by Judith Curry
How did the state of New Jersey come to adopt sea level rise projections for their adaptation planning that are more than twice as high as the IPCC’s values?
Part I introduced the challenges facing New Jersey associated with sea level rise, and their general adaptation strategy. Their adaptation strategy is driven by a report written by sea level researchers at Rutgers University in New Jersey [link]. The main results from the Rutgers Report is this table of sea level rise projections for NJ:
My involvement in this started on December 21, when I received an email from Ray Cantor seeking my evaluation of the Rutgers Report. After a phone call to discuss, a small amount of funding was approved on January 28 for me to assess the Rutgers Report. They needed my evaluation report by February 28. With one month, a small amount of funding, and a schedule that was already heavily committed, I put together an evaluation report with the help of a very capable assistant. This project was just too irresistible to turn down.
My evaluation report can be downloaded [NJ SLR report CFAN final]. This blog post addresses scenarios of 21st century sea level rise outcomes, including a critique of the Rutgers scenarios. The main points are excerpted in this blog post; see the full report for background and details.
Also see this press release from the NJ Business and Industry Assoc [link]
2.1 21st century sea level rise projections
Global sea level rise projections provide a basis for projecting local sea level rise, such as for the New Jersey coast. The analysis in this section places the projections provided by the Rutgers Report in context of other assessment reports. The analysis provided below demonstrates that the Rutgers projections of sea level rise are substantially higher than those provided by the Intergovernmental Panel on Climate Change (IPCC).
In 2019, the IPCC published a “Special Report on Oceans, Cryosphere and Climate Change” (SROCC), which included updated sea level rise projections (Table 2.2) based on the same CMIP5 climate model simulations that were used in the IPCC AR5. It is instructive to compare the sea level rise projections in the second-order draft of the SROCC versus the values in the final report. The SROCC values for RCP2.6 and RCP4.5 are comparable to the AR5 values. However, the SROCC sea level rise projections for RCP8.5 (high emissions scenario) are significantly higher than the AR5 values (upper end of the likely range is 0.98 m). It is notable that the RCP8.5 values in the final SROCC report are substantially lower than in the second order draft. The volatility of the sea level rise projections for RCP8.5 (high emissions scenario) reflects deep uncertainties in understanding of the dynamics and potential instability of the West Antarctic ice sheet and its influence on future sea level rise.
Table 2.2: Projections of global mean sea level rise for 2100 from the IPCC SROCC (baseline period 1986-2005).
The sea level rise scenarios from the forthcoming IPCC AR6 are not yet available. However, a publicly available letter (King et al. 2020) cited the following values from the 2nd order draft of the AR6: projections for 2100 range from 0.82 – 0.98 meters (32.3 – 38.6 inches), for a surface temperature increase of 4.3 – 4.8oC (7.8 – 8.6 oF). While the AR6 temperature projections are higher than the AR5, the CMIP6 sea level rise projections for the emissions/ concentration scenario equivalent to RCP8.5 are lower than the values in the SROCC. It remains to be seen whether the sea level rise projections will change in the AR6 final report.
While there is general convergence in the IPCC reports on the projected likely range for sea level rise for the low and intermediate emissions/concentration scenarios (RCP2.6, RCP4.5), there is substantial uncertainty and disagreement regarding sea level rise for the high emissions scenario (RCP8.5). This uncertainty is associated with the impact of potential instabilities in the West Antarctic Ice Sheet, which could be triggered by large values of warming.
2.3 Rutgers Report
The emissions scenarios used in the Rutgers Report do not directly relate to the emissions scenarios used in the IPCC AR5 and SROCC (e.g. RCP2.6, RCP4.5, RCP8.5). Rather, the Rutgers Report selects two scenarios based on the amount of warming since early industrial (1850-1900): 2oC (low emissions) and 5oC (high emissions). The Rutgers high emissions scenario is close to RCP8.5 through 2100. However, the Rutgers low emissions scenario reflects more warming than RCP2.6. The Rutgers team then averaged the sea level rise projections for their high and low emissions scenario to create a moderate emissions scenario, nominally associated with a temperature increase of 3.5oC.
The rationale for using these new definitions of high and low emissions scenarios is to accommodate the Bamber et al. (2019) expert elicitation on sea level rise associated with potential instability in the West Antarctic ice sheet, which used the 2oC and 5oC scenarios. It is somewhat surprising that the Rutgers Report elected to structure their scenarios following Bamber et al., since the Report states that “SEJ [structured expert judgment], however, is not fully accepted by the ice-sheet modeling community, as it relies on the calibrated mental models of the participating experts rather than explicit physical models.”
