by Judith Curry
“The results show that the extreme sea levels observed during Hurricane Katrina will become ten times more likely if average global temperatures increase by 2°C”, said Dr Jevrejeva. That would mean a storm surge of Katrina proportions every other year. - Alex Kirby
This PNAS paper received a lot of press in March, but I didn’t get around to doing a post. With the North Atlantic hurricane season just about underway, consideration of this paper is timely.
Projected Atlantic hurricane surge threat from rising temperatures
Aslak Grinsted, John Moore, Svetlana Jevrejeva
Abstract. Detection and attribution of past changes in cyclone activity are hampered by biased cyclone records due to changes in observational capabilities. Here, we relate a homogeneous record of Atlantic tropical cyclone activity based on storm surge statistics from tide gauges to changes in global temperature patterns. We examine 10 competing hypotheses using nonstationary generalized extreme value analysis with different predictors (North Atlantic Oscillation, Southern Oscillation, Pacific Decadal Oscillation, Sahel rainfall, Quasi-Biennial Oscillation, radiative forcing, Main Development Region temperatures and its anomaly, globaltemperatures, and gridded temperatures). We find that gridded temperatures, Main Development Region, and global average temperature explain the observations best. The most extreme events are especially sensitive to temperature changes, and we estimate a doubling of Katrina magnitude events associated with the warming over the 20th century. The increased risk depends on the spatial distribution of the temperature rise with highest sensitivity from tropical Atlantic, Central America, and the Indian Ocean. Statistically downscaling 21st century warming patterns from six climate models results in a twofold to sevenfold increase in the frequency of Katrina magnitude events for a 1 °C rise in global temperature (using BNU-ESM, BCC-CSM-1.1, CanESM2, HadGEM2-ES, INM-CM4, and NorESM1-M).
Published online by PNAS [abstract]. This paper is based upon a data set that was described in a previous paper:
Homogeneous record of Atlantic hurricane surge threat since 1923
Aslak Grinsted, John Moore, Svetlanta Jevrejeva
Abstract. Detection and attribution of past changes in cyclone activity are hampered by biased cyclone records due to changes in observational capabilities. Here we construct an independent record of Atlantic tropical cyclone activity on the basis of storm surge statistics from tide gauges. We demonstrate that the major events in our surge index record can be attributed to landfalling tropical cyclones; these events also correspond with the most economically damaging Atlantic cyclones. We find that warm years in general were more active in all cyclone size ranges than cold years. The largest cyclones are most affected by warmer conditions and we detect a statistically significant trend in the frequency of large surge events (roughly corresponding to tropical storm size) since 1923. In particular, we estimate that Katrina-magnitude events have been twice as frequent in warm years compared with cold years (P < 0.02).
[link] to abstract
Critiques of this paper can be found at:
The latest online issue of PNAS has a letter (comment) [link] by Andrew Kennedy et al. and response from Grinsted et al. [link]. Kennedy et al. argues that:
Because the Pensacola elevations are so low (<2 m NAVD88), their use by
Grinsted et al. (2) degrades the Katrina standard to such an extent that it becomes possible to conceive of multiple Katrina events per decade; this would not be possible using more appropriate surge values. The danger of using spatially distant measurements to represent the magnitude of surge events thus becomes clear. Inappropriate comparisons with Hurricane Katrina have their own unique dangers, as this storm has great emotional resonance. Instead of a Katrina-magnitude event, the authors computed the probabilities of far more moderate surges.
Grinsted et al. respond (my mind boggles):
If Kennedy et al. are correct that our measure of storm surge for Katrina magnitude events is too low, then we can examine the implications of choosing a more extreme benchmark. One of our results is that the most extreme surges are also the most sensitive to warming, in agreement with many other hurricane studies. Thus, we would have projected an even greater relative frequency increase if a more extreme benchmark for Katrinas had been chosen [as argued by Kennedy et al.].
I agree with all of these critiques, and have a few additional criticisms of my own. What I object to most is their turning their rather dubious analysis into alarming predictions of future storm surge activity:
Statistically downscaling 21st century warming patterns from six climate models results in a twofold to sevenfold increase in the frequency of Katrina magnitude events for a 1 °C rise in global temperature
I find the reasoning to get to this prediction to be mind boggling.
Deep uncertainty in long term hurricane risk
The key issue of concern is the future risk from major hurricanes and storm surges. A much better perspective on this problem is provided by the following paper.
Deep uncertainty in long-term hurricane risk: Scenario generation and implications for future climate experiments
Nicola Ranger and Falk Niehorster
Abstract. Current projections of long-term trends in Atlantic hurricane activity due to climate change are deeply uncertain, both in magnitude and sign. This creates challenges for adaptation planning in exposed coastal communities. We present a framework to support the interpretation of current long-term tropical cyclone projections, which accommodates the nature of the uncertainty and aims to facilitate robust decision making using the information that is available today. The framework is populated with projections taken from the recent literature to develop a set of scenarios of long-term hurricane hazard. Hazard scenarios are then used to generate risk scenarios for Florida using a coupled climate–catastrophe modeling approach. The scenarios represent a broad range of plausible futures; from wind-related hurricane losses in Florida halving by the end of the century to more than a four-fold increase due to climate change alone. We suggest that it is not possible, based on current evidence, to meaningfully quantify the relative confidence of each scenario. The analyses also suggest that natural variability is likely to be the dominant driver of the level and volatility of wind-related risk over the coming decade; however, under the highest scenario, the superposition of this natural variability and anthropogenic climate change could mean notably increased levels of risk within the decade. Finally, we present a series of analyses to better understand the relative adequacy of the different models that underpin the scenarios and draw conclusions for the design of future climate science and modeling experiments to be most informative for adaptation.
