And of course the clouds increase the albedo. When the rain hits the ground it carries heat from the warm soil below the surface. Perhaps Trenberth’s missing heat is in ground water. There is a lot of water down there.
http://www.guardian.co.uk/environment/2011/aug/26/underground-river-amazon
The Univ of Vienna link is busted.
But more to the point, that’s Wolfgang Wagner’s turf.
Can a retraction or resignation and apology to Trenberth for not consulting the climate modeling community before publishing a paper showing the models are wrong be far behind?
If not…. why not?
Pamela Gray says:
September 12, 2012 at 6:38 pm
I take it most climate researchers have never asked farmers one single question.
Why would they want to do that for free when they can get the government to pay them for “cutting edge research”?
Hmmmmmmmm. You mean those daily rain showers we experienced in the tropics of Southeast Asia (during the wet monsoon) were the result of drier soil? And those rainless, relatively pleasant days during the dry monsoon were, maybe, the result of wetter soil? Who’da thunk?
If the soil is dry, the thermals are stronger, and our new research shows that this makes rain more likely.”
I very much doubt this. Wet soil has an albedo of close to .1, whereas dry soil has an albedo around .3. Which means wet soil absorbs at least 20% more solar energy than dry soil and that energy has only one place to go. Back into the atmosphere above the soil.
Ask any hang glider and they will tell you thermals occur over dark (low albedo) surfaces.
Otherwise, there are more dust aerosols over dry soils, which play a role in cloud formation (seeding) and can increase precipitation (although some aerosols decrease precipitation). And this may be at least a partial cause of more rain over drier surfaces.
When and if temperatures reach +3.0C, water vapour in the atmosphere is supposed to increase by +21.0%. That is literally what the climate models have built in (and yes I checked).
All that means is that it is going to rain way more often everywhere on the planet and it will just evaporate a little faster as well.
If water vapour is 21% higher, then it is just going to rain everywhere much more often. No way is the Sahara going to become more dry in that situation. Water vapour does not just avoid certain places on the planet. It is everywhere. There are places where prevailing weather patterns concentrate it more and places where it is concentrated less, but increase the overall global level by 21% and it is just going to be 21% higher everywhere.
Big deal, more rain and more evaporation. Almost all plants love this scenario. Now throw in some increased CO2 so that the C3 plants need less open stomata, and the entire planet is now capable of becoming one big forests.
The last time it was 1C to 2C warmer, 8,000 years ago, the Sahara was forest/Savanna, and large lakes and rivers occurred there. Our human ancestors thrived in the Sahara until about 4,000 years ago.
The last time, it was 2C to 3C warmer, 9 million years ago, the entire planet was one big forest and there were 50 different species of Apes on the planet, one of which lead to us.
Now, in reality, water vapour is not increasing in the atmosphere so all this theory/fantasy is for not until it actually does increase.
Thank god … now we will never have to worry about deserts ever forming … anywhere, ever and ever.
Not counter-intuitive. Given sunlit moist soil and sunlit dry soil, the warmed, moist air will do the best convection as water vapor seriously decreases air density, more than warming does. Thus, air will convect up more efficiently over moist and and overcome the rising dry air and descend over the dry land. Not so hard to understand, really.
Isn’t this finding, a direct rebuttal to the alarmist claim, that increased rainfall (warming climate) will only fall, where we are already receiving excessive rainfall? The warmists had stolen the benefits of increased rainfall and converted it into catastrophic desertification and flooding, despite the apparent contradiction. It would be nice to put THAT claim in the grave. GK
Monsoons heat the land, hot air rises, moist air is drawn in from the ocean and rain results as moisture condenses in the updrafts. This effect greened the Sahara for about 3,000 years after the Ice Age, and darker vegetation and less exposure of lighter-colored, reflective sand and soil, increased warming and the power of rising hot air to draw in moist air from the seas and ocean. However, soon the wind patterns reverted to the Ice Age pattern and desert reclaimed the Sahara. The effects are totally different comparing areas near oceans to those far removed, and these differences are not contradictory. Since personal anecdotes abound in these comments, let me add mine. After over 21 years in the Air Force, living and working in many lands foreign and domestic, I experienced both: many sudden, powerful thunderstorms in the dry summers of Wichita Falls, Texas (almost in Oklahoma), and the monsoon rains of the Philippines and South-East Asia. They were quite different, but both soaked me to the skin.
