Some of the Earth's tiniest naturally occurring particles may play a much different role in global climate change than previously thought.  New findings reveal that models scientists have long used to estimate the causes and effects of global warming may be flawed due to errors in one of their significant inputs -- atmospheric dust.

This revelation is an indirect conclusion of a new study published in the journal Proceedings of the National Academy of Sciences by Jasper Kok, a climatology researcher with The National Center for Atmospheric Research (NCAR).

I. Ice Melt, Cooling, and Large Dust Particles

The study's key conclusion was to show that the ratio of small soil dust particles (clay), which cool the atmosphere, to large soil dust particles (silt), which may heat or cool depending on their size, may be much lower than previous estimated using empirical (observation based models).  While the number of small dust particles, according to our discussion with Professor Kok, is approximately the same as previously thought, the number of large particles are much higher.

This is a critical finding because it means that current global circulation models (GCMs) used in broader climate change models, may significantly overestimate the net cooling effect per volume of dust (given the greater occurrence of large silt, which favors heating).  The overestimation of the cooling is reported especially high at the top of the atmosphere (TOA), where it may be overestimated by as much as a factor of 15.

It is also critical as these large silt particles tend to precipitate and can play an important role of accelerating glacial melting.  Thus atmospheric dust may not only warm the earth more than previously estimated, but may also lead directly to one of the observed effects often associated with climate change -- glacial melting.

II. A Dusty Surprise

To better understand the study, it's important to get a picture of what these particles are. 

Mineral dust originates when sand particles are blown by the wind into soil, shattering it into microparticles.  Major sources of mineral dust include the southwestern United States, northern Africa, Northeast Asia (the Gobi Desert) and Australia.

The sand essentially acts like tiny bullets, hammering dirt particles, which tend to be 20 microns or smaller and are thus known as PM20 sediment.  This process, known as saltation ejects a mixed distribution of dust microparticles into the atmosphere, as the cracked particles' aerodynamic forces overwhelm their cohesive forces. 

The smallest resulting microparticles measure less than 2 microns (2000 nm).  These particles are known as clay.  Past studies have shown that they tend to stay in the atmosphere for long periods of time, reflecting light and cooling the Earth.  

Large microparticles, known as silt, can reach 50 microns (50,000 nm) -- about the width of a human hair).  Their weight causes them to quickly fall out of the atmosphere.  Depending on their size they can either exert a net cooling effect or a net heating effect.  Silt larger than 5 microns in diameter tends to warm the earth (as reflection of outgoing radiation back towards the Earth overcomes sun-blocking effects), while silt between 2 and 5 microns tends to cool (as the sun blocking effects are more effective than the reflective effects).  In that regard small silt acts much like clay, albeit being less effective at cooling.

Despite their quick precipitation all types of silt have another significant direct effect on the climate.  After falling through the atmosphere, they tend to accumulate in mountain polar ice, concentrating sunlight, absorbing heat and accelerating melting.

In order to find the true ratio of the particles types, Professor Kok cleverly combined mathematical theory and statistical data.  To determine the breaking method, he used brittle object breaking formulas developed by mathematicians.  Brittle objects, like glass, rocks -- or soil -- break into a predictable distribution of small, medium, and large particles.  

Using these formulas, the researcher turned to statistical information on arid soil, published in a 1983 study Guillaume d'Almeida and Lothar Schüth from the Institute for Meteorology at the University of Mainz in Germany.  By combining the two, he was able to arrive at what is thought to be the most accurate statistical distribution for particle sizes resulting from soil breaking published to date.

And the results yielded a major surprise.  They showed that the ratio of larger particles to small particles (responsible for heating) was two to eight times higher than originally thought.  In other words, this additional large dust may be both exerting more of a net warming effect on the atmosphere and more of a melting effect on glacial ice.

The difference in ratio is large enough that previous models may have even missed the sign of the net forcing from dust (there may have been a net warming due to global dust, rather than a net cooling).  At the least, the dust's per unit cooling effect is much smaller than previously though.

This raises puzzling questions about historic cooling, which traditionally is accompanied by an increase in global dust.  According to Professor Kok, "We know from paleo-records, such as ice cores and lake/ocean sediment cores, that the atmospheric dust concentration during, for example, ice ages, was several times higher than it is now."

The author believes that this is because current GCMs and climate change models significantly underestimate the current volumes of atmospheric dust circulation, and thus come close to the correct net forcing, despite overstating the dust's per-unit net cooling effect.  If true, this could have important ramifications in the biology field, as the Earth's nutrient circulation model would change.

