What would it take to achieve the Paris temperature targets?

Abstract

The 2015 Paris Agreement aims to limit warming to 2 or 1.5°C above preindustrial level, although combined Intended Nationally Determined Contributions (INDCs) are likely insufficient to achieve these targets. We propose a set of idealized emission pathways consistent with the targets. If countries reduce emissions in line with their INDCs, the 2°C threshold could be avoided only if net zero greenhouse gas emissions (GHGEs) are achieved by 2085 and late century negative emissions are considerably in excess of those assumed in Representative Concentration Pathway (RCP) 2.6 (net −5 Gt CO₂/yr, compared with −1.5 Gt CO₂/yr in RCP2.6). More aggressive near-term reductions would allow 2°C to be avoided with less end-of-century carbon removal capacity. A 10% cut in GHGEs by 2030 (relative to 2015) could likely achieve 2°C with RCP2.6 level negative emissions. The 1.5°C target requires GHGEs to be reduced by almost a third by 2030 and net zero by 2050, while a 50 year overshoot of 1.5°C allows net zero GHGEs by 2060.

Key Points

• 2030 emissions in line with INDCs are unlikely to avoid 2 degrees warming
• Small increases in near-term mitigation allow large reductions in negative emissions
• A 50 year overshoot of 1.5 degrees requires net zero emissions by 2060

1 Introduction

Although the value of a single global-average temperature threshold to represent dangerous climate change can be debated [Victor and Kennel, 2014; Knopf et al., 2012], the 2° threshold representing dangerous climate change has endured in science [Smith et al., 2009; Rogelj et al., 2011; Meinshausen et al., 2009] and in the recent agreements in Paris [Paris Agreements, 2015], which also considers a more aggressive target of 1.5°, argued by some to be necessary [Tschakert, 2015] and feasible [Rogelj et al., 2015a]. The last decade has seen global CO₂ emissions tracking the upper end [Friedlingstein et al., 2014a] of Representative Concentration Pathways (RCPs) [Van Vuuren et al., 2011] considered in the last Intergovernmental Panel on Climate Change (IPCC) report, while a number of studies have suggested that short mitigation delays would necessitate much more rapid decarbonization later in the century [Rogelj et al., 2013; Vaughan et al., 2009; Huntingford et al., 2012] and will likely quickly render the 2° target infeasible [O'Neill et al., 2010; House et al., 2008; Mignone et al., 2008; Stocker, 2013; Friedlingstein et al., 2011].

Invariably, results of such studies depend crucially on assumptions about two factors: first, the maximum feasible rate of decarbonization which is difficult to bind because it is implicitly tied to complex questions of geopolitics and economics. Although recent integrated assessment simulations have achieved 7% yr⁻¹ reductions [Rogelj et al., 2015a; Riahi et al., 2015], this is only by massive transformation of the global energy system in the near term [Bertram et al., 2015], achieved by rapid phaseout of existing infrastructure and by decoupling carbon emissions from economic growth [Rozenberg et al., 2015]. Second is the question of negative emissions capacity which some argue is necessary to achieve the 2° goal [Edmonds et al., 2013], while others argue that the technologies associated with negative emissions are speculative [Fuss et al., 2014] and resource consuming [Smith et al., 2015]. Estimates of net potential removal from all biological and chemical approaches vary significantly [Lenton, 2010].

Many recent studies have framed mitigation choices in the context of cumulative emissions, exploiting a near-linear relationship with peak global mean temperatures in transient scenarios [Allen et al., 2009]. These relationships have been used to produce carbon budgets on a regional level [Raupach et al., 2014] and to produce constraints on the carbon intensity of economic production in the near- and medium-term future [Rozenberg et al., 2015]. A number of papers have noted subtleties in this simplification. Rogelj et al. [2015b] point out that the non-CO₂ forcing agents (methane, black carbon, etc.) can alter allowable carbon budgets to remain below 2° by up to 25% in an extreme case. LoPresti et al. [2015] point out that two scenarios with similar cumulative carbon emissions can result in different rates of change of warming, which can in turn produce very different impacts on natural and human systems, while Krasting et al. [2014] show that the transient response to cumulative emissions is likely to be itself function of emissions rate (with highest values in the limits of very high and very low emission rates).