I’m afraid this study does not match what I and many Australian farmers have noted over many many years. An early decent big storm rain in one area and not in another very very often means
that the area that got the decent big storm rain keeps getting the decent rains all season, while the area that missed out gains much less all season.
Shower and storm clouds are also seen to follow national parks areas where extra moisture lurks and runs along rivers creeks where trees and more moisture lurk.
Sure, I have been in an area where a big storm rain came one day and all around did not, and yes the next day the storms developed over the areas all around first, but then storm downburst lines came in from all sides later in the afternoon and collided over us giving us an even larger huge rainfall the next day, again way more than the surrounding areas.
Instability depends on temp but also moisture increase the instability dramatically, as anypone who has plotted the old scew-t atmospheric storm diagrams will know!
And what happens when a front moves through dry areas and moist areas, the moist areas get more rain almost always. And a storm over dry ground has much less rainfall in it than one over wet ground, as someone has already suggested.
So, although I like the study knocking some of the big AGW computer programs, I feel strongly that this new study is not correct .
What you are describing seems to be a monsoon effect, quite different from the dry, far inland thunderstorm this study is focused on.
I should mention that I am not a farmer but an agricultural weather forecaster, so my experience of the dry-wet ground rain effects are gained not just from my experience but from the experience of listening to many farmers comments over the years.
I am describing SE Australian growing season Autumn to Spring weather not monsoon weather.
Philip Bradley: September 12, 2012 at 8:36 pm
“…wet soil absorbs at least 20% more solar energy than dry soil and that energy has only one place to go …”.
Over moist soil some of the energy will go into evaportion.
Putting aside island rain environments & elsewhere variables of temperature/atmosphere instability/irradiation/ humidity, it seems that the local aerosol size convected upward would explain how this report finds drier soil (rather than moister soil) influences rain to fall best. My comment is in the context of certain types of rain, namely convective and strati-form rainfall; but not in the context of a shallow rain type (characterized by a sparse daily amount falling) which doesn’t seem to be in the same dynamic.
Too small a radius dust (or smoke) particles will impede the droplets from precipitating. The moister soil offers a relatively smaller aerosol mote, whereas the drier soil provides a larger dust (aerosol) particle. In the context of semi-deserts/deserts there are lots of dust particles loading the sky, but their radius is too small to give results that were revealed in this posting’s study. (for orientation try J.Huang et.al’s 2009 “Large Scale Effects of African Aerosol on Precipitation of the West African Monsoon” & “The Spatial and Temporal Variability of African Dust Outbreak”)
Philip Bradley (September 12 at 8:36 pm) says:
“I very much doubt this. Wet soil has an albedo of close to .1, whereas dry soil has an albedo around .3. Which means wet soil absorbs at least 20% more solar energy than dry soil and that energy has only one place to go. Back into the atmosphere above the soil.”
No, it also has to warm up the ground, and water has a very large heat capacity compared with dry soil. To warm the wet soil to the same temp as the dry soil will take a lot more energy, and conversely when there’s only so much energy coming in, the wet soil (and the humid air above it) will warm far less than a dry system.
In the N. Atlantic rainfall is synchronized with the SST (the AMO), warmer the ocean more rain as this reconstruction (1700-2000) from N.W. Scotland shows
http://www.vukcevic.talktalk.net/AMO-recon.htm
Steve C, without additional rainfall wet soil will become dry soil. Thus all the additional energy absorbed by wet soil goes into the atmosphere by a combination of conduction, LWR and evaporative cooling over a time scale that varies with climate and season.
I strongly suspect some more complex process is at work here, involving lateral movement of air over scales of 100Ks+.
Its long puzzled me that convective thunderstorms are very rare here in Perth, despite our hot summer climate, high solar insolation and periods of high humidity during the summer.
It can be over 40C, with stifling humidity (near the ground), yet we have 100% blue skies all day.
The answer may lie in another unusual aspect of our weather, which is blindingly intense sunlight in the hour or so after dawn and before sunset. I am reasonably sure this is due to low levels of aerosols in the mid-troposphere and minimal scattering of sunlight. As the air over Perth comes from over the Indian and Southern oceans, it seems land surfaces get more aerosols higher in the troposphere than generally realized.
