Pielke Jr and the Misrepresentation of AR6

from IPCC AR6 WG1 Chapter 11

With all the misrepresentation of the scientific literature found in the recent DOE report, I thought it might be helpful to show that misrepresenting the scientific literature has a long tradition in contrarian circles, even among those who give the appearance of supporting the overall science presented in the IPCC technical reports. The strategy can serve an important rhetorical goal if you want to show that the science is on your side while taking a counterfactual stance on the science. If your goal is to show that position X (found in the media or in some scientific study) is wrong or not sufficiently supported by evidence, you can cite a series of papers with extremely selective quotations to give the false impression that this series of papers either falsifies Position X or undermines its credibility. This strategy appears to be a go-to strategy in the DOE report. 

An interesting twist on this strategy is frequently used by those who want to say that the IPCC's technical reports are misrepresented by the media, by other scientists, and even by the the IPCC's popular summaries, like the Summary for Policy Makers (SPM). Even though this is a couple years old, I thought it might be helpful to document this in a blogpost by Roger Pielke Jr from March 29, 2023 entitled "Misinformation in the IPCC." Pielke Jr wants to support his claim that there are "quality control lapses" in producing the SPM that are "serious" and "simply unacceptable."  This post makes two arguments to this effect, but I think Pielke Jr's strategy is most clearly seen in the second of the two. I'm showing below a screenshot of his blogpost so that you can clearly see where he is quoting from either the IPCC AR6 SPM (a popular summary) with IPCC AR6 WG1 (technical report):

Pielke Jr's argument here is that the SPM misrepresents WG1 on tropical cyclones. While SPM says there evidence of "observed changes" in tropical cyclones , including "their attribution to human influence" has "strengthened since AR5." But WG1 says there is "low confidence" in most long-term trends in tropical cyclone frequency and intensity metrics. Then Pielke Jr asks, "how did a strong claim of detection and attribution for trends in tropical cyclones make it into the SPM of the AR6 SR? Let's get to the bottom of the mistaken claim." Yes, let's get to the bottom of this. It turns out the "bottom" isn't what he claims.

I think it should first be noted that the quote from SPM does not make a "a strong claim of detection and attribution for trends in tropical cyclones." Rather, it says evidence of observed changes in tropical cyclones and their attribution to human influence has strengthened since AR5. All this sentence is saying that evidence is stronger in AR6 than it was in AR5. And AR6 WG1 absolutely supports this. But to see this, we first have to show the complete quote from WG1, since Pielke Jr conveniently cut off the quote mid-sentence and ignored the context. Here's the full quote from AR6 WG1:

Identifying past trends in TC metrics remains a challenge due to the heterogeneous character of the historical instrumental data, which are known as ‘best-track’ data (Schreck et al., 2014). There is low confidence in most reported long-term (multi-decadal to centennial) trends in TC frequency- or intensity-based metrics due to changes in the technology used to collect the best-track data. This should not be interpreted as implying that no physical (real) trends exist, but rather as indicating that either the quality or the temporal length of the data is not adequate to provide robust trend detection statements, particularly in the presence of multi-decadal variability (p. 1585).

WG1 is claiming that for two metrics having to do with frequency and intensity, identifying past trends is challenging because of a lack of consistent "best-track" data. So on multi-decadal to centennial time scales, we can have low confidence in those two metrics. However, they also clearly indicate that "this should not be interpreted as implying that no physical (real) trends exist."  The issue has to do with the heterogeneous character of historical data, not positive evidence of a lack of trends. And on the next page, WG1 says that there are other metrics regarding tropical cyclones that do not suffer from the same issues as frequency and intensity metrics. WG1 writes,

Subsequent to AR5, two metrics have been analysed that are argued to be comparatively less sensitive to data issues than frequency- and intensity-based metrics. Trends in these metrics have been identified over the past 70 years or more (Knutson et al., 2019).

This section is explicitly affirming the quote Pielke Jr. gave us from SPM. The report goes on to identify and explain the two metrics that now have stronger evidence:

1. "the mean latitude where TCs reach their peak intensity – exhibits a global and regional poleward migration during the satellite period (Kossin et al., 2014)"[1]
2. "TC translation speed (Kossin, 2018), which exhibits a global slowdown in the best-track data over the period 1949–2016."[2]

In the citations below, I reproduced the full quotes from WG1 related to these. But from this alone it should be absolutely clear that the SPM is being consistent with WG1. There are two metrics in which there is stronger evidence in AR6 than there was in AR5 regarding tropical cyclones and their attribution to human activity, though they are not specifically in metrics having to do with frequency and intensity. On top of this, WG1 concludes this section with:

There is mounting evidence that a variety of TC characteristics have changed over various time periods. It is likely that the global proportion of Category 3–5 tropical cyclone instances and the frequency of rapid intensification events have increased globally over the past 40 years. It is very likely that the average location where TCs reach their peak wind intensity has migrated poleward in the western North Pacific Ocean since the 1940s. It is likely that TC translation speed has slowed over the USA since 1900.