Although these studies are informative, they do not easily translate into the language of the Paris agreement which does not define commitments in terms of carbon budgets, but rather in terms of near-term emission goals (as represented by the collection of country-level Intended Nationally Determined Contributions (INDCs)) and a desire for greenhouse gas emissions neutrality in the latter part of the century. In this context, it is useful to frame the likelihood of achieving temperature goals as a function of these characteristics of future emissions pathways.

Recent studies have extrapolated the INDCs to assess whether severe climate change can be avoided with their implied level of action [Fawcett et al., 2015], but the degrees of freedom in those pathways are not explored comprehensively. In this study, we define emission pathways with a minimal set of parameters and set out to determine the complete set of scenarios which would be consistent with the 1.5° and 2° temperature targets within this idealized framework. We consider scenarios where emissions depart from an unmitigated baseline scenario (RCP8.5 [Van Vuuren et al., 2011]) at or before 2030 and effectively explore the space of idealized scenarios allowing for a delayed start of mitigation and a modified long-term emission level and rate of decarbonization.

Uncertainty in climate response is sampled using an ensemble of perturbed versions of the Integrated Science Assessment Model (ISAM) simple climate model [Cao and Jain, 2005; Jain and Cao, 2005] such that the ensemble range represents the diversity in global response seen in ESMs (Earth System Models). By our sampling design, we can consider the fraction of perturbed models exceeding a given temperature goal as a proxy for the likelihood of exceeding said goal in reality (Figure S2 in the supporting information shows the likelihoods plotted in terms of the sampled scenario parameters).

2 Methods

Future emissions are considered to follow an unconstrained growth pathway (RCP8.5) until a certain date t_m, which represents the start date for climate mitigation. Global CO₂ emissions (E(t)) after this date are represented by a simple functional form described in the next section (in equation 1).

The idealized pathways ensure that emissions (and the rate of change of emissions) are continuous from RCP8.5 at the date of start of mitigation, t_m. The simple functional form assumes a smooth pathway in which CO₂ emissions peak at some point and then decline, decaying toward a long-term emission floor . This “floor” represents the net anthropogenic CO₂ flux achieved in the far future (but not necessarily in 2100). The time of the peak, the rate of the decay, and the long-term emission floor are adjustable with the equations parameters (Figure 1). Non-CO₂ parameters are treated in a comparable fashion, such that they shift from the RCP8.5 pathway to the RCP2.6 pathway on a comparable timescale.

Figure 1

An illustration of the 3 degrees of freedom for the idealized CO₂ emission pathways considered for this study, varying (a) the start year of mitigation, (b) the time taken after the start year for global emissions to be half way between the start and long-term emissions level, and (c) the long-term emission level.

3 Deriving Functional Forms

The anthropogenic annual carbon fluxes relating to land use, land cover, and fossil fuel combustion are combined into a net anthropogenic source/sink term which is represented in this study by a simple functional form. However, the ISAM framework treats both land use and fossil fuel emissions as separate inputs to the model, and so we must define them both explicitly. Land use emissions (E^LU(t)) are drawn from the terrestrial carbon pool and are prescribed such that they follow RCP8.5 (as reported in the RCP database) until a time t_m and then decay to the RCP2.6 pathway with a decay constant τ, such that

Total emissions (land use and fossil combined, E(t)) are assumed to follow RCP8.5 until a time t_m, after which total CO₂ emissions (fossil, cement, and land use) are assumed to follow the following equation:

(1)

with four parameters: A, t_e, τ, and , for which we solve with the following boundary conditions:

(2)

(3)

(4)

(5)

As such, when the pathway for E(t) diverges from RCP8.5, three parameters define the trajectory: t_m is the time at which emissions diverge from RCP8.5, t₅₀ is the number of years required for emissions to get halfway between the starting emissions E(t_m) and , which represents the asymptotic emissions level for the far future. We aim to span a wide range of solutions which have at least some chance of physically achieving the temperature targets, not all of which will be economically plausible. Hence, we sample values of which range from a small positive flux of 11 Gt CO₂/yr (equivalent to 2100 emissions in shared socioeconomic pathways (SSP)2-RCP4.5—a scenario very likely to exceed 2° [International Institute for Applied Systems Analysis (IIASA), 2016]) to a very large negative flux of −55 Gt CO₂/yr. For context, the largest net removal rate in the SSP database [International Institute for Applied Systems Analysis (IIASA), 2016] is in 2100 in SSP4-RCP2.6, where emissions are −25 Gt CO₂/yr and still decreasing at a rate of −5 Gt CO₂/yr/yr.