This is why climate scientists need to fiddle less with models and go for observations. Satellites can be pesky flies in the ointment of CAGW unless they show the Arctic minimum ice extent. 😉
Also
moist air moves up …
See: http://docs.wind-watch.org/More-Rain-for-the-Town-of-Norden.pdf
Juergen Michele
In the UK, summer often consists of two hot days and a thunder storm, so I guess that just about fits their findings.
Reading the article and the comments would suggest that, although the theory may be correct globally, there are areas on earth where the opposite appears to be true.
Suggests the most common conclusion in climate science: forming unified theories is not always useful to those working on the ground.
100 sites/areas globally to repeat this study to see what gives?
I think I see what the problem is, that makes this result counter-intuitive.
Climate Science™ default position:
The Earth’s systems are so delicate and finely balanced that mankind’s miniscule influence can catastrophically disrupt them. Assume positive feedbacks, like that which would lead wet areas to become wetter and dry areas to remain dry. Program models accordingly.
Reality:
The Earth’s systems are robust and so highly resistant to change that mankind’s miniscule influence is hardly noticed with its resulting effects temporary at best. Assume negative feedbacks as part of interconnected global self-correcting mechanisms that maintain the stability of Earth’s systems.
I think it would be best to ask an experienced forecaster when dry land gets a sort of vicious cycle going, and when it does not. I know I have witnessed different patterns in the USA.
In the plains a “heat high” can get established, as it did last summer, and the storms seem to form all around the edges but not in the middle, as a sort of “ring of fire,” (or perhaps “ring of lightning.”) The dry land seems to perpetuate drought as long as the weather remains hot.
On the other hand, in the high deserts of the Four Corners area, the heat of May and June seems to lead to puffy clouds or perhaps “lady rains,” (Navajo description of rain that doesn’t hit the ground.) In July and August the clouds get bigger and produce “man rain,” though the ground is drier, and you experience what some call the “monsoon,” (though others say it isn’t the same as a true monsoon.)
It seems scientists have discovered heat thunderstorms.
I’m slowly getting used to this. It’s like green shoots of science under a big crumbling block of concrete ruins left over from the climate model wars.
WebWires says:
September 12, 2012 at 9:10 pm
“Thank god … now we will never have to worry about deserts ever forming … anywhere, ever and ever.”
If you worry about desertification, you should strive to emit as much CO2 as possible, preferrably from fossil fuels, as a higher CO2 content helps plants to grow in drier areas – they need less stomata to breath and therefore lose less water through evaporation. Parts of the Sahara are already greening. I expect a strong positive feedback as new root systems hold moisture.
This is a more serious problem with the models than even this article indicates. The problem is that the empirical observation does not explain why there is a higher probability of rain over parched soils. It at best offers an heuristic hypothesis.
And here’s the problem. I’m quite certain one can go into existing models and either tweak a few parameters or program in some variant of the heuristic hypothesis and get improved correspondence. Why not? One has many parameters to play with, and if they still won’t do it a new phenomenological model with still more parameters can almost certainly do the trick. This does not address the deeper question: Why didn’t the models get this right in the first place?
The answer there, at least, we know. They have a whole chunk of climate physics dead wrong. Worse, the physics they do have has been tuned to what were clearly biased assumptions that were not founded on measurement at all!
Here’s the way climate modeling should work. You look at the entire climate system, looking at pieces of “microscopic” physics that should bear, trying to at least formulate the actual global Navier-Stokes equation(s) with full radiative coupling, on an unbiased tessellation of the surface of the (presumed, but of course not exactly) sphere, including all of the gross geometry associated with a tipped axis of rotation, precession, and orbital eccentricity and at least trying to parameterize what is known about the sun both as a source of direct heating via light and indirect modulation of the climate system via e.g. magnetic effects and modulation of solar wind (which I admit is not much, not well). Don’t forget to make the surface model three dimensional, don’t forget to make the coupled oceanic model three dimensional too. Don’t assume good mixing of greenhouse gases, aerosols, particulates, either in the atmosphere or in the ocean — allow for arbitrary localized sources and sinks of variable power (both those related to human activity and those that aren’t). Don’t forget to include oceanic salinity, at least, and you may even need a biofeedback loop there (as you certainly will on the land surface!) to account for e.g. variations in absorption and emission and vaporization in ocean water (and the larger lakes) due to dynamical phenomena such as algae bloom or river-borne silts or storm action that can affect the visible light transparency of water as well as its albedo. Finally, be sure to make the model non-Markovian — it has to be initialized, and run from, not just today’s data but from data collected over at least a decade into the past, more likely 30 to 100 year, because the actual climate changes on time scales that average over at least multiple solar cycles and multiple global decadal oscillations — that (and observations of past temperatures and climate changes by proxies) suggests that a century is a pretty reasonable minimum — out to a thousand years or more (timescales visible in the ocean, a truly vast and slowly evolving heat sink for the entire Earth).
