Serendipity

Well, I had a fabulous week at the EGU. I tried to take in many different aspects of climate research, but inevitably ended up at lots of sessions on earth systems informatics (to satisfy my techie streak), and sessions looking at current cutting edge research on earth systems models, such as integrating weather forecast and climate models, model ensembles, and probabilistic predictions. Lots of interesting things going on in this space. 

Here’s what I would regard as the major themes of the conference from my perspective:

• Ocean Acidification. It’s pretty easy to predict because it’s linear in the concentration of CO2 in the atmosphere – i.e. there’s no uncertainty at all. When we kill off life in the seas we also lose a major carbon sink.
• Feedbacks. I learned at least nine different definitions of the word feedback, and also that there are a huge number of feedbacks that we might want to put into an earth system model, so someone’s got to work out which ones are most likely to be important.
• Abrupt Climate Change. I learned that the paleontological record tells us that the earth is quite likely to be twitchy, and we still don’t know anywhere near enough about the triggers. Oh, and lots of climate scientists think we’ve already hit some of those triggers.
• Probabilistic forecasting. I learned a lot about the use of model ensembles (both multi-models, and perturbed physics experiments with single models) to quantify our uncertainties. There’s a strong move in the climate community to replace single predictions of climate change with probabilistic forecasts. The simplest exposition of this idea is MIT’s wheels of fortune.
• Simpler targets for policy makers. I’m very taken with the analysis from Chris Jones and colleagues that show that if we want to stay below the 2°C temperature rise, we have a total budget of One Trillion Tonnes of Carbon to emit, and since the dawn of industrialization, we used up more than half of it. 
• Geo-Engineering. Suddenly it’s okay for climate scientists to start talking about geo-engineering. For years, this has been anathema, on the basis that even just talking about this possibility can undermine the efforts to reduce carbon emissions (which is always the most sensible way to tackle the problem). But now it appears that many scientists have concluded that it’s too late anyway to do the right thing, and now we have to start thinking the unthinkable.

Plus some things that I missed that I wish I’d seen (based on what others told me afterwards):

• Stefan Rahmstorf‘s views on the risks of an abrupt shift in the ocean circulation. Still, I consoled myself with this webcast from one of his talks last year: A greater risk of sea level rise? (Stefan writes great stuff: this primer from 2004 on the risk of abrupt change is fabulous).
• The session on The Cryosphere – for how much longer?, particularly fellow Canadian Luke Copland‘s talk on accelerated collapse of the arctic ice shelves; I’m told this was both spectacular and scary stuff.

Okay, I finally got some of the webstreaming working (I needed an updated plugin). I managed to watch some of the medal award lectures after the fact from the EGU webstream page. Turns out the medal award lectures are not so interesting (although Leonard Bengtsson’s lecture on extra-tropical cyclones is worth it for his observations about the current state of the art in modeling cyclones).

However, the press conferences are far more interesting:

• The press conference on uncertainties in climate change is definitely worth it to hear scientists separate what we know from what we don’t know, along with a basic introduction to the principles of climate modeling. Make sure you watch at least until the “wheel of fortune” bit (about halfway through). Bottom line: quantifying uncertainty is crucial. Over dinner this evening we were wishing that other fields (e.g. economics) would be anywhere near this willing to quantify their uncertainties…
• The press conference on improving outreach and education in the cryosphere is great for lots of facts and figures about the frightening rate at which glaciers and sea ice are melting, and the wide ranging implications (it’s a little slow to get going, but worth it once the panelists start).
• The press conference on ocean acidification packs a powerful punch. It starts with the screening of a film about how absorption of CO2 by the oceans leads to dramatic change, told from the perspective of how it will affect the generation of kids growing up today.

I missed out on liveblogging the last session on Tuesday on Seamless approaches in weather and climate, because the room had no power outlets at all, and my battery died. Which is a shame, as it was very interesting. The aim of seamless assessment is to be able to move back and forth between weather forecast models and climate models.

The last speaker in the session, Randall Dole, give a good explanation for the reasons why this is an emerging priority, with his three challenges:

• Understanding and modeling organized tropical convection and its global impacts. This is a key problem in predictability of weather beyond about a week, and a major factor in the regional differences in climate variations within the overall climate change trends.
• Predicting weather and climate extremes in a changing climate (e.g. tropical cyclones, floods, droughts, coastal inundation, etc)
• Integrating earth system models and observations. Or: how to build a scientifically-based, internally consistent record of how the earth system is evolving over time.