The solution and its derivative are continuous with RCP8.5 at t_m (and thus, this condition accounts for 2 degrees of freedom in equation 1). is assessed by numerically differentiating a cubic spline fit to the RCP8.5 emissions pathway.

(6)

ISAM default historical emissions are followed from 1760 to 2005. From 2005 to t_m, the net anthropogenic CO₂ flux follows RCP8.5, and from t_m to 2300, CO₂ emissions follow equation 1. Finally, fossil fuel and cement emissions (E^FF(t)) are calculated as the residual between the total emissions and the land use emissions:

(7)

Similarly to the land use case, emissions trajectories (E^j) are also calculated for CH₄, N₂O, CO, VOCs, NO_x, CFC-11, CFC-12, HFC-134a, HCFC-22, and SO_x. Each follows RCP8.5 until t_m and then decays to the respective RCP2.6 pathway with the decay constant τ, such that

Greenhouse gas emission (GHGE) is calculated by combining all CO₂ emissions with equivalent emissions for CH₄, N₂O, CFC-11, and CFC-12 using 100 year greenhouse warming potentials from the IPCC Second Assessment Report, as in United Nations Environment Programme (UNEP) [2015].

3.1 Creating an Ensemble of Perturbed Simple Models

Calibration. Parameters in ISAM are sequentially calibrated to approximately reproduce the global climate trajectory of Earth System Models in the C4MIP archive [Friedlingstein et al., 2014b] (see Table S1 in the supporting information). For each model, outcomes are aggregated into a nine-element vector X_m. We define a subjective tolerance t for each outcome (informed by the spread seen in the C4MIP experiments), which acts as a normalization term to allow the elements of X_m to be combined into a single metric. A high weight is put on the emulator's ability to predict temperatures and CO₂ concentrations in 2100 by setting the tolerance for these factors to be significantly less than the standard deviation of the values in the C4MIP archive. Other climate outcomes are more weakly constrained to be simply in the range of the C4MIP values by setting the tolerance to be comparable to the standard deviation of outcomes in the C4MIP archive. The perturbed physical parameters in ISAM relate to climate sensitivity, aerosol forcing, ocean heat uptake, and various CO₂-driven and temperature-driven components of carbon cycle feedback (Table S2 in the supporting information) and form a vector p. An ISAM historical/RCP8.5 emission-driven simulation can be used to derive X(p), a nine-element vector corresponding to the ISAM-derived global mean quantities (with each element corresponding to the values from each C4MIP model in X_m). In order to calibrate ISAM to best replicate the behavior of each of the C4MIP models, we define a cost function C_m(p):

(8)

where t(n) is the tolerance for variable n. We then use an interior trust region algorithm [Coleman and Li, 1996] to determine a local minimum for C_m(p), for each ESM subject to the prior parameter constraints (listed in Table S2 in the supporting information) representing the limits of physical plausibility, informed by existing literature [Kheshgi et al., 1999]. Figure S1 in the supporting information shows that in most cases the calibrated ISAM versions can emulate the global response of their C4MIP counterparts.

Parameter-constrained ensemble. Once calibrated versions of ISAM have been produced, we define a flat prior for each ISAM model parameter in between the minimum and maximum values required in the previous step. These values are shown in supporting information Table S2 in the “post” columns. In many cases, the optimization algorithm converges on the minimum or maximum values allowed in the original constraints—so in these cases the bounds remain unchanged. In other cases, all C4MIP models can be emulated using only a subset of the original parameter range.

We then use the updated parameter ranges to conduct a 1000-member Latin hypercube parameter sample [McKay et al., 1979] within the updated bounds of the nine-dimensional parameter space (illustrated in Figure S4 in the supporting information). An ensemble of historical and RCP8.5 simulations is conducted for each member of the ensemble. The distribution of climate system outcomes of this ensemble is shown in the subplots of Figure S1 (in the supporting information) by the light grey histograms.

Outcome-constrained ensemble. Finally, we reject any ensemble member which falls outside of the C4MIP range for any given element of X_m. The remaining ensemble consists of 67 members. The distribution for the key climate system outcomes is illustrated in Figure S1 by the dark grey histograms.