This won’t be a terribly easy project, because to do a good job on a uniform tessellation of the sphere — one that has a prayer of actually corresponding to the climate — one will have to very likely consider the Earth’s surface cell by cell — and I’m thinking that the granularity of the tessellation will need to be far smaller than the current laughable and non-uniform 5×5 latitude/longitude cell decomposition. One will have to do the moral equivalent of what Google Earth does (and might even need to use GE as a source) and look down on each and every tessera/cell, consider both its height and structure and the land use and whether or not it is all or partly water and whether or not humans live there and of course its absolute location on a rotating tilted Earth as it wobbles around a remarkably variable star as it makes its way through an almost totally uncharted galactic interstellar medium filled with more or less invisible “stuff” at unknown but variable density, stuff that is sufficiently dense that it can and does accrue to where it collapses into big balls of matter a million kilometers in radius that are hot enough in the center to ignite fusion.
Be lavish! Include every bit of physics that MIGHT have a bearing! Tidal heating? Estimate it, locally (cell by cell), set from the actual physics of the tide, the actual orbital state of Earth, Moon and Sun, and at least a simple model of plastic deformation of the solid crust and a rotating waveform caused by the deformation of the oceans. Don’t forget the fact that the tide also causes a similar breathing oscillation of the atmosphere — maybe it is insignificant, maybe not. Let the model (in the end) decide. Anisotropic heat bleed from the Earth’s interior at places where the crust is relatively thin? Have a parameter that can describe it and do your damnedest to set that parameter (and all parameters!) from data, not to make the model fit past observations better. A forest fire? A meteor? A nuclear exchange? The model should be able to cope with all of these, in real time, in its inputs. Small perturbations can grown. And don’t forget volcanoes, both known active and dormant and ones that could start up tomorrow out in the middle of a farm plain in Mexico!
If the parameters, set accurately from local observations, fail to reproduce the past better, you have precisely two options. If you think the parameters are incorrect, go measure them better. This is entirely fair; many of the parameters in your model will not be known at the granularity of your tessellation. For example, the local rate of heat escape through the crust can only be measured by drilling a rather deep borehole, waiting for the heat produced by the drilling itself to go away, and then measuring the temperature at a series of points of different depth. The thermal gradient plus a measured knowledge of the thermal conductivity of the borehole core as a function of depth permits one to determine the rate of heat flow. There are only a finite, rather small, number of cores at this time, order of 20K to cover the globe, with the deep oceans entirely neglected and with nothing like a uniform coverage, so one perforce has to interpolate the grid, making assumptions of some sort of continuity or uniformity that may or may not be even close to correct. By all means, drill more holes, but do not tweak the parameter “by hand” or recompute the interpolent model in an optimization process to improve hindcast accuracy! That’s a no-no.
The other option you have is to make the model better in one of precisely two ways. You can always add more physics or correct the physics you’ve got. Butterfly wing flaps important? Add a parameter for the observed density of butterflies per tessera, seasonally adjusted. Forget to add a parameter that describes the local utilization of nitrate fertilizer in the Mississippi River basin, which in turn affects algae bloom in the Gulf of Mexico, which in turn modulates its emissivity and albedo and the rate at which it absorbs or emits CO_2? Have at it, as long as that parameter is set from measurements made at a sufficient cellular granularity that it has a prayer of actually being statistically sound, tessera by tessera. A sign error in a coupled term? Fix it.
Or, you can change the granularity of the model. Initially, this may well involve stepping back to larger tessera, as smaller ones (with the resulting finer grained coverage) may either exceed your ability to compute or result in a poor averaging of badly sampled parameters, sort of like taking a handful of thermometric measurements made densely in one tiny part of Antarctica and then using assumptions that smear them over the entire, largely unsampled continent. Differential/adaptive tessellation may even be required and is certainly permitted, using larger cells in relatively homogeneous parts of the surface or places where we have relatively few measurements and a lot of uncertainty and smaller ones where we can do better.
