Climate Sensitivity is 3 Degrees is a post on Dr. James Annan's blog about the issue of Climate Sensitivity. It is a good summary of the situation, but I would like to add my uniformed two cents worth.
The dry air sensitivity to a doubling of CO2 is about 1 degrees C. I give a range of 1.0 to 1.6 degrees C to include what "I" believe is the originator of the greenhouse theory's, Arrhenius' estimate of minimum climate sensitivity. The uncertainty in specifying climate sensitivity is the response of water vapor which is "the" predominate greenhouse gas.
The term "greenhouse effect" is not that good a description of how CO2 variations in the atmosphere works. Heat flows by conduction, convection and radiation. A real greenhouse impacts all three methods of heat transfer and Arrhenius' "Greenhouse Effect" deals only with radiative heat transfer. Convection and to a lesser extent, conduction are still a part of the heat transfer in the atmosphere up to a layer called the tropopause in the atmosphere. The layer we live in, the troposphere, has a lot of convection, as in rising air masses, falling air masses, precipitation and generally air moving around. There is also variations in solar radiation converted into heat caused by reflection by snow, clouds and dust in the air. The greenhouse effect applies only to outgoing heat radiation called long wave radiation or Infrared radiation. Since things change in the atmosphere, climate sensitivity changes somewhat and the estimates of climate sensitivity would of course be somewhat difficult to nail down. The more we learn the better the estimates will become. Still these estimates will never "guarantee" that climate sensitivity remains exactly in the range we estimate.
Models of the Earth's climate are tools to help understand what is going on, which can help predict what will happen. No model is totally accurate, so the results of several models and various model runs are used determine best guesses. Using different statistical approaches, these best guesses can provide a likelihood range of climate sensitivity that is narrower than indicated by just averaging out the guesses.
Model output is dependent on model input, so the more that is understood about the interactions of the atmosphere, the better the model results should be. Simplicity is a wonderful thing, so keeping it simple in the models is a desired thing. One person's simple is not the same as another person's simple which is where I get into a little trouble with the guys that make the models.
Natural variability is one thing the models do not do a good job of using. Some natural variability is fairly well understood and can be calculated or measured. Other kinds of natural variability is poorly understood and the main source of the uncertainty. Water vapor is the main pain in the butt when it comes to fine tuning the models.
Since the dry air 2xCO2 sensitivity is pretty much agreed upon, the poles of the planet should show a more predictable response to 2xCO2. In a perfect world, warming at both the Arctic and Antarctic caused by CO2 would be somewhat equal or at least discernible in the instrumental temperature records. For a variety of reasons it is not. Before you stop reading, this doesn't prove or disprove anything other than our temperature records at the poles are less than perfect. To improve our understanding of the temperature average at the Antarctic, a recent paper fine tuned some statistical methods to reconstruct a temperature record for the past fifty years for the Antarctic. This discussion of the paper and the paper it challenged is here. The results is that the Antarctic continent maybe warming or may not.
There is definitely more warming at the western Antarctic peninsular than anywhere else which is not a good indication that the warming is due to increased CO2. It does indicate that natural variability, or at least unknown variability, is more a factor than CO2 increase.
In the Arctic, there is a lot of warming. A recent study using ocean sediment cores determined "unprecedented warming" due to Atlantic waters of 3.5 degrees C.
These two extreme examples illustrate our lack of understanding of what is happening in our biosphere. Natural oscillations in our climate patterns complicate our analysis of the impact of 2xCO2.
Natural oscillations are kinda sorta different than natural climate variability. These oscillation move heat energy around where that change in energy can impact global temperature through other processes. The oscillations do not add or remove heat. A cold front, for example, moves air with less heat content to an area where the air has more heat content. When that colder air meets warmer air you normally get some kind of precipitation which does change the heat content of the air. Without the precipitation, air is just being moved around. Now that is not a perfect analogy, there is heat transfer without precipitation, there is just a heck of a lot more with precipitation due to the latent heat change in the air. Sensible heat change, what we can feel, is about 1/4 of the total heat change when moisture is removed from the air. This year there is a whole lot of precipitation going on as the floods in Australia and Brazil indicate, plus record snowfalls in the Northeastern US and Britain.