Randall also identified an opportunity to provide better information for energy and climate policy, for example to assess the likely unintended consequences of major new energy projects, geo-engineering proposals, etc.

David Williamson from NCAR described the Transpose-AMIP project, which takes a set of models (both numerical weather prediction (NWP) and climate models) and runs them against a benchmark of 2 months worth of real weather observations. The aim is to analyze the primary errors, and is especially useful for comparing parameterization schemes with the field data, to track down which parameterizations cause which forecast errors. The NWP models did dramatically better than the climate models, but probably because they are highly tuned to give high quality forecasts of things like precipitation.

Keith Williams from the UK Met Office Hadley Centre talked about progress on a new initiative there on seamless assessment. The aim is to get all of the Met Office models to use the same physics schemes, from the 2-day weather forecast model all the way to the centennial and regional climate models. (Except in cases where it is scientifically justifiable to use an alternative scheme). Work by Rodwell and Palmer paved the way for this. One of the big challenges is to predict extreme events (e.g. heavy storms and flash floods) under climate change. Keith demonstrated why this is hard, with an example of a particular flood in North Cornwall, which is only predicted by high resolution weather forecast models on a 1.5km grid, and not by climate models working on a 20km grid). The problem is we can’t say anything about the frequency of such events under future climate change scenarios if the models don’t capture them.

Frank Selten gave a talk on EC-Earth, a project aimed at extending the ECMWF weather forecast model, currently rated as the best in the world for medium range weather forecasts, and creating a longer range climate model. Interestingly, the plan is to synchronize this effort with the ECMWF’s seasonal model, rather than forking the code. [Note: In conversations after the talk, we speculated on what software engineering problems they might encounter with this, given that the two will be developed at different sites in different countries. My work at the Met Office suggested that a major factor in their success at keeping the weather and climate models integrated is that everything is done in a single building at a single site. EC-Earth might make a good case study for me]. Oh, and they’ll be using the climate prediction index introduced by Murphy et al to assess progress.

Finally, Prashant Sardeshmukh blew my mind, with his description of the twentieth century reanalysis project. The aim of this project is to recreate an entire record of 6-hour measurements of near-surface and tropospheric temperatures, extending back to the start of the 20th century, using all the available observational data and a 56-model ensemble of climate models. Once they’ve done that they plan to go all the way back to 1871. They do this by iteratively improving the estimates until the models and the available field data converge. I amused myself by speculating whether it would be easier to invent a time machine and send a satellite back in time to take the measurements instead…

Not much to report from this morning, but here’s a few interesting talks from this afternoon:

15:30: Dick Schaap, speaking about SeaDataNet. Another big European project: 49 partners and 40 data centres. Most of the effort focusses on establishing standard data formats and metadata descriptions. The aim is to collect all the data providers into a federated system, with single portal, with a shopping basket for users to search for data they need, and secure access to them through a single sign-on. Oh, and they use Ocean Data View (ODV) for interactive exploration and visualization.

15:45: Roy Lowry, of the British Oceanographic Data Centre, whose talk is A RESTful way to manage ontologies. He covered some of the recent history of the NERC Datagrid, and some of the current challenges. 100,000 concepts, organised into about 100 collections. Key idea was to give each concept its own URN throughout the data and metadata, with a resolving service to get URLs from URNs. URLs instantiated as SKOS documents. Key issues:

• Versioning – if you embed version numbers in the URNs, you have many URNs per concept. So the lesson is to define the URN syntax so that it doesn’t include anything that varies over time. 
• Deprecation – you can deprecate conecpts by moving the collection, so that the URN now refers to the replacement. But that means the URN of the deprecated concept changes. Lesson: deprecation implemented by change of status, rather than change of address.
• WSDL structure – RDF triples are implemented as complex types in WSDL. So adding new relationships requires a change in the WSDL, and changing the WSDL during operation breaks the system. 

Oh, and this project supports several climate science initiatives: the Climate Science Modeling Language, and, of course, Metafor.