Adjusting the input data — correcting for measurement biases and so on — is permitted, but it is a place where angels fear to tread because it is an open invitation to once again permit your biases to influence the model, which is of course against the rules of good science. An example of a good use of adjustment — when I examine the weather station output of the half dozen or so personal stations sited within five or six miles of my house, the one that is closest to me, sadly, invariably reads a full 2-4C warmer than all of the rest! It is literally a half-mile from my house and yet nearly useless. I am almost certain that the station is, for example, sitting right above or slightly downwind of the family’s air conditioner, or next to their south-facing driveway, or lacks a proper white box with sufficient ventilation, or perhaps its thermometer is just plain broken, but I am completely certain that it is broken somehow because I can look at my own outdoor thermometer and observe the discrepancy.
An acceptable way to fix this would be to apply a statistical test that eliminates sufficiently deviant outliers, under the hypothesis that the local surface air temperature is unlikely to actually be 2-4 C warmer, consistently, anywhere but in a local hot spot caused by poor siting of the thermometer in question. A better way, of course, would be to visit the owners of this weather station and ask them to check its siting and perhaps suggest that it would work better if it weren’t mounted on their sunnyside rear asphalted porch or whatever, or to point out that the covering box is supposed to be white, not dark brown (even if the latter matches their house), or to note that setting the box on the ground so that there is no airflow means that they are creating a mini-“greenhouse” of restricted convection and trapped air that actively traps even the heat generated by the station’s electronics.
Once the full model was built and set up at some fine granularity, only then could one face the reality of the difficulty of the actual computation and decide how to proceed. Need to approximate, coarse grain, neglect things that you think might be negligible? Have at it. Just remember, you can’t adjust the parameters when you do this, and your path from the parameterization on the fine scale to the coarser scale must be clearly documented and justified by objective statistical analysis and above all, included in the final error estimate! A perfectly acceptable outcome will be computations with error bars much larger than observed climate variations, meaning that the model (as best you can compute it so far is meaningless. Your only option then is to refine the model, improve the empirical base of the parameters (make more, better measurements, not “adjust” the parameters you already have and pretend that this lowers your error estimate or makes the result more meaningful) and try to solve it at a finer tessellation and see if it helps.
Another perfectly acceptable result will be a model that utterly fails to predict gross observations of the way the climate works — such as getting the sign wrong in the dependency between soil dryness and summer rainfall. Sadly, in a thoroughly coupled Navier-Stokes problem, an error like this simply means that your entire result is meaningless — you have fundamental physics wrong, you have a completely inadequate (and/or geometrically stupid) tessellation, you lack the computational power to solve the problem correctly, you are averaging wrong, or — dare I suggest — the model has decent physics in it somewhere but the parameters were set to optimize agreement with both incorrect past measurements and a set of beliefs about what missing or unmade measurements “ought” to be in some entirely subjective heuristic in the minds of the modelers.
This is the sad, sad problem with modeling — and I say this as an experienced, professional modeler that has run massive large scale computational models in physics for over twenty years, and as CTO of a company devoted to advanced bleeding-edge predictive models. It is enormously difficult to separate the biases and beliefs of the modeler from the construction of the model. It requires a considerable amount of discipline to ensure that the model has only the right physics (or other prior knowledge) in it and that it utilizes randomness (via e.g. Monte Carlo) in key ways to erase possible parametric or conditional bias. It is so very easy with a sufficiently complex parametric model to get good correlation between model predictions and any given past data set and still end up with a model that sucks when it comes to predicting the future. There are ways to avoid doing this — ways that any experienced modeler knows better than to evade — but in the end the risk remains, and should properly be expressed as future time error bars that grow, quite possibly grown exponentially, initially, until they saturate out at some sort of common sense threshold determined by the past observed variability of the system in question.
Predictions of “catastrophe” — egregious deviation outside of those empirical limits — should be viewed with the greatest of suspicion, especially in the extrapolated solution to an enormously nonlinear and complex set of coupled Navier-Stokes equations with multiple non-Markovian feedbacks over comparatively huge timescales, arguably the most difficult problem the human race has attempted to solve so far. Climate modeling is a hard problem, and its difficulty is reflected in poor predictive power or reliability in its predictions, so far. The egregious failure of Hansen’s many “catastrophic” predictions in the past stand as a perfect example of this.