Another possibility not fully analyzed in the paper is that the cooling effect could instead be lesser, but the warming effect from other inputs e.g. carbon aerosols could be less than previously thought.  This would be very significant if true, as it could have a bearing concerning to what degree current emissions need to be regulated.

A crucial question not discussed by the author, but that also seems apparent, is whether warming can trigger an increase in dust, which in turn reins in the system and results in a cooling.  That is a critical question to assessing the possible impact of aerosol emissions, as it could indicate that the planet as the ability to counteract aerosol emissions, via an increase in dust, at least to an extent.

Again, Professor Kok emphasizes that how atmospheric dust levels change over time is still poorly understood.  But based on our discussion with him, we received the impression that this is absolutely critical to determining how our climate will change in response to a variety of factors, including carbon dioxide emissions.

III. Back to the Drawing Board

The new findings by no means devalue the idea of using computer modeling to study the Earth's climate.  But they are an important reminder that climate models are only as good as their inputs, and in many cases those inputs are based on information that's lacking.

It would be rash not to reevaluate at least some of the modeling work done to date, taking into account this new perspective on atmospheric dust.

How much or how little impact this study has on modeling will rest largely on determining more accurately how atmospheric dust levels vary with time.  According to a brief conversation we had with Professor Kok, the levels of atmospheric dust and how they change with time is a poorly "understood" topic.  Thus the study could potentially invalidate currently collected data from current warming models, or leave it largely unscathed.  But it does indicate that a new round of data collection and verification is necessary to determine which is the case.

Despite the controversy that peripheral effect of the study will bring in the non-scientific community, which is preoccuppied with the kind of instant, absolute answers that can justify billion dollar policy decisions, most objective members of the climate change community will likely take the need for correction in stride. 

The author states:

Climate scientists are acutely aware of the many uncertainties in climate models. They are an imperfect tool to estimate future climate, but they are probably the best tool we have. However, despite the many uncertainties, climate scientists are still able to say that, over long time periods (decades) over which greenhouse warming will overwhelm natural climate variability, the Earth will very likely warm. Exactly how much this warming will be, and how this varies regionally, is still uncertain... For this reason, much of the research in the climate sciences is directed towards reducing uncertainties in climate models by both better understanding processes that are already in the models (such as what my study has done), and by adding additional processes that had not yet been accounted for (the glacier melting by dust deposition has for example been added to some climate models in the last few years). My study is thus a small part of this much bigger effort.

And moreover, Professor Kok thinks that its important not to forget the results' beauty from a mathematical perspective, in the rush to apply them to improving modeling.  He states in the previous NSF press release, "As small as [the particles] are, conglomerates of dust particles in soils behave the same way on impact as a glass dropped on a kitchen floor.  Knowing this pattern can help us put together a clearer picture of what our future climate will look like.  The idea that all these objects shatter in the same way is a beautiful thing, actually.  It's nature's way of creating order in chaos."

Beautiful indeed.  Professor Kok's ability to focus on truly objective mathematical and scientific analysis with regard to what is an increasing politicized topic is exemplary as well.  Given recent revelations [1] [2] [3], this kind of objective dedication to scientific truth is not always present in this field, so it's as refreshing to see that as it is fascinating to observe the symmetry that underlies many laws of our universe.

Update: Dec 31, 2010 11:10 a.m.-

We had a lengthy discussion with Professor Kok to clarify a couple of conclusions that the NSF press release seemed to be pointing at.  Most importantly, Professor Kok explained to us that the levels of small dust was not less than previously expected, rather they were the same and the levels of large dust were higher.

He also explained that solar variance (generally within the range of +/- 0.1 percent of total energy transfered into the atmosphere) and dust's reflectivity (about +/- 0.3 percent of total energy transfered into the atmosphere) indicate that any correlation between dust and the impact of solar activity would be very weak (around 3 ppm).  At 3000 ppm, dust's direct effects via atmospheric dust levels are the more important topic to look at, as confirmed by our further discussions with Professor Kok.

That said the key conclusion of this article -- that warming models need to be re-run with accurate dust info or risk offering misleading conclusions -- still stands, albeit via a different mechanism than we previously thought.  We apologize for the confusion concerning the relationship between solar activity and dust levels.

We'd also like to thank Professor Kok for taking the time to discuss some of this study's finer points.  We're looking forward to doing an interview with him on the topic of climate modeling in the near future.