Clearly the SPM quote is supported by the technical material in WG1. Most of the discussion following this has to do with a separate statement altogether, having to do with the observed increase in global proportion of Cat 3+ storms. Pielke objects to the change in wording from "instances" (WG1) to "occurrences" (SPM). Here again is a screenshot of his objection to SPM:

He claims this change in vocabulary "completely changes the meaning of the claim," if we consider the source of the terminology in Kossin et al 2020[3], which distinguishes between observational estimates of hurricanes and hurricane occurrence. He quotes, "Over the period 1979–2017 considered here, there are about 225,000 ADT-HURSAT intensity estimates in about 4,000 individual TCs worldwide." So there can be more "intensity estimates" (called "fixes") than tropical cyclone occurrences because these fixes can happen more than once per occurrence. Pielke Jr assures us that Kossin is concerned with an increasing proportion of these fixes, or observational estimates, rather than increase in the proportion of occurrences of Cat 3+ storms. Given that a popular summary needs to be more understandable to laypeople than the language of a technical report, I think the only legitimate reason to argue that the shift in terminology is misleading is if Kossin et al 2020 shows that there has been an increase in proportion of Cat 3+ fixes without a correlating increase in the proportion of Cat 3+ occurrences. But that is not the case. The corrected version of Kossin et al claims,

Here the homogenized global TC intensity record is extended to the 39-y period 1979–2017, and statistically significant (at the 95% confidence level) increases are identified. Increases and trends are found in the exceedance probability and proportion of major (Saffir−Simpson categories 3 to 5) TC intensities, which is consistent with expectations based on theoretical understanding and trends identified in numerical simulations in warming scenarios. Major TCs pose, by far, the greatest threat to lives and property. Between the early and latter halves of the time period, the major TC exceedance probability increases by about 8% per decade, with a 95% CI of 2 to 15% per decade.

This is graphically displayed in Figure 2.

Time series of fractional proportion of global major hurricane estimates to all hurricane estimates for the period 1979–2017.
Each point, except the earliest, represents the data in a sequence of 3-y periods. The first data point is based on only 2 y (1979
and 1981) to avoid the years with no eastern hemisphere coverage. The linear Theil−Sen trend (black line) is significant at the
98% confidence level (Mann−Kendall P value = 0.02). The proportion increases by 25% in the 39-y period
(about 6% per decade).

Now here is the IBTraCSv4 occurrence data for 1980 - 2023. While I haven't changed this to 3-year periods, see if you can spot a significant difference in the percent change between 1980 and 2017.

The best fit trend through this data shows an increase in the proportion of Cat 3+ storms from 44% to 56% between 1980-2017, or about 27%. This is about the same as that shown when the data is reported in terms of fixes. Pielke Jr's interpretation of the change in terminology is that "somewhere along the way the IPCC attached strong confidence in the detection and attribution of changes in 'occurrences,' — both of which are completely detached from the substance of the article being cited to support the claims." But alternatively, the writers of the SPM knew that the trends in the proportion of occurrences was roughly the same as the trend in instances, and they decided it was fair to summarize the literature with the less technical term "occurrences," since that is also faithful to the exact same data.

It appears that Pielke Jr didn't demonstrate any disagreement between SPM and WG1 or any misinformation in the SPM, at least in this particular claim. Pielke appears to have not done his homework. WG1 backs up and supports the SPM's summary statement.

References and extended quotes:

[1] The first metric –the mean latitude where TCs reach their peak intensity – exhibits a global and regional poleward migration during the satellite period (Kossin et al., 2014). The poleward migration can influence TC hazard exposure and risk (Kossin et al., 2016a) and is consistent with the independently observed expansion of the tropics (Lucas et al., 2014).The migration has been linked to changes in the Hadley circulation(Altman et al., 2018; Sharmila and Walsh, 2018; Studholme and Gulev, 2018). The migration is also apparent in the mean locations where TCs exhibit eyes (Knapp et al., 2018), which is when TCs are most intense. Part of the Northern Hemisphere poleward migration is due to basin-wide changes in TC frequency (Kossin et al., 2014,2016b; Moon et al., 2015, 2016) and the trends, as expected, can be sensitive to the time period chosen (Tennille and Ellis, 2017;Kossin, 2018; Song and Klotzbach, 2018) and to subsetting of the data by intensity (Zhan and Wang, 2017). The poleward migration is particularly pronounced and well-documented in the western North Pacific basin (Kossin et al., 2016a; Oey and Chou, 2016; Liang et al.,2017; Nakamura et al., 2017; Altman et al., 2018; Daloz and Camargo, 2018; J. Sun et al., 2019; T.-C. Lee et al., 2020; Yamaguchi and Maeda,2020a; Kubota et al., 2021) pp. 1586-87.

[2] A second metric that is argued to be comparatively less sensitive to data issues than frequency- and intensity-based metrics is TC translation speed (Kossin, 2018), which exhibits a global slowdown in the best-track data over the period 1949–2016. TC translation speed is a measure of the speed at which TCs move across the Earth’s surface, and is very closely related to local rainfall amounts(i.e., a slower translation speed causes greater local rainfall). TC translation speed also affects structural wind damage and coastal storm surge by changing the hazard event duration. The slowdown is observed in the best-track data from all basins except the Northern Indian Ocean, and is also found in a number of regions where TCs interact directly with land. The slowing trends identified in the best-track data by Kossin (2018) have been argued to be largely due todata heterogeneity. Moon et al. (2019) and Lanzante (2019) provide evidence that meridional TC track shifts project onto the slowing trends, and argue that these shifts are due to the introduction of satellite data. Kossin (2019) provides evidence that the slowing trend is real by focusing on Atlantic TC track data over the contiguous USA in the 118-year period 1900–2017, which are generally considered reliable. In this period, mean TC translation speed has decreased by17%. The slowing TC translation speed is expected to increase local rainfall amounts, which would increase coastal and inland flooding. In combination with slowing translation speed, abrupt TC track direction changes – that can be associated with track ‘meanders’ or ‘stalls’ – have become increasingly common along the North American coast since the mid-20th century, leading to more rainfall in the region (Hall and Kossin, 2019). pp. 1587

[3] J.P. Kossin, K.R. Knapp, T.L. Olander, & C.S. Velden, Global increase in major tropical cyclone exceedance probability over the past four decades, Proc. Natl. Acad. Sci. U.S.A. 117 (22) 11975-11980, https://doi.org/10.1073/pnas.1920849117 (2020).