Climate scientists released the new CMIP7 scenario set in May. The headline that the upper-end scenario is lower than RCP8.5 has been read in some quarters as evidence that perhaps councils have been overreacting all along.

That is very misleading for a range of reasons.

The revision of the upper-end pathway reflects changes in energy markets, technology, and policy trajectories. RCP8.5 made assumptions that are now less plausible. That is, in a narrow sense, good news. However, CMIP7 has also removed the lower scenario that stayed below 1.5°C, because that outcome is no longer plausible either. We have acted too slowly for it to remain on the table. The range has shifted, not because risk has diminished, but because the plausible futures have changed.

The point of climate scenarios has never been to predict the future. They are structured ways of testing how infrastructure, communities, and economies perform under different conditions, and what thresholds, once crossed, change the decision. A helpful way of thinking about how the scenarios work is to imagine the base of a tent - the tent pegs are the scenarios, they approximately mark out the broad range of futures that could be possible.  The future isn’t any of those pegs, but the future is likely in the tent somewhere (based on our current knowledge). This tells us what you need to be prepared for across the whole space.  What has happened is that we can now define the edges of the tent a little better because some of the futures have been ruled less plausible.  Scenarios will always change and be refined as we learn more, but the question that matters for adaptation is not which emissions label a council uses, it is what physical condition causes the system to fail.

Consider sea-level rise in New Zealand. Under SSP5-8.5, median projections reach around 83cm by 2100. Under SSP3-7.0, that same 83cm arrives around 2110. For an asset or land-use decision with a 50 to 100-year life, that is a minor difference. 

For risk management and adaptation planning, the relevant variable is when an asset fails or a property floods, not the emissions label attached to the pathway. This type of stress testing is how high-end scenarios have often been used, not because planners assumed those exact pathways would occur, but because they identify the physical conditions at which infrastructure stops functioning.

This is why the critique of "overdesign" based on RCP8.5 misunderstands what planners were doing.

Adaptation planning is often misunderstood as committing to expensive interventions based on worst-case futures. In practice, its purpose is almost the opposite. Every adaptation decision carries an opportunity cost. Building too early can lock resources into infrastructure or protection that may not yet be needed, while waiting too long transfers larger costs into emergency response, asset failure, insurance retreat, or abrupt disruption. 

Managing this trade-off is the purpose of adaptive planning. It identifies the trip wires that signal when action is actually needed, e.g. a sea-level increment, a flood frequency threshold, repeated infrastructure failure, insurance becoming unaffordable. These conditions act as signals and when they are reached, the next decision is triggered.

While RCP8.5 is no longer the reference upper-bound scenario, this unfortunately does not remove those trip wires, it mainly shifts when they may be triggered.

If governments and councils now move toward lower-end assumptions to reduce short-term costs, what happens if those thresholds arrive earlier than expected? Ratepayers still pay, but it will be far more disruptive, through emergency repairs, rate shocks, stranded infrastructure, and insurance withdrawal. 

So what should councils do?

First, check what physical boundary conditions your models were built with (the sea-level increment, the storm return period, the rainfall intensity) rather than the scenario label. Those conditions are likely still plausible under CMIP7. If your existing work was grounded in thresholds rather than pathway labels, it remains defensible. If it was not, that is the problem to fix, and CMIP7 is a good prompt.

Second, reframe how you communicate risk to elected members and communities. Lead with the physical condition, not the emissions label. "This road overtops regularly at 0.4m of sea-level rise, and that condition is expected to arrive somewhere between 2060 and 2090 depending on how global emissions unfold" is something a councillor can act on. "This road is at risk under RCP8.5" invites an argument about whether the scenario is still reasonable, an argument that will now be had every time the scenario set is updated. The range expresses the uncertainty honestly and the assumptions visible.

Third, your existing analysis should be anchored to physical increments and thresholds and this is a good opportunity to reframe it. The same physical increment can be reached under multiple emissions pathways, earlier under higher emissions, later under lower ones. What happens when a threshold is crossed does not change depending on the pathway that got you there. Analysis built around conditions rather than scenario labels survives updates to the scenario set. CMIP7 has shifted the timing of some thresholds slightly, but it has not changed what happens when they are reached.

There’s also a wider lesson for the hazard modelling community. For hazard management and adaptation planning, the most useful outputs are not tied to an emissions pathway or a fixed year. Instead, it is more useful to tie model outputs to increments so they can be used for thresholds, tolerance testing, and therefore decision making. Clearer guidance and greater consistency for hazard models would improve planning.

No one wants ratepayers to bear unnecessary costs. But the harder question is what counts as unnecessary: paying now to plan for conditions we can see coming, or paying later, in emergency response, insurance withdrawal, or loss of life, when the options have run out.

Read More:

• For Resilience Explorer users: 

  • Using Scenarios to Explore Uncertain Futures in Planning

  • CMIP7: What it means for existing climate risk assessments and adaptation plans

• Rather than short-term fixes, communities need flexible plans to prepare for a range of likely climate impacts 

• A warning from the future: the risk if NZ gets climate adaptation policy wrong today