Published in Global Environmental Change [link]
From the Discussion:
From this series of arguments, we conclude firstly that the direct use of single or sets of current GCMs in risk management, without an appropriate treatment of uncertainty, could lead to potentially costly maladaptation and unnecessary risks.
This research has suggested a number of specific lessons for the design of future climate modeling experiments and climate analyses to meet these needs; firstly, the importance of better understanding the role of natural cycles (and therefore, climate change) in driving variability in tropical cyclone activity and the climate of the Atlantic.
There is a need for more study of the adequacy of models. There are many fundamental questions that need to be addressed: for example, does the fact that current GCMs are unable to fully represent historical co-variations of MDR SSTs and windshear suggest an inadequacy for forecasting future conditions for tropical cyclone formation and evolution? Is it sufficient for a model to represent current tropical cyclone climatology and multiannual variability, even if there are suggestions that they are right for the wrong reasons? From such questions, one might be able to define a set of necessary, not sufficient, tests for model adequacy for a given application. Such analyses may in time enable one to exclude certain scenarios or estimate relative confidence and therefore, refine adaptation decisions.
Thirdly, the need to explore the full range of uncertainty in future states. In the main, climate modelers have attempted to generate projections that represent a ‘best guess’ conditioned on a particular model structure. Recently, some studies have set out to more fully explore the range of uncertainties in future climate projections. We argue that while these studies represent a significant step forward in exploring uncertainty, they still do not explore the full range of uncertainties because they are conditioned on one (or a handful of nonindependent) GCMs. This means that uncertainties associated with model structure are not explored. We suggest that to inform adaptation, climate experiments include analyses that leave the confines of current GCM structures and attempt to explore the range of possible outcomes, for example, using simple models or considering theoretical limits.
Empirically derived scenarios of future hurricane activity
Arguments for using a broader range of scenarios than those provided by climate modes were made on this previous post Alternative approach to assessing climate risk.
In the work that I am starting to do on empirically-based scenario development for future hurricane landfall impacts, I have been looking at the following frameworks for developing scenarios on decadal time scales (out to 20 years) :
- Climatology: The simplest climate scenario assumes that the tropical cyclone climatology for the next two decades is the same as that for the previous 30 years.
- Persistence: A persistence-based scenario applies a trend or other statistics using a shorter period than climatology that was selected using some physical rationale.
- Dynamic climatology: Use teleconnection indices (e.g. AMO, PDO) and the expression of Atlantic hurricane characteristics to construct future scenarios, accounting for uncertainty in length of regime periods and timings of regime transitions.
- Secular global warming: Additional scenarios are created based on the other scenarios by including (as a multiplier effect) possible impacts from secular global warming: small secular increase in tropical cyclone intensity and the percent of category 4 and 5 hurricanes, decrease in the ratio of U.S. landfalls to total Atlantic hurricanes, increase in rainfall, and a small secular sea level rise.
Several years ago I conducted a study for World Bank Latin America, ”Potential Economic Impacts of Hurricanes in Mexico, Central America, and the Caribbean ca. 2020-2025” that used a similar empirically-based scenario approach.
Specifically with regards to storm surge. In addition to the frequency and intensity of hurricanes at landfall, a key factor is the horizontal size of the hurricane, which was a major factor in the storm surge magnitude for Hurricanes Katrina and Sandy. How hurricane horizontal size changes with climate is a new topic that is being studied by several group.
The historical record for U.S. landfalling hurricanes since about 1900 is pretty robust, and storm surge can be estimated fairly reliably using models in locations and periods where storm surge observations were lacking. Tide guage measurements at a limited number of locations (where it is impossible to separate out hurricane induced surges from other causes) are a very poor proxy for hurricane activity. Erroneous inferences from the tide guage measurements combined with dubious applications of climate models produces a faux storm surge hockey stick.
Ranger and Niehorster correctly argue that climate model based scenarios of future hurricane activity have numerous problems and also do not introduce a sufficiently broad range of scenarios. On decadal timescales, I think that generating empirically-based scenarios is a far better approach.
Based upon my understanding, here is what I think is the most likely scenario. North Atlantic hurricane activity will remain high until the transition to the cool phase of the AMO. Sometime during mid-century (when the temperatures start rising), North Atlantic hurricane activity will calm down, analogous to the 1970s and 1980s. While intensity may show some increase with rising temperatures, U.S. landfalls could be less if the hurricanes continue to form more frequently in the eastern Atlantic, and curve northward on the open ocean.