EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

High sensitivity of seasonal tropical precipitation to local sea-surface temperature

Robin Chadwick1,2, Peter Good1, Christopher Holloway3, John Kennedy1, Jason Lowe1,4, Romain Roehrig5, and Stephanie Rushley6
Robin Chadwick et al.
  • 1Met Office Hadley Centre, Exeter, UK (
  • 2Global Systems Institute, University of Exeter. Exeter, UK
  • 3Department of Meteorology, University of Reading, Reading, UK
  • 4Priestley International Centre for Climate, University of Leeds, UK
  • 5CNRM, Université de Toulouse, Météo-France, CNRS, Toulouse, France
  • 6Department of Atmospheric Sciences, University of Washington, Seattle, Washington 98195, USA.

Seasonal mean tropical precipitation at any location is controlled by a tangle of local and remote effects, including influences from SSTs across the globe. This, along with uncertainty in precipitation observations, and extremely limited observations of atmospheric circulation, makes understanding the relevant physics challenging. Climate model precipitation biases persisting across multiple generations of models point towards stubborn gaps in understanding and reduce confidence in seasonal forecasts and climate projections.  This includes the 'double ITCZ problem': excessive rainfall in the southern tropical Pacific, first reported in 1995.  Model ITCZs also tend to be too wide.

Our study shows that in the real world, the sensitivity of tropical precipitation to local sea surface temperature is high, associated with strong shallow circulations.  This rests on a novel analysis of observations, unpicking local and remote controls on precipitation, and navigating a path through observational uncertainty.  Models with appropriate sensitivity to local sea surface temperature, perform well across many conditions.  Improvements in this sensitivity from the fifth to the sixth model intercomparison project are small, highlighting the need for new understanding.  By further linking model biases to shallow convection, our results highlight a target process for focused research: accelerating improvements in seasonal forecasts through to multi-decadal climate projections.

Wider Met Office work linking precipitation evaluation between climate, seasonal and weather timescales will also be summarised.

How to cite: Chadwick, R., Good, P., Holloway, C., Kennedy, J., Lowe, J., Roehrig, R., and Rushley, S.: High sensitivity of seasonal tropical precipitation to local sea-surface temperature, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-7708,, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 01 May 2020
  • CC1: Comment on EGU2020-7708, Julia Hargreaves, 03 May 2020

    I would like to be able to understand your display.  Can you upload a new version with some explanations perhaps?  What are all the lines in the first set of plots? What do the terms in the equation mean? Are you discussing precip over land or ocean? If over land, how nearby is the SST? As you can tell, I'm totes confused at present. Sorry! But thanks for uploading a display! 



    • AC1: Reply to CC1, Robin Chadwick, 03 May 2020

      Hi Julia, thanks for your comment, there are some captions under the slides which provide some minimal explanation (including what the lines mean on the second slide) did you see those? We're doing zoom talks for this session so I just uploaded my talk with some captions, I didn't have time to make a whole separate poster as well as a talk, sorry if that means the slides are not self-explanatory. Let me know if you want the zoom link for tomorrow morning and I can send it to you.


      • CC2: Reply to AC1, Julia Hargreaves, 03 May 2020

        No I didn't see captions. I was planning to attend the chat for this session. I guess they will post the zoom link there and then? At least that explains why the next display down is just a teaser. 

        • AC2: Reply to CC2, Robin Chadwick, 03 May 2020

          Yes I think the plan is to post the zoom link in the chat session tomorrow so people are aware it's going on

  • CC3: Changes in precipitation without changes in temperature (geoengineering / Dietmar's question), Angeline Pendergrass, 05 May 2020

    During the synchronous part of the session yesteday, Dietmar Dommenget asked a you a question, that I wanted to comment on and also get your (Rob's) opinion on. Dietmar asked (correct me if I'm wrong) about the suppression of precip despite constant temperature. 

    I figured that what he was referring to is the solar geoengineering situation. If GHG forcing increases, and some solar radiation in blocked from reaching the earth/troposphere (eg by stratospheric aerosol geoengineering) at a rate where global mean surface temperature is held fixed, then global mean precipitation decreases (see eg Figure 1 of McCusker et al 2012:  ). This happens because of the direct effect/fast response of precipitation to GHG forcing. If you think about global mean precipitation change as having a temperature-dependent part and a direct effect/intercept (a la Allen and Ingram 2002), then the intercept is the only part that matters in this case. We know solar forcings have little direct effect on global mean precip, while the direct effect of GHGs is to suppress it (eg Fläschner et al 2016: or my work: ). We recently commented on this in a paper on geoengineering too (Simpson et al 2019:

    But that's pretty much all global mean. So my question for Rob is, do you have any thoughts or insights on what could / should / might happen to the spatial pattern of precipitation change in response to solar geoengineering? 

  • AC3: Comment on EGU2020-7708, Robin Chadwick, 05 May 2020

    Hi Angie,

    This paper by Cao et al. shows precip pattern change in response to solar geoengineering and I think there are some other more recent papers which have looked at this as well:

    As you say, the direct solar atmospheric absorption effect is small in the global mean, but actually we found there could be some surprisingly large regional circulation changes in experiments where we fixed SSTs and land surface temperature, but increased the solar constant. See supporting info for this paper:

    In any case, I wouldn't expect solar geoengineering to exactly offset the effect of CO2 forcing in terms of regional precipitation change, and I think that could be a significant danger of solar geoengineering. Particularly as our understanding of regional precip responses to forcing is low, and inter-model uncertainty in projections high.


    • CC4: Reply to AC3, Angeline Pendergrass, 05 May 2020

      Hi Rob, 

      Thanks for your response and for pointing me to these relevant studies. Indeed, I agree that uncertainty is high and there are so many potential unintended consequences, so I don't think solar geoengineering is a good idea to actually undertake. And I'm also aware that some of the more recent work on the topic has more complex forcing (eg GEOMIP) than the earlier studies did. Nonetheless I think the thought experiment provides an example that challenges us to articulate our understanding of precipitation and its changes. I imagine this was what Dietmar was getting at with his question. 



  • AC4: Comment on EGU2020-7708, Robin Chadwick, 05 May 2020

    Yes absolutely, I think looking at idealised experiments across a whole range of different forcings can give some really good (and sometimes surprising!) insights into precipitation change.