TROPICAL CYCLONES and CLIMATE CHANGE ASSESSMENT Part I: Detection and Attribution

TROPICAL CYCLONES and CLIMATE CHANGE ASSESSMENT Part I: Detection and Attribution

TROPICAL CYCLONES AND CLIMATE CHANGE ASSESSMENT Part I: Detection and Attribution THOMAS KNUTSON, SUZANA J. CAMARGO, JOHNNY C. L. CHAN, KERRY EMANUEL, CHANG-HOI HO, JAMES KOSSIN, MRUTYUNJAY MOHAPATRA, MASAKI SATOH, MASATO SUGI, KEVIN WALSH, AND LIGUANG WU We assess whether detectable changes in tropical cyclone activity have been identified in observations and whether any changes can be attributed to anthropogenic climate change. he question of whether anthropogenic influence To conduct this assessment, we identified studies on tropical cyclone (TC) activity is detectable in where claims were made about detection of long-term T observations is important, particularly owing to (multidecadal to century scale) changes in TC activi- the large societal impacts from TCs. Detection and ty, or about attribution of past TC changes or events to attribution studies can also inform our confidence in anthropogenic forcing. We next developed potential future TC projections associated with anthropogenic detection/attribution statements for the author team warming. This report updates TC climate change as- to evaluate. In a few cases—for completeness—we sessments by a World Meteorological Organization developed and evaluated a potential detection and (WMO) expert team (Knutson et al. 2010) and by the attribution statement for which no published claim Intergovernmental Panel on Climate Change Fifth has yet been made. We used an elicitation procedure Assessment Report (IPCC AR5; IPCC 2013; Bindoff whereby each author provided confidence levels or et al. 2013). Part II of the assessment (Knutson et al. agree–disagree opinion on each potential statement. 2019) focuses on future TC projections. Walsh et al. The summary assessment statements highlighted (2016) provides a recent extensive literature review of in the “Summary” section represent cases where a TCs and climate change studies. majority of authors either support the statement or The authors of this report include some former support an even stronger version of the statement. members of the expert team for the WMO 2010 A full list of opinions for each individual author assessment (Knutson et al. 2010) along with cur- and each potential statement is contained in the on- rent membership of a WMO Task Team on Tropi- line supplemental material (https://doi.org/10.1175 cal Cyclones and Climate Change. The Task Team /BAMS-D-18-0189.2). members were invited to membership by the WMO Some related topics, such as long-term trends in World Weather Research Program’s Working Group TC damages (Pielke et al. 2008; Mendelsohn et al. on Tropical Meteorology Research. 2012; Estrada et al. 2015; Klotzbach et al. 2018) or AMERICAN METEOROLOGICAL SOCIETY OCTOBER 2019 | 1987 changes in mortality risk (Peduzzi et al. 2012), are detectable change refers to one that is highly unlikely beyond the scope of our assessment. While total (typically p < 0.05) to occur from natural variability storm damages have increased over time, these are alone. Natural variability here refers to changes (either strongly influenced by socioeconomic trends (Pielke forced or from internal variability) that occur because et al. 2008; Hsiang 2010; Hsiang and Narita 2012; of natural processes alone, without human influence. Camargo and Hsiang 2015). Attribution involves evaluating the relative con- tributions of different causal factors to an observed BACKGROUND ON DETECTION AND AT- change, along with an assignment of statistical con- TRIBUTION OF TC CHANGES. Detection fidence (Hegerl et al. 2010; Bindoff et al. 2013). This and attribution studies explore the causes of observed can involve multivariate methods comparing spatial climate changes or events, typically by comparing and temporal patterns of modeled forced signals models with observations. TC activity has been a very to observations (Hegerl et al. 2010), or univariate challenging topic for detection/attribution, owing methods comparing individual observed regional or partly to the comparative rarity of TCs and the dif- global time series to models (Knutson et al. 2013b). ficulty obtaining accurate and temporally consistent While attribution is normally assessed once a cli- measures of TC properties for climate studies (e.g., mate change has been established as detectable, it Landsea et al. 2006; Kossin et al. 2013), particularly has become more common practice for investigators in the pre–satellite era (Vecchi and Knutson 2008). to make attribution claims even in the absence of Observed TC time series are often limited by data establishing a detectable change, particularly in the quality issues and by the length of available records case of extreme event attribution. We refer to such from various monitoring techniques (e.g., satellites). attribution