16:05: Massimo Santoro, on SeaDataNet interoperability, but I’m still too busy exploring the NDG website to pay much attention. Oh, this one’s interesting: Data Mashups based on Google Earth.

16:45: Oh, darn, I’ve missed Fred Spilhaus’s lecture on Boundless Science. Fred was executive director of the AGU for 39 years, until he retired last year. I’m in the wrong room, and tried the webstreaming, but of course it didn’t work. Curse this technology…

17:30: Now for something completely different: Geoengineering. Jason Blackstock, talking on Climate Engineering Responses to Climate Emergencies. Given that climate emergencies are possible, we need to know as much as possible about possible “plan B”s. Jason’s talk is about the outcome of a workshop last year, to investigate what research would be needed to understand the effects of geoengineering. They ignored the basic “should we” question, along with questions on whether consideration of geoengineering approaches might undercut efforts to reduce GHG emissions.

Here’s the premise: we cannot rule out the possibility that the planet is “twitchy“, and might respond suddenly and irreversibly to tipping points. In which case we might need some emergency responses to cool the planet again. Two basic categories of geogengineering – remove CO2 (which is likely to be very slow), or increase the albedo of the earth just a little bit (which could be very fast). The latter options are the most plausible. The most realistic of these are cloud whitening and stratospheric aerosols, so that’s what the workshop focussed on. We know aerosols can work fast because of the data from the eruption of Mt Pinatubo. Ken Caldeira and Lowell Wood did some initial modeling that demonstrated how geoengineering through aerosols might work.

But there are major uncertainties: transient vs. equilibrium response. Controllability and reversability; ocean acidification continues unaffected; plus we don’t know about regional effects, and effects on weather systems. Cost is not really an issue: $10B – $100B per year. But how do we minimize the potential for unanticipated consequences? 

• Engineering questions: which aerosols? Most likely sulphates. How and where to deploy them? Lots of options.
• Climate science questions: What climate parameters will be affected by the intervention? What would we need to monitor? We need a ‘red team’ of scientists on hand to calculate the effects, and assess different options. 
• Climate monitoring: what to we need to measure, with what precision, coverage, and duration, to keep track of how the deployment proceeding?

If we need to be ready with good answers in a ten-year timeframe, what research needs to be done to get there? Phase 1: Non-intervention research. Big issues: hard to learn much without intervention. Phase II: field experiments. Big issues: can’t learn much from a small ‘poke’; need to understand scaling. Phase III: Monitored deployment.

Non-technical issues: What are sensible trigger conditions? Who should decide whether to even undertake this research? Ethics of field tests? Dealing with winners and losers from deployment. And of course the risk of ‘rogue’ geoengineering efforts.

Takehome messages: research into geoengineering responses is no longer “all or nothing” – there are incremental efforts that can be undertaken now. Development of an ‘on the shelf’ plan B option requires a comprehensive and integrated research program – this is a 10-year research program at least.

Some questions: How would this affect acid rain? Not much, because we’re talking about something of the order of 1% of our global output of sulphurous aerosols, plus problems of acid rain are reducing steadily anyway. A more worrying concern would be effect on the tropospheric ozone.

Who decides? There are some scientists saying already we’ve reached a climate emergency. If the aim is to avoid dangerous tipping points (e.g. melting of the poles, destruction of the rainforests), at what point do we pull the trigger? No good answer to this one.

Read more: Journal special issue on geo-engineering.

Google tells me I’m not the only one blogging from the EGU meeting this week:

• Bente Lilja Bye is blogging about the sessions on Tsunami Risk;
• Dave Petley is blogging about landslides;
• Quirin Schiermeier is blogging on mountain hydrology and in particular the effect of climate change on glaciers in the alps; and comments on Weigel’s talk on seasonal forecasting;

And some others who might blog this week:

• Bryan Lawrence appears to be too busy giving talks to blog much this week (update: he did)
• Robert Huber tells me there is a biodiversity informatics session (found it);

But by and large, Google also seems to be telling me that this community doesn’t blog very much.

Update: 

• RealClimate blogs about the token skeptic;
• Liz Kalaugher does some excellent summaries of the threat to Canada’s ice fields, MIT’s wheel of fortune; and Chris Jones’ trillionth tonne;

A leisurely breakfast this morning, chatting with Tim, so we didn’t make it to the conference until the coffee break. 