All members of the outcome-constrained ensemble are used to simulate the climate associated with the idealized emissions pathways introduced in the previous section. Each pathway is evaluated from 1760 until 2300, where we evaluate the fraction of perturbed simulations which exceed a given threshold above preindustrial global mean temperatures during the model simulation.

3.2 Emulator Skill

Our analysis uses ISAM to emulate the results of models in the C4MIP archive, and in an ideal case, one would like to validate this emulation with a range of scenarios. An out of sample calibration is not possible because only one scenario (RCP8.5) was completed for the C4MIP experiment. We show in Figure S7 in the supporting information that the emulator trained on RCP8.5 in C4MIP produces a fairly good prediction of each model's concentration-driven response to RCP2.6 in CMIP (midcentury global mean temperature change under RCP2.6 predicted from the emulator exhibits a correlation of 0.82 with the CMIP-derived values). However, a true test of an emission-driven emulator would require additional coupled carbon simulations with a different emissions scenario. Hence, for the purposes of the present study, we are making the assumption that an appropriate set of ISAM configurations which represent uncertainty in climate response can be assessed by the RCP8.5 simulation alone and that the structural assumptions within the ISAM model are appropriate for translating these uncertainties to different emissions pathways.

4 Results

Figure 2 shows the likelihoods of achieving certain temperature targets as a function of GHGEs at two points in the future, 2030 and 2100. First, consider climate mitigation starting in 2005, the year at which the RCP2.6 and RCP8.5 scenarios diverge; Figure 2a shows the RCP2.6 scenario would likely achieve a maximum temperature of less than 2°C above preindustrial conditions (where likely is defined as a greater than 66% chance of the event occurring). However, given that the world has followed an unmitigated emission pathway closely approximating the RCP8.5 since 2005 [Friedlingstein et al., 2014a], the exact trajectory described in RCP2.6 is now impossible. RCP2.6 has a maximum yearly reduction in emissions of 3% (all percentages are expressed relative to actual 2014 emissions in Le Quéré et al. [2014]) and a long-term carbon removal of −4 Gt CO₂/yr. With mitigation starting in 2015 or later, such level of action would no longer have a likely chance of achieving the 2° target. Similar action starting in 2015 would have a 52% chance of staying below 2°, and action starting in 2020 would have only a 28% chance (see Figures 2b–2d and Figure S5 in the supporting information).

Figure 2

An illustration of the exceeded global temperature thresholds as a function of GHGEs in 2030 and in 2100, for four potential start dates. The 2030 emission levels consistent with combined INDCs and pre-Paris policy [UNEP, 2015] are shown for comparison. Colored regions show combinations which likely result in staying below a given target. Black contour lines show the maximum percentage reduction in emissions per year (relative to 2015 values) required during the scenario. Dashed colored lines show the threshold for an allowable 50 year period over a given temperature target. (a) The green circle shows RCP2.6 emissions; (b–d) green circle illustrates the level of action corresponding to the rate of decarbonization and long-term carbon extraction level of RCP2.6.

4.1 The 2° Target

By considering the 2030 total GHGEs associated with the combined INDCs [UNEP, 2015], we can consider what long-term emissions are implied by the various levels of near-term commitment if 2° is to be avoided. Emission policy before the Paris agreement as estimated in UNEP [2015] would result in 2030 emissions some 15% lower than RCP8.5, but this would be insufficient to achieve 2°, irrespective of negative emissions capacity in the future because inertia in the climate system would guarantee exceeding 2°C even with rapid emissions (Figure 2b).

The process of translating the INDCs into global emissions is itself subject to some uncertainty: UNFCCC [2015] finds that 2030 emissions under the INDCs would result in an 8–10% increase on 2015 levels, whereas UNEP [2015] finds that the unconditional INDCs (the portion of the contributions which are not conditional on additional laws or regulations being approved) would result in 2030 emissions being very similar to 2015 (RCP8.5) emissions.

Using the UNEP [2015] values for 2030 emissions, we find avoiding warming greater than 2°C can be avoided with 66% likelihood if net zero GHGEs are achieved by 2085 (i.e., with anthropogenic uptake of carbon dioxide compensating for all emitted greenhouse gases, see Figure 3a). But it should be noted that such a scenario ultimately requires long-term net negative carbon emissions of at least −26 Gt CO₂/yr (for context, RCP2.6 reaches −4 Gt CO₂/yr in net negative emissions, and 2014 CO₂ emissions were +37 Gt CO₂ [Le Quéré et al., 2014]).