Over time, of course, the models will indeed improve. Compute power that can be applied to the problem continues to increase. The physics — and empirical basis of the parameters — is gradually being improved. Observations like those in the top article — if they are independently verified and subsequently found to hold and not be a “local” feature of the timescale of observation that inverts to a completely different behavior in five years, then inverts once again to this behavior in six more — provide useful tests of the quality and reliability of the model. Climate modelers need to be a lot more willing to say we don’t know what the climate is going to do in ten years, not even approximately, certainly not with “60% confidence” or “90% confidence”, not when the models can’t even get the sign of an important feedback right unless/until it is more or less entered by hand or the models parametrically tweaked to “force” agreement.
After all, it is only the truth!
rgb
So, how long before my Las Vegas area property becomes lush farmland? [/sarc]
****
stefanthedenier says:
September 12, 2012 at 6:39 pm
This is a 24carat crap: ”moist soils lead to even more rain, whereas dry regions tend to remain dry.” But observations suggest otherwise: “We have analyzed data from different satellites measuring soil moisture and precipitation all over the globe,” OTHERWISE?!?!
No, you have NOT!!! Where is topsoil moisture, water storages and ”water vapor” in the air -> attracts rain-clouds from the sea like a magnet . Where is dry – clouds go around the land as cars around a traffic island!!! Australia is a perfect example, so is Sahara, so is Brazil.
****
I agree w/you. Something like 50% of Amazon rainfall comes from water evaporated from the surrounding forest, not the ocean. If I watch summer convective rainfall in the US over time, areas that are moist continue getting subsequent rain compared to drought areas. Why? Because the moist areas evaporate & water vapor is lighter & encourages uplift (and eventually rain). Nearby (regional-wise) dry areas provide the “downdraft” to supply air for the updraft areas. Wet areas keep getting rain at the expense of dry areas.
Of course, synoptic conditions can sometimes overcome this tendency (fortunately). And winter frontal-type precip seems to be much less influenced by this effect as it is not so dependent on local ground conditions.
OT, but today I saw our own Leif Svalgaard on a National Geographic Channel program called “Solar Force” (concerning solar effects on climate). Congrats & about time!
Maybe, Just maybe. Average is just that? Drier areas are apt to get moisture. Wetter areas not so much.
Complete nonsense. Yes, dry soil might heat up more than wet soil because the heat goes directly to warming the surface as opposed to evaporation. But look at the effects on dew point. The dew point is much lower over drier soils in the same air mass than over wetter soils. It’s not the heat, it’s the humidity!
john robertson says:
“Does anyone know of a naturally occurring positive feedback mechanism?”
They are quite common. Avalanches. Nuclear chain reactions. Ice albedo feedback.
Most of what rgb suggests is already common practice. In particular GCM resolution of 1 degree is considered coarse nowadays and is only used for first passes or very long paleoclimate runs. Here is a visualization of a run at 25 km. http://www.youtube.com/watch?v=YlT6mJzNqEQ
Jay Davis on September 12, 2012 at 5:49 pm
I love the “models wrong” in the title.
————-
I know you love it but it’s not really down to models. The models simply incorporate what science is known at the time. So the real issue here is whether the basic science behind soil evaporation and rain is correct. I am betting there is a whole lot of research already done into the issue. This new paper appears to contradict that (hypothetical) research.
No doubt there will be some to and fro about whether this new paper is correct and under what conditions it is correct. It’s quite possible on a local scale in some geographies it’s is not correct and this may have led to faulty global conclusions.
john robertson says:
“Does anyone know of a naturally occurring positive feedback mechanism?”
Ice sheet elevation increases during cold periods causing increased snow/ice accumulation, and the reverse during warm periods.
Last month, global water vapour levels were 0.13 mms/m2 or 0.13 kg/m2 (0.6%) above average while the climate model runs included in IPCC AR5 were forecasting that it would be 6.1% above average by this time.
The ENSO, as the biggest weather phenomenon on the planet, actually has more influence on global water vapour levels than anything else and it has NO trend over the long-term. Thus, there is no real trend in water vapour levels either.
More than half of the expected warming comes from the expected increased water vapour levels (which is not happening unless one cherrypicks a time period of low levels contrasted with a period of high levels which the IPCC/climate scientists have consistently done in the past).
http://s13.postimage.org/z4ux4t5pz/Water_Vapour_IPCC_AR5_Forecast_ENSO_Aug_2012.png
How to fix the models?