claims as “attribution without detection.” TC historical time series show far less compelling evi- An example case of this is Murakami et al. (2015), who dence for long-term increases than does global mean infer, using model simulations, that anthropogenic temperature (e.g., Bindoff et al. 2013). Several long forcing has increased the occurrence rate of TCs near TC-related observed time series are shown in Fig. 1 Hawaii and thus contributed to the active TC season (discussed below). Longer TC records are potentially in that region in 2014, despite the lack of a significant available from historical documents (e.g., Chenoweth observed long-term increasing trend in TCs in the and Divine 2008). None of these observed TC time region. While such statements will typically have series demonstrate clear evidence for a century-scale relatively low confidence, they can help identify pos- increase similar to that observed for global mean sible anthropogenic climate change signals that might temperature. later become detectable. Within this report, we use the term “detectable Natural variability contributions can be assessed anthropogenic change” to refer to a change (e.g., via long preindustrial climate simulations with trend) that is both clearly distinguishable from natural constant forcing (which simulate internal variabil- variability, and is consistent—at least in sign—with ity) or by historical simulations forced by natural a modeled (statistically significant) expected change forcing agents only (e.g., volcanic and solar forcing). due to anthropogenic influence. More narrowly, a Paleoclimate proxies can aid in evaluating how AFFILIATIONS: KNUTSON—NOAA/Geophysical Fluid Dynamics Melbourne, Parkville, Victoria, Australia; WU—Nanjing University of Laboratory, Princeton, New Jersey; CAMARGO—Lamont-Doherty Information Science and Technology, Nanjing, Jiangsu, China Earth Observatory, Columbia University, Palisades, New York; CORRESPONDING AUTHOR: Thomas R. Knutson, CHAN—Guy Carpenter Asia-Pacific Climate Impact Centre, City [email protected] University of Hong Kong, Hong Kong, China; EMANUEL—Depart- The abstract for this article can be found in this issue, following the table ment of Earth, Atmospheric and Planetary Sciences, Massachusetts of contents. Institute of Technology, Cambridge, Massachusetts; HO—School of DOI:10.1175/BAMS-D-18-0189.1 Earth and Environmental Sciences, Seoul National University, Seoul, South Korea; KOSSIN—Center for Weather and Climate, NOAA/Na- A supplement to this article is available online (10.1175/BAMS-D-18-0189.2) tional Centers for Environmental Information, Madison, Wisconsin; In final form 20 May 2019 MOHAPATRA—India Meteorological Department, New Delhi, India; ©2019 American Meteorological Society SATOH—Atmosphere and Ocean Research Institute, The University For information regarding reuse of this content and general copyright of Tokyo, Chiba, Japan; SUGI—Meteorological Research Institute, information, consult the AMS Copyright Policy. Tsukuba, Japan; WALSH—School of Earth Sciences, University of 1988 | OCTOBER 2019 unusual an observed climate variability/change is Adopting only the type I perspective has substantial compared to preindustrial behavior, but have so far potential for missing anthropogenic influences that been limited to a relatively small number of regions/ are present but have not yet emerged or been identified locations. In some TC-climate studies, a statistically to a high level of confidence. Therefore, we also make significant linear trend or secular change is identi- some assessment statements considering the type II fied based on conventional statistical tests (e.g., t error perspective. For this, we consider whether there test). This type of statistical test does not necessar- is a balance of evidence (more than 50% chance) that ily demonstrate that the observed change is highly anthropogenic climate change contributed nontrivially unusual compared to natural variability. If the time in a specified direction to an observed change. Further, series is long enough, an estimated contribution from for climate change detection, we adopt weaker criteria natural variability can be statistically removed (e.g., (e.g., p = 0.1) for statistical significance and also assess Kossin et al. 2016a; Kossin 2018b) to assess whether whether the balance of evidence suggests that a de- a significant trend remains, which would increase tectable change has been identified—a lower burden confidence in a climate change detection. Confidence of evidence than used for type I error avoidance. We in anthropogenic climate change detection depends recognize that the alternative type II error framing will on a number of factors: confidence in the estimates of lead to more speculative TC detection and/or attribu-

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    22 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us