Figure 3

An illustration of the exceeded global temperature thresholds as a function of GHGEs in 2030, and the year in the future in which all anthropogenic greenhouse gas sources and sinks first become net zero. The 2030 emissions levels consistent with combined INDCs and projected pre-Paris policy from [UNEP, 2015] are shown for comparison. Colored regions show combinations which likely result in staying below a given target. Black contour lines show the maximum percentage reduction in emissions per year (relative to present-day emissions) required during the scenario. Dashed colored lines show the threshold for an allowable 50 year period over a given temperature target. The plot is shown for two potential start dates, 2015 and 2020, and only solutions with long-term negative GHGEs are plotted because other solutions never achieve neutrality.

If emissions meet the conditional INDC levels as detailed in UNEP [2015] (this would require 2030 emissions to be 3% lower than 2015 levels), the minimum long-term negative carbon emissions would be reduced to −15 Gt CO₂/yr and net zero GHGEs could be achieved a few years later in 2093.

It is evident from considering the shallow contours of the 2° region in Figure 2b that for mitigation beginning in 2015, small differences in the emissions level achieved in 2030 translate to large differences in the required 2100 emissions and the long-term emission level if one is to achieve the 2° goal. This occurs because the small differences in the emission pathways are integrated cumulatively until the time at which emissions become net zero, but there is potentially only a short time available to remove these emissions in the negative phase before the temperature threshold is exceeded. This is illustrated in Figure 4c which shows a number of scenarios, picked from the contour of the 2° region in Figure 2b such that they each have a 66% chance of exhibiting a peak temperature of 2 K warming or less. It is evident that very small perturbations in near-term mitigation rates result in large increases in the future negative emissions capacity in order to likely remain below 2 K warming.

Figure 4

A plot showing a range of pathways for effective CO₂ emissions, all of which begin mitigation action in 2015 and satisfying certain criteria. (a) Pathways which each have a 66% chance of not exceeding 1.5 K warming above preindustrial temperatures, illustrating the trade-off between rapid action and long-term negative emission commitments. (b) Pathways which are likely to overshoot 1.5° for 50 years, (c) pathways which are likely to peak at 2° warming, and (d) pathways which are likely to overshoot 2° for 50 years. RCP8.5 and RCP2.6, as well as the UNEP [2015] estimates for conditional and unconditional INDC emissions are plotted for context.

Hence, more aggressive action in the near-term allows the temperature to be much less sensitive to the end-of-century carbon removal capacity; if we consider the solution which achieves 2° with the least extreme decadal emission reductions relative to present-day emissions (2.9% yr⁻¹, Figure 2b), this would require 2030 emissions of at most 50 Gt CO₂ eq/yr (a 10% cut from 2015 levels) and a long-term carbon emission level of −4 Gt CO₂/yr (comparable to RCP2.6 requirements). All solutions with a delay until 2020 would require significantly faster reductions later in the century (with RCP2.6 long-term emission levels, a reduction of at least 4% yr⁻¹ would be required (relative to present-day emissions), Figure 2c).

In this study, all emissions pathways decay to an emissions floor and are held there until the end of the simulation. Clearly, in reality, negative emissions might eventually be able to be scaled back down to allow atmospheric carbon concentrations to stabilize at an acceptable value. However, because such pathways require a higher dimensional representation of future emission trajectories, and any ramp-down of negative emissions would occur after peak temperatures had occurred, we leave to further study strategies for ramping down negative emissions activity.

4.2 The 1.5° Target

For mitigation beginning in 2015, the 1.5° target requires substantially faster action than the 2° target. All solutions require 2030 emissions levels to be less than 38 Gt CO₂/yr by 2030 (a 30% cut from 2015 levels, Figures 2b and 4a). The solution which minimizes the peak rate of GHGE reduction (4.8% yr⁻¹ relative to 2015 emissions) requires 2030 emissions of 36 Gt CO₂/yr, GHGE neutrality by 2060, and a long-term net carbon removal of −14.7 Gt CO₂/yr. However, the end-of-century carbon removal requirement rises rapidly if the near-term action is less aggressive (Figures 2c and 4a).