See the paper at: Afternoon rain more likely over drier soils Christopher M. Taylor, Richard A. M. de Jeu, Françoise Guichard, Phil P. Harris & Wouter A. Dorigo, Nature Letter DOI: doi:10.1038/nature11377 online 12 September 2012
David Stockwell (2008) showed that the CSIRO’s drought predictions were contrary to the historical evidence with a hindcast/forecast using half the data for tuning and the other for prediction. Tests of Regional Climate Model Validity in the Drought Exceptional Circumstances Report.
rgbatduke above observes:
The models might have been corrected earlier had researchers incorporated WJR Alexander’s findings that precipitation and runoff vary strongly with the 21 year Hale cycle but NOT with surface evaporation.
Linkages between solar activity, climate predictability and water resource development* J. So. African Inst. Civil Engineering Vol 49 No 2, June 2007, Pages 32–44, Paper 659
Now the challenge of applying the scientific method, discovering the missing physics, and correcting the models to match the evidence – rather than political alarmism.
Bill Illis September 13, 2012 at 5:44 pm
Bill, 1] water vapor has nothing to do with the phony GLOBAL warming. 2] more water vapor = milder climate + more snow on the polar caps. 3] more water vapor = better climate; but NOTHING to do with the imaginary GLOBAL warmings.
4] most of the Faulty Line is on the bottom of the oceans – where is tectonic movements – submarine volcanoes / hot vents get more active -> currents spread 100% of that heat = get La Nina or El Nino; depends where is more activity – that increase evaporation. You people are falling for their trick that water vapor increases the OVERALL global temp = they are feeding you with more bullshine. If you keep using ENSO’s lies, to prove them wrong – they just invent more lies.
Get some real proofs about water vapor: http://globalwarmingdenier.wordpress.com/2012/07/20/water-vapor/
Dorigo says: “The air over dry soils heats up more easily.”
BUT, Dorigo obviously doesn’t know that: ”The air over dry soils COOLS faster, at night. It should start getting too obvious to you people that: as long as there are ”bullshine addicts – there will be plenty Bullshine Merchants. From IPCC to CSIRO – they have to feed their bulldung consumers; bon-apetit
If somebody believes that: they know overall Water vapor in the atmosphere, with precision at given times; shouldn’t blame the climatologist; but the person in his mirror.
Increased / decreased evaporation in the oceans depends on the activity of submarine volcanoes / hot vents – if they can ”predict’ where AND when is that activity going to increase – they would have became the BEST seismologists = they wouldn’t had to confuse big / small regular climatic changes with any phony GLOBAL warmings; to fleece the Urban Sheep
Dorigo says: “The air over dry soils heats up more easily.”
That’s because the air over drier soil is less humid. Less humid air has a lower thermal capacity, but is more dense than more humid air. The net effect is more humid air over wet soil rises, even though it is cooler than warmer drier air over dry soils.
@Rich Lambert says:
September 12, 2012 at 7:04 pm
I live in Oklahoma where it has been dry the past two summers. The sun quickly heats the ground, but I’ve failed to see those thunderstorms and rain. Recently, I was in eastern North Carolina which is quite humid this year and it rained about every other day. I guess the model applies somewhere else.
Another Oklahoman here. I agree.
From mtobis (@mtobis) on September 13, 2012 at 1:45 pm:
Most people kinow of feedback as what happens when a microphone that’s the input to an amplifier gets too close to the speaker that’s the output, when the gain is too high. That’s positive feedback, a small signal gets amplified into a much larger one when the output feeds back into the amplifier.
An avalanche is a failure of supporting structure, normally thought of when snow at the bottom can no longer support the weight of the snow on top of it. Other “avalanches” are landslides and mudslides. They are a catastrophic failure.
Naturally occurring nuclear chain reactions are far from quite common. There is only one known location in the world, which occurred about 1.7 billion years ago. They occurred there because of rich uranium deposits and groundwater.
Inherent in the concept of positive feedback is if you remove the input then the effect will stop. If you reduce the input then the effect will lessen.
Avalanches are catastrophic one-time events. Nuclear chain reactions are self-sustaining. Both are triggered, “turned on”, and then run their course with no further input required.
If you think the ice albedo effect is a similar positive feedback, you clearly have no idea what really is a positive feedback, at all.