10:30: From climate predictability to end user applications: on the route to more reliable seasonal ensemble forecasts, by Andreas Weigel, who is also a Young Scientist award winner. Most of the analysis is based on the ECMWF model. Uses probabilistic forecasts for seasonal predictions – e.g. 41 runs, with perturbed physics, and use probability density functions to create the forecasts. Uses RPSS (Ranked Probabilistic Skill Score) to compare model predictions with observations. Interesting point: if you get lots of random models, and do ensemble forecasts with them, they approach 0 on this skill scale. This kinds of analysis helps to identify bias in the skill score metric, so that this bias can be removed. Multi-models ensembles have been shown to outperform individual models, but this is a bit of a paradox, because the multi-models include less skillful models. Which implies you can improve your forecasts by adding lower skill models to the ensemble. The answer is to do with reducing overconfidence in the forecasts. Last topic for the talk: how can prediction skill be communicated to the public? Introduce an intuitive skill score that makes sense to the public. Rather than just adding up the accuracy of a series of forecasts, look at two different specific observations, and test whether the forecasts correctly distinguish them (this is known as 2AFC). Then add up the skill as the sum of these tests (okay, I’m not sure I’ve got my head around how this works – I’ll need to read the paper…). Oh, and I like the cartoon on the second slide of this talk.

11:15: change of sessions, and I’ve come in partway through Chris Jones’ talk – “Impact of cumulative emissions of carbon dioxide: the trillionth tonne” (Chris is from the UK Met Office, and I had lots of interesting discussions with him last summer). He’s talking about modeling experiments to determine what reduction in emissions is needed to meet the target of stabilizing climate to the 2°C target. Here’s an interesting emergent result from the modeling: peak warming is related strongly to the total cumulative emissions, rather than the specific pathway (i.e. when the emissions occur). This leads to an observation that you should set a total emissions budget as a policy, without constraining when these emissions should occur. Best answer: total emissions should be no more that 1 trillion tonnes of carbon. We’re halfway there right now! So over the next 40 years or so, we mustn’t emit more than 1/2 trillion tonnes. But the longer we leave it before peak emissions, the more dramatic the cuts after that will have to be. The bottom line is that this analysis greatly simplifies climate negotiations, because it makes the target very clear.

11:30: “Marine oxygen holes as a consequence of oceanic acidification“, presented by Matthias Hofmann. It’s well known that higher CO2 levels leads to ocean acidification, which reduces the ability of shellfish and coral to grow, because it inhibits calcification. But how quickly does this occur under different emissions scenarios? There is one bit of good news: there’s a negative feedback – reduced biogenic calcification has a negative effect on atmospheric CO2. But there are also some other effects that are more worrying: the massive growth of oxygen holes, because of oxidation of organic matter in shallow water. This has very worrying implications for marine life. (Here’s the paper).

11:45: last talk before lunch: Quantifying DMS-cloud-climate interactions using the ECHAM5-HAMMOZ model. The CLAW hypothesis suggests there is a negative feedback loop between the ocean and atmosphere, because warmer oceans enhance the growth of phytoplankton leads to increased So2 and hence more clouds (here’s a nice diagram that explains the feedback). Not sure I can summarize the results of the study presented here, except that they showed the effect is seasonal in nature.

Note to self: get to the sessions earlier and find a seat near a power outlet.

Lunch: I managed to visit the exhibition and pick up a couple of books:

• Avoiding Dangerous Climate Change, a collection of papers resulting from a workshop held at the Met Office in 2005.
• A History of the Science and Politics of Climate Change, written by Bert Bolin, the first chair of the IPCC. 

13:30: Ray Bates, giving a talk entitled Climate Feedbacks: Some Conceptual and Physical Issues. Ray is receiving the Vilhelm Bjerknes Medal, and this is the lecture associated with the medal. Standing room only (but I got here first and nabbed one of the only power outlets). Ray started off by giving a little restrospective on his career, starting with his PhD with Charney. Likes the idea of being an Irishman studying tropical dynamics!