If mitigation begins in 2020, a maximum 7% cut in 2020 global emissions per year would be required to avoid 1.5° warming (Figure 2c) and all solutions satisfying the 1.5° target with mitigation starting in 2020 require net zero GHGEs by 2043. Emission thresholds change if a 50 year period of overshoot of the targets was permitted (illustrated by the dashed lines in Figures 2 and 3); however, the magnitude of overshoot for such a timescale is not large. Figure S6 in the supporting information shows that a 50 year overshoot of 1.5° could not exceed 1.8° of warming. It seems clear that the 1.5° target is made significantly easier with a 50 year permitted overshoot, allowing GHG neutrality to be reached 10–20 years later (see Figures 3a and 4b). For the 2° target, the flexibility that a 50 year overshoot allows affects only those solutions which are heavily dependent on long-term negative emissions (see Figures 2b and 4d)—and temperatures do not exceed 2.2° during the overshoot (Figure S6 in the supporting information).

5 Conclusions

These broad conclusions give us some early indication of what should be expected in the maximum mitigation experiments in CMIP6 (which will be conducted as part of the ScenarioMIP experimental design [O'Neill et al., 2016]). The current draft of the experimental design includes an update of RCP2.6 (SSP1-RCP2.6) and a more aggressive mitigation scenario (SSP1-RCP2.0) which will achieve 2.6 and 2.0 W m⁻2 of forcing, respectively, by 2100. But as the present study demonstrates, if mitigation begins in 2015 rather than 2005 this must come at a cost—and in the case of the current version of SSP1-RCP2.6, this is achieved with net end-of-century negative emissions significantly greater than those seen in the original RCP2.6 together with more rapid mitigation of non-CO₂ greenhouse gases. ScenarioMIP also contains a more aggressive mitigation scenario (tentatively planned as SSP1-RCP2.0), which will explore the possibility of reducing forcing to 2.0 W m⁻² by 2100. This is achieved though very rapid near-term mitigation (relative to the original RCP2.6), with net zero CO₂ emissions occurring by 2060 (in line with our estimates of what would be required for a 1.5° overshoot scenario). The realism of such a scenario can be debated, but the present study would imply that unless such rapid near-term mitigation is realized in the coming decade, future phases of CMIP will find it increasingly difficult or impossible to justify the assumptions required to produce scenarios which avoid 1.5° of warming.

In summary, the chances of avoiding the 2° and especially the 1.5° temperature target are highly sensitive to the timing of climate mitigation, which is in line with the findings of previous studies [den Elzen et al., 2010; Rogelj et al., 2013]. Current INDCs could not avoid 2° of warming without relying on substantially greater net negative emissions later in the century than was proposed in RCP2.6, a capability that is not certain to be realized [Fuss et al., 2014; Smith et al., 2015]. In order to achieve 2° with an RCP2.6 level of long-term carbon removal, 2030 net GHGEs must be reduced by 10% from 2015 levels, significantly more than the unconditional INDCs (which allow 2030 and 2015 emissions to be effectively equal).

Avoiding 1.5° of warming altogether, even with immediate action, would require considerably greater effort—at least a 25% cut in effective global CO2 emissions from present-day levels by 2030, a rate of reduction which many would consider impossible. However, allowing for a 50 year overshoot of 1.5° gives more flexibility; we find a scenario in which immediate action achieves a 10% cut in effective CO₂ emissions by 2030, combined with large-scale deployment of negative emission technologies which enable net zero effective greenhouse gas emissions in 2060 would likely allow temperatures to exceed 1.5° for 50 years, with temperatures peaking at 1.8°. Such a 1.5° overshoot scenario would still require emissions in the 2030s to be reduced by over 4% of present-day values per year, and if the world follows RCP8.5 for another 5 years, then a 6% reduction in present-day emissions per year would be necessary. The likelihood of a 1.5° world and even a 2° world can be maximized with substantial and prompt global action, and each year following the RCP8.5 pathway lowers the probability that either target can be achieved.

Acknowledgments

DOE/UCAR CA: This research was supported by the Regional and Global Climate Modeling Program (RGCM) of the U.S. Department of Energy's, Office of Science (BER), cooperative agreement DE-FC02-97ER62402.