Since the Rutgers Report didn’t use the same emissions scenarios as the IPCC, it is not straightforward to compare them. Global sea level rise projections that are cited in the Rutgers Report as providing the basis for their local sea level rise projections are: Kopp et al. (2014), Kopp et al. (2017), Rasmussen et al. (2018) and Bamber et al. (2019). Kopp et al. (2014) yields projections of likely global mean sea level changes that are broadly consistent with IPCC SROCC. Kopp et al. (2017) replaced the original Antarctic ice-sheet mass loss projections of Kopp et al. (2014) with those from the Antarctic ice-sheet modeling study of DeConto and Pollard (2016). Bamber et al. (2019) replaced the Greenland and Antarctic ice-sheet projections of Kopp et al. (2014) with projections based on an expert elicitation of ice-sheet changes associated with climate scenarios leading to 2°C and 5°C of warming by 2100.
Table 2.4 compares the likely range predicted by Kopp (2014) and Kopp (2017) with the IPCC projections. While the Kopp (2014) projections are fairly close to the SROCC, the Kopp (2017) projections are almost twice as high as the IPCC projections for RCP4.5 and RCP8.5. While not directly comparable to the other projections, the Bamber et al. (2019) values are even higher.
Table 2.4: Comparative projections of global mean sea level rise (meters) for 2100 based on different emission scenarios.The Rutgers scenarios of future climate change are formulated in a fundamentally different way than the IPCC. The IPCC provides a range of temperature and sea level rise projections for a discrete number of emissions scenarios. This results in a continuum of temperature and sea level rise projections whose likely range overlaps among the different emissions scenarios. By contrast, the Rutgers Report provides distributions of sea level rise for two temperature scenarios: 2oC and 5oC. A third, intermediate scenario is provided by simply averaging the temperature and the sea level rise percentile. The Rutgers approach does not discriminate in a meaningful way the range of temperature scenarios (including the lower bound) for which the potential instability of the West Antarctic ice sheet might contribute substantially to sea level rise in the 21st century – the relevant processes are nonlinear with temperature and cannot be interpolated based on temperature in a meaningful way.
4.3 Evaluation of predictions from the Rutgers Report: 2000-2020
The projections provided in the Rutgers report for NJ sea level rise use as a baseline the period 1991-2009 (nominally the year 2000). Hence, the observed tide gauge data through 2020 can be evaluated against the sea level rise projections in the Rutgers Report (Table 4.1). With regards to the 2030 projections, we are already more than 2/3 through this period. When based on the trends since 1980, there is 3.51 inches (0.29 feet) of sea level rise at Atlantic City during the period 2000-2020. Reaching 9.6 inches (0.8 feet) by 2030 (the Rutgers scenario with 50% chance) would require a very substantial acceleration for the remainder of the 2020’s.
Figure 4.2 overlays the observed time series since 1980 of sea level on top of the Rutgers sea level rise projections for Atlantic City. The projections for 2030 and 2050 are independent of emissions scenario. The solid blue line is the Rutgers ~50%; the dash line reflects the likely bounds (17-83%) and the dotted line reflects the very likely bounds (5-95%).
The short-term trend in the observed sea level record is dominated by large year-to-year variability. The observed sea level record between the period 2000 and 2020 appears to be tracking between bottom of the likely and very likely ranges of the Rutgers forecast.
Figure 4.2: Historic and projected relative sea level in centimeters (y axis) over time (x axis) for the period 1980 to 2050 in Atlantic City. Historic data is based on data utilized in figure 4.1 from NOAA tidal records. Future data is based on values found in the Rutgers Report Table 6.
4.4 Recommended scenarios of sea level rise for New Jersey
Based upon the recent historical record since 2000, there seems little justification for a 2030 prediction that exceeds the bottom of the likely range (>83% chance exceedance).
For the period to 2050, consideration is needed of the natural modes of ocean circulation patterns, especially given the current period with the Northeast U.S. coastal ‘hot spot.’ A further issue of relevance is the influence of these same ocean circulation patterns in the North Atlantic on the mass balance of Greenland. The bottom of the likely range in the Rutgers Report also seems like a good bet out to 2050.
For projections to 2100, emissions scenario RCP4.5 seems the most appropriate to use. However as described in Section 2.3, the projections of sea level rise in the Rutgers Report for the moderate emissions scenarios do not relate to RCP4.5. The IPCC’s projections for RCP4.5 are more consistent with the Rutgers low emissions scenario.
