« Workshop Discussion Threads
Questions for Tom Knutson Notify me of new posts »
1: Karin Kirk 10:39 AM Oct 20 2008 973:2833
Please post your questions and comments for Tom Knutson's presentation here.
2: Catherine Gautier 01:58 PM Oct 20 2008 973:2852
Hi Tom, Great talk. I tried ot ask questions during the talk but they didn't seem to get to you. Anyway...
My main issue has to do with the higher intensity hurricanes and whether you have any idea what causes your model, even at the highest resolution, to underestimate those Cat 4 and 5 hurricanes which seem like those that could increase in a warmer world based on limited observations (e.g., Webster et al and also ELsner et al.). It looks like your more recent results improved a little with the cat 4 hurricanes. What gave this improvement (resolution?) and what direction are you looking into to improve your results further?
My main issue has to do with the higher intensity hurricanes and whether you have any idea what causes your model, even at the highest resolution, to underestimate those Cat 4 and 5 hurricanes which seem like those that could increase in a warmer world based on limited observations (e.g., Webster et al and also ELsner et al.). It looks like your more recent results improved a little with the cat 4 hurricanes. What gave this improvement (resolution?) and what direction are you looking into to improve your results further?
3: Robert Kuhlman 02:08 PM Oct 20 2008 973:2853
Thanks for your great presentation, Tom! Quick question based upon the 'threshold temperature' concept introduced in slide 27, I believe:
Why is it that the threshold temperature rises with an overall change in SST? Is it the *contrast* in SST of the tropics vis-a-vis mid-latitudes which is important to hurricane formation? In other words, as the mid-latitude SST increases, the tropic SST has to get that much warmer? Or is it not quite that simple?
Why is it that the threshold temperature rises with an overall change in SST? Is it the *contrast* in SST of the tropics vis-a-vis mid-latitudes which is important to hurricane formation? In other words, as the mid-latitude SST increases, the tropic SST has to get that much warmer? Or is it not quite that simple?
4: Tom Knutson 02:14 PM Oct 20 2008 973:2854
Hi Catherine. I lost my Yugma connection about half way
through, but I was too busy in the early part with my talk to notice the online questions...Anyway, hopefully I can address everyone's questions here.
Hurricane intensities as simulated by models will depend on both model resolution and some details of the physics in the models. Our 18-km grid regional climate model tops out at around 47 m/s, but we can improve on this a lot by doubling the resolution and tinkering with the surface flux parameterization (e.g., evaporation and momentum fluxes).
The GFDL hurricane prediction model (9 km grid resolution) simulates cat 3's and 4's fairly well in terms of climatological number but is a bit deficient on cat 5's. It's also used operationally for hurricane prediction, so we get to learn more about its biases based on operational experience. This work with the GFDL hurricane model that I showed has not yet been published. We were keen to move to a model that could simulate cat 4-5 storms as we saw a tendency for a different response to climate warming at the high end vs the run-of-the-mill tropical storm or hurricane.
Further improvements will involve illustrating how dependent some of our downscaling results are on the particular climate model we choose for the downscaling. That is a measure of uncertainty about the whole problem, and one that the whole climate change community will have to help in solving.
through, but I was too busy in the early part with my talk to notice the online questions...Anyway, hopefully I can address everyone's questions here.
Hurricane intensities as simulated by models will depend on both model resolution and some details of the physics in the models. Our 18-km grid regional climate model tops out at around 47 m/s, but we can improve on this a lot by doubling the resolution and tinkering with the surface flux parameterization (e.g., evaporation and momentum fluxes).
The GFDL hurricane prediction model (9 km grid resolution) simulates cat 3's and 4's fairly well in terms of climatological number but is a bit deficient on cat 5's. It's also used operationally for hurricane prediction, so we get to learn more about its biases based on operational experience. This work with the GFDL hurricane model that I showed has not yet been published. We were keen to move to a model that could simulate cat 4-5 storms as we saw a tendency for a different response to climate warming at the high end vs the run-of-the-mill tropical storm or hurricane.
Further improvements will involve illustrating how dependent some of our downscaling results are on the particular climate model we choose for the downscaling. That is a measure of uncertainty about the whole problem, and one that the whole climate change community will have to help in solving.
5: Tom Knutson 02:21 PM Oct 20 2008 973:2856
Robert,
My guess is that the threshold temperature is determined based on the sea surface temperature in the tropical locations relative to the rest of the tropics. Whether the tropical vs extratropical contrast could also affect things is an interesting question that we haven't really looked into. We think that the tropics is special in that you cannot have large contrasts of temperature in the upper troposphere, even though you can have such contrasts at the surface. So the place where tropical storms will be most active will be where the surface is warmest, because that will have the greatest contrast exists (in the vertical) between the surface and the (relatively uniform) upper tropospheric temperature.
My guess is that the threshold temperature is determined based on the sea surface temperature in the tropical locations relative to the rest of the tropics. Whether the tropical vs extratropical contrast could also affect things is an interesting question that we haven't really looked into. We think that the tropics is special in that you cannot have large contrasts of temperature in the upper troposphere, even though you can have such contrasts at the surface. So the place where tropical storms will be most active will be where the surface is warmest, because that will have the greatest contrast exists (in the vertical) between the surface and the (relatively uniform) upper tropospheric temperature.
6: E. Christa Farmer 02:21 PM Oct 21 2008 973:2885
Re: threshold temperature. I learned that SST never gets above 28-30degC, for various thermodynamic reasons having to do with wind speed, evaporation, etc... true? If so, is there some kind of asymptotic relationship between SST (or difference in SST between E & W equatorial regions?) and "threshold temperature" for hurricane formation?
7: Tom Knutson 02:54 PM Oct 21 2008 973:2889
Re: threshold temperature. I'm no expert on this, but according to http://www.infoplease.com/ipa/A0001375.html the highest recorded SST was 36deg C (96F) in the Persian Gulf (Aug. 5, 1924). Certainly we obtain large scale SSTs well in excess of 30 deg C in climate change projections from climate models. For example, see Fig. 1 of this paper for a look at SSTs from a 4xCO2 experiment. [http://www.gfdl.noaa.gov/reference/bibliography/1995/tk9501.pdf]. Note the large area of SST in the Western Pacific Warm Pool simulated at over 33 deg C under 4xCO2 conditions. There is no apparent "thermostat" operating here...only slightly less warming occurs in the warm pool than in the eastern tropical Pacific for the reasons we discuss in that paper.
8: E. Christa Farmer 02:29 PM Oct 22 2008 973:2919
OK, so not 30degC, but maybe never above 36degC? Isn't there still something keeping the water from getting any warmer than that? When actual temperature gets close to that max value, do you think the threshold temperature for hurricane formation would reach a max as well?
Join the Discussion
In order to post to this discussion you need to be logged in.
« Workshop Discussion Threads 