Here’s the key idea: most dynamical systems are characterized by negative feedbacks – which keep the system stable. Climate scientists appear to be an exception – they assume climate systems are subject to positive feedbacks that lead to runaway warming. So scientists outside of climate science are often skeptical. To understand this, you first have to understand what the zero feedback case is, and then figure out what we mean by positive/negative feedback. Ray presents four different definitions of “feedback”, F1 from control theory, F2 from electronics, and then two from climate science: F3: a stability altering feedback, and F4, a sensitivity-altering feedback. Ray then points out that any pair of these can give the opposite sign when applied in a particular way to the same system. He then goes on to give several more definitions of different types of feedback in the climate literature. (here’s the paper). Bottom line: an urgent need for a common definition (or set of definitions), so that readers of the climate literature know what we’re talking about.

Ray then gives a long account of Lindzen’s BAMS paper on cloud feedback effects – the paper that causes Lindzen to argue that climate scientists are being alarmist about global warming, because their model (the LCH model) gives a much lower figure for climate sensitivity. Several problems with the LCH model: e.g. it doesn’t include explicit heat transport between the tropics and extra-tropics. Adding these in explicitly gives a very different set of dynamics. With an extended LCH model (with these heat transports) it’s possible to choose parameters that give the opposite feedback effects than when those same parameters are used in the LCH model. (alright, this is a gross simplification of the analysis…) Bottom line: unless we’re much clearer about what we mean by feedback, a lot of the confusion will remain.

14:15: Martin Claussen, giving a talk entitled Is the Sahara a Tipping Element? This work looked at periods in prehistory when the Sahara region was green – covered with grassland. From both the models and the marine sediment cores, it appears that the Sahara flips readily between a ‘green’ state and the desert state, and it only takes a small increase in rainfall to reach this tipping point. As the general circulation models suggest such an increased rainfall as a result of global warming, it’s possible that the Sahara could change dramatically in the coming decades. However, it’s not clear whether it’s a single tipping point, or multiple swings (e.g. different swings for the Eastern vs. Western Sahara). Here’s a summary of the work.

14:30: Peter Brockhaus, giving a talk on soil-moisture feedback effects. Here’s another dilemma about feedbacks. Two different runs of a model (the CCLM) at different resolutions (2.2km and 25km) give soil-moisture feedback effects that are opposite in sign. (Here’s the paper)

14:45: Hezi Gildor, on Lightning-biota feedback effects. This one is fascinating: increased temperature leads to increased incidence of lightning, which generates nitrogen compounds that stimulate plant growth. It also makes the grass greener! The analysis indicates this feedback effect is small, but not necessarily insignificant, so it might need to be investigated in earth system models. [my thought: this begs the question – how many of these different feedback effects do we need to track down and incorporate into the general models? Each new effect that we add increases the complexity of the model, and increases the complexity of the coupling…] Now it gets complicated: one of the questioners points out that lightning also causes forest fires, which burns vegetation (in the short term) but which also stimulate more forest growth (in the long term). More feedback effects to account for!

Time for a break, and some ice cream in the hot Austrian sun.

15:30: Larry Hinzman, talking about Hydrological Changes in the Polar Regions: An Analysis of Linkages and Feedbacks. It’s already getting noticeably drier in many polar regions (many lakes are shrinking), but as the permafrost melts, it generally subsides and significantly increases groundwater, which makes these regions wetter. The connections between different processes here are complex, and Larry indicated they are making progress on sorting them out an quantifying them. [He mentioned a new paper (in submission) that has some nice graphics indicating the linkages]. I did find this recent paper, which summarizes many of the changes in Actic hydrology that have already been seen.

16:45: Emma Stone: Could vegetation feedbacks determine whether the Greenland ice sheet regrows after deglaciation?  This is a long-term question – if we lose the Greenland ice-sheet, will it eventually re-grow once greenhouse gas concentrations stabilize? Two previous studies offer conflicting answers: Lunt’s work suggested it might regrow in 20,000 years, while Toniazzo’s study indicated that it would not happen at all. Emma is running a series of experiments using HadCM3 from the UK Met Office to investigate. She initializes the model with bare soil (for one treatment) and needle leaf (for another treatment), tested under a return to pre-industrial CO2 concentrations. She found that in some runs, some glaciation reappears on the Greenland’s eastern coast, but it depends on assumptions about vegetation. In other words, vegetation feedback effects are critical here for answering the question. Of course, this all pre-supposes that we ever do return to pre-industrial CO2 concentrations…