Precipitation accounts for a very large amount of heat transfer. The average Atlantic hurricane transfers about 600 trillion Watts per day from the troposphere to the low stratosphere. I lost the link to that estimate by NASA, but using 2260 Kilojoules per Kilogram of water, an average storm diameter of roughly 1250 kilometers and an average rainfall per day over that area of roughly 1 centimeter, you should get in that ballpark. Then there is other heat transfer like mixing of the ocean thanks to the waves which involves less energy but has an impact on sea surface temperatures. While 600 trillion watts per day is a big number, it is small in comparison to daily solar irradiation that is on the order of Zeta Watts (10^21). Then each year approximately 505,000 cubic kilometers of rain falls per year around the globe. There is 1x10^12 liters(kilograms of water)per cubic kilometer which yields 5.05x10^17 kilograms of rain at 2260 joules per kilogram yields 11.4x10^20 joules or about 1.14 zetajoules. Phew! These are getting to be big numbers!
One zetawatt is nothing compared to approximately 5500 zetawatts per year of solar irradiation at the top of the atmosphere(TOA). Rain though is not at the TOA, it is part of the turbulent climate activity at the bottom of the atmosphere where we live. So to get some kinda perspective the area of the Earth's surface is about 5.1x10^8 Kilometers squared which is 5.1x10^14 meters squared. Combining the two, precipitation accounts for about 2.2x10^6 Joules per meter squared per year of heat transfer, which is about 0.1 watt-hr/meter squared.
In addition to heat loss due to precipitation there is also energy reflected by the clouds associated with the weather systems that produce the rain. I haven't found a good rule of thumb for cloud cover per watt of heat loss due to precipitation. So I will give it a shot. Back to the average hurricane estimate we have roughly 1250 kilometer diameter of cloud cover for 600 trillion watts per day. The area of the storm is about 1.2 million kilometers squared which is 1.2 x 10^12 meters squared. That should give me 2 meters per watt-day or 0.08 meters per watt hour. Let's just round that off to 0.1 since it is my guesstimation. Since the tops of clouds are nice and bright white they should reflect nearly all the short wave radiation. The average incident radiation for the Earth, per the World Meteorological Organization for sunshine duration, is 120 watts per meter squared. For diurnal illumination I will assume 60 watts per square meter. That would yield about 6 watts per square meter reflected radiation. So reflected radiation due to cloud cover should be a good order of magnitude greater than the heat loss due to latent cooling of precipitation.
This rough rule of thumb is pretty crude, but it may be useful in figuring out negative cloud feed back based on precipitation change. Clouds are considered to be a positive feedback since they block outgoing radiation especially in dry environments. Winter clouds do tend to keep temperatures warmer as long as there is no significant precipitation associated with the cloud cover. Storm clouds are a different critter. They tend to be a negative feedback because of the associated convection. (This is a point of dispute, the whole reason I am considering this rule of thumb.)
The next question is how much more negative is the feedback compared to typical positive feedback? There is a lot of ongoing research in this area. The Tropical Warm Pool International Cloud Experiment is one group I know of. The problem is complex enough that I have not seen any definitive results yet.
This is kind of how it goes. The Earth is a gray body that radiates heat into space based on the Stefan-Boltzmann equation, W = e*delta*T^4, where W is watts per meter squared, e is the emissivity, delta is Stefan's constant 5.67x10^8 Joules/(second*meters^2*K^4) and T is absolute temperature in degrees K. So the right side of the equation is in Joules per second per meter squared which is the same as Watts per meter squared. The dimensionless term e for a black body is about 0.924 according to the Wein approximation. The Earth is not a true black body so the average emissivity is approximately 0.68 and typical cloud emissivity is thought to be about 0.5, lot of approximates and abouts aren't there? Here is a paper about one approach to the problem, Spectral Cloud Emissivities...
The problem is that clouds are part of the Earth and will radiate heat to space just like the planet surface. That radiation rate depends upon the same variables only at the top of the clouds the emissivity is approximately 0.924 (it is this emissivity the 2XCO2 will most greatly impact). Below the cloud, the surface radiation is being blocked by the clouds with an emissivity that varies. Since heat can flow due to conduction and convection, active moving clouds mix the heat through the cloud body negating some portion of the apparent lower emissivity. Blankets and insulation work because there is dead or virtually not moving air to reduce heat transfer due to conduction, convection and radiation. Those winter clouds that keep things warmer than you would think are like a blanket. They are just laying there nice a quiet. If he wind kicks up to 20 knots it is more than just wind chill that makes it colder, it is the clouds mixing temperatures which allows heat to escape also.