The wild card (potential Dragon King) and largest uncertainty is associated with a potential large contribution from instability in the West Antarctic ice sheet. The IPCC SROCC assessed this contribution to be considerably lower than that provided by the expert elicitation of Bamber et al. (2019), which is used in the Rutgers Report.
For decision making and policy purposes, the most important outstanding issue is continued investigation and assessment of the contribution of West Antarctic ice sheet instability under moderate temperature increases associated with emissions scenario RCP4.5. They key issue is assess whether or not there is a plausible Dragon King scenario that should be considered for the moderate emissions scenario.
7.Conclusions and recommendations
The sea level projections provided by the Rutgers Report are substantially higher than those provided by the IPCC, which is generally regarded as the authoritative source for policy making. The sea level rise projections provided in the Rutgers Report, if taken at face value, could lead to premature decisions related to coastal adaptation that are unnecessarily expensive and disruptive.
The Rutgers Report develops its own scenarios of sea level rise and does not reference the sea level rise scenarios of the IPCC AR5, SROCC or the AR6. However, the approach used in the Rutgers Report that relies heavily on publications by Kopp et al. (2017) and Bamber et al. (2019) does not pass muster in the IPCC SROCC (2019) in context of their choices for sea level rise projections:
“For this reason, their results and probabilistic (e.g., Kopp et al., 2017; Le Bars et al., 2017) and statistical emulation estimates that build on them (Edwards et al., 2019), are not used in SROCC sea level projections.”
“The expert elicitation approach (Bamber et al., 2019), which applied elicitation to both ice sheets, suggests considerably higher values for total SLR for RCP2.6, RCP4.5 and RCP8.5 than provided in Table 4.3.”
Experts inevitably disagree owing to inadequate data, insufficient understanding, different evaluations of the various classes of evidence, and different logical frameworks for linking the available evidence. ‘Which experts’ are included in a particular assessment report or expert elicitation makes a difference to the outcome conclusions. Therefore, it is important for a practitioner developing scenarios for policy applications to provide context from other assessments (particularly the IPCC) and other experts, as well as the temporal rate of change of expert opinion on the topic at hand. None of this context is provided in the Rutgers Report.
Individual experts can be out ahead of the IPCC assessments in developing better scenarios. However, the CFAN Review does not judge this to be the case with the Rutgers Report.
The Rutgers Report characterizes their projections as “Consensus Science to Support Planning for Sea Level Rise in New Jersey.” While their projections may reflect a ‘consensus’ among the authors of the Rutgers Report, they do not reflect a consensus of international experts on climate change and global sea level rise. The consensus on climate change and sea level rise is better represented by the IPCC assessment reports.
Well, it will be interesting to see what kind of response my report gets from the Rutgers team and the NJ EPA. As far as I can tell, the points I make are pretty unassailable. We’ll see.
Back to the opening statement for this post:
How did the state of New Jersey come to adopt sea level rise projections for their adaptation planning that are more than twice as high as the IPCC’s values?
The lead authors on the Rutgers Report – R. Kopp and W. Sweet – were also lead authors on the 2017 NOAA Report on Sea Level Rise Scenarios and also on the 4th U.S. National Climate Assessment chapter on sea level rise. Kopp is a lead author on the CMIP6 sea level chapter. Kopp in particular is a prolific publisher on the topic of sea level rise.
Being a successful academic is associated with different motives and a different skill set than being a successful practitioner in supporting adaptation decision making. Relative to private sector practitioners, academic scientists have no ‘skin in the game’ – there are no adverse consequences for an overconfident forecast with a 5-30 year time horizon that turns out to be really wrong. Private sector practitioners in climate services have skin in the game in the sense that they will lose money or go out of business if their confident forecasts turn out to be wrong. While 80 year predictions are beyond the lifetime of the practitioners and most likely their companies also, the time period for many relevant decisions is the 5-30 year time frame.
Academic scientists promote their own research in order to get recognition, promotion, funding etc. However, this his is not helpful in supporting decision making or in the process of synthesis and assessment. On the other hand, private sector practitioners will consider any tool that they deem relevant or helpful.
And finally, synthesis and assessment is another skill that runs counter to the normal mode of operation for a research scientist, requiring subsuming one’s professional ego in context of a broad survey and evaluation of the relevant research. Individual scientists tend not to be very objective about their own research. While a diverse group of scientists on an assessment panel (e.g. IPCC) helps ensure that individual hubris and promotion of the individual’s own work doesn’t dominate the assessment, we have certainly seen examples in the past where this is not the case.
Part III addresses what I have learned about best practices in climate adaptation and how climate science can be useful, based on my real world experience with real world decision making.