With a planet, energy in is pretty much equal to energy out at the top of the atmosphere. That makes it easier to get a guess how much the difference is between the local radiation balance of an area with precipitation cloud cover. Rain clouds block or reflect, most of the incoming radiation and about half of the outgoing radiation. So a good first guess is that rain clouds produce about 2 to 3 watts per meter squared of negative feedback. Note that is not all clouds, only clouds with precipitation.
If my rule of thumb happens to be in the ballpark, changes in precipitation would be a factor in global temperature change. Fine tuning the rule of thumb is bit more difficult.
There are a number of scientists and groups studying the nature of cloud impact on global climate. So this feeble attempt of mine has been done more accurately than I have in the past and will be further fine tuned in the future, I am sure. I rarely have truly original ideas, so here is very little likelihood that there is anything earth shattering here. It is a reasonable explanation of the complexity of cloud feedback.
In case you were wondering, the positive feedback of clouds has to be considered as well and it is more difficult track changes over time because of the lack of precipitation. So here is a brief look at positive cloud feedback from my perspective.
For positive feedback, clouds would have a rather low emissivity, ~0.5, and fairly high diffusion. How much light the clouds let through is an indication of how much solar radiation they are letting through. Simply, positive feedback cloud block more outgoing radiation than incoming.
This is more difficult to "see" because the infrared range of light outgoing is not in our visible spectrum. Patchy cirrus clouds let in a lot of visible light, so it would seem they would let out just as much infrared. which is not really the case.
At higher latitudes, where the air is typically drier, these clouds have a positive feedback or at least indicate the possibility of positive feedback. That sounds like I am hedging my bet, but there is a little logic to my reasoning.
Greenhouse gases have a higher impact in areas of lower relative or specific humidity. A desert is a good example. With clear skies, daytime temperatures can be over 100 degrees F and nighttime temperatures drop below 50 degrees F. With cloud cover that range can drop to below 100 F in the day and over 50 degrees F at night. The same cloud conditions in the humid tropics much less dramatic temperature wise because there is not as large a swing in emissivity which can be indicated by the specific humidity.
That may sound confusing, but you have to remember that water vapor is "the" main greenhouse gas. Water vapor is responsible for somewhere between 60% and 80% of the greenhouse effect while trace elements are responsible for between 6% and 15% of the greenhouse effect on a global average. As water vapor decreases, the well mixed trace gases remain about the same, so their percentage impact increases in a dry air environment. So cloud location impacts its positive feedback, high latitude clouds and clouds over desert areas produce more positive feedback than they would over the tropics.
Changes in average cloud cover can be an indicator of 2XCO2 feedback, but the local conditions where the increased cloud cover occurs determines how strong the feedback is. That is about as simple as I can explain it. Local dew point temperatures use to determine relative and specific humidity are probably the most inaccurate temperature measurements recorded. Satellite measurements are more accurate, but coverage and the length of records limit their usefulness.
Precipitation records, while not totally accurate, are long enough and have good enough accuracy to be useful. In addition, rainfall reconstructions tend to be more accurate than temperature reconstructions because rainfall has a greater impact on most proxies than temperature.
The US and Global Precipitation Anomalies to the left are from a US government site which I will include a link to soon. The Data for the US appears to be more complete so it is likely that a comparison to US temperature anomalies would be appropriate for fine tuning the rule of thumb. A quick eyeball indicates a correlation between temperature and precipitation trends. A more intelligent statistical approach will be needed to see if and how much correlation exists.
As I mentioned, the rule of thumb is crude at this point. Comparisons to other cloud/precipitation studies shows a variety of results on the magnitude and sign of cloud feedback.
My conclusion so far is to reconstruct past climate change and to provide model input to predict future climate change, precipitation change by region would appear to be an important factor. A reasonable rule of thumb for energy change associated with precipitation would help fine tune the model skill. Regional based models, where the accuracy of instrumental measurements can be more easily determined, should be better suited to determining a precipitation variation input for general climate models.
2/11/11
Just to add a few thoughts before I forget them. Low latitude precipitation reconstructions agree well with oscillation variations, but mid and high latitude reconstructions are more difficult to attribute to a particular oscillation due to the interaction of various oscillation with each other. Since high latitude temperature variation is much greater than tropical, it would be very easy to jump to wrong conclusions should precipitation lag/lead temperature.
No comments:
Post a Comment