Models aren’t the world.
I’m writing this from a café in Rimini, where the air smells like espresso and sea salt, and my wheels hum softly over the worn tiles. A city planner just asked me, “So how hot will our summers be in 2040—really?” The clink of cups and the hiss of steam framed a truth I’ve had to learn the hard way: if you want one number, you’ll get a comforting fiction; if you want wisdom, you’ll need the full messy range—noise, grit, and all. For clarity, I’m simplifying complex scientific ideas here—on purpose—so you can use them.
Three Ideas We Need to Question
We assume bigger, pricier climate models guarantee reliable, local forecasts—the kind that tells your street when floods hit and how hard. You can almost hear the server rooms, cold and loud, promising certainty through sheer computational muscle. Yet today’s complex Earth System Models, brilliant as they are, aren’t equivalent to reality, especially on regional and local scales; they miss or approximate crucial processes, which means we can’t lean on them for detailed local predictions .
We treat precision as the goal—fine-grained maps with razor-sharp lines, like the crisp rustle of a new paper map. But climate prediction isn’t about polishing decimals; it’s about being robust enough to hold when the wind changes and the air smells like rain and ozone. Models differ from reality in ways that matter for local futures, and even when they replay the past, they often wander degrees off for regions like central Europe—too far to trust the vegetation and rainfall patterns they spit out .
We act like uncertainty is ignorance—the squeak you try to tune out, the sand underfoot you wish were smooth. Flip it: uncertainty is information we can use. It’s part of what we know, and it should guide adaptation and policy, not get swept under the rug like crumbs after dinner .
A Story Over Coffee in Rimini
Last spring, in this same café that smells of fresh biscotti and wet pavement, a council member asked me for the “most likely” rainfall change for our coast. The espresso machine roared; my fingertip traced a ring of condensation on the table. I told her about a massive experiment—6,203 model versions, all run to explore different plausible responses—and every single one warmed the planet while pointing to wetter winters in northern Europe and drier summers around the Mediterranean, with the magnitude ranging wildly: less than 10% to more than 50% for winter rain in the north, and temperatures in that region climbing from under 2°C to over 8°C in those scenarios .
Here’s the kicker you can feel in your bones like the shift in air pressure before a storm: the exact numbers aren’t the prize; the range is. The study used an older model and a deliberately unrealistic greenhouse-gas scenario for research purposes, so don’t frame it on the wall—treat it like gravel under your tyres that keeps you from sliding . The takeaway is loud as forks on saucers—plan for bands of futures, not a single outcome, and you’ll still roll when the road changes.
Tomorrow’s budget gets better when we stop pretending tomorrow is a line and start treating it like a corridor. You can almost hear the doors along that corridor clicking open.
What Models Are (And Aren’t)
Think of Earth System Models as grand mechanical clocks you can feel ticking, with cogs of air and sea turning together. Their “dynamical core” solves sturdy physics for fluid motion on grids of about 20–100 km—wind, heat, and currents, like the steady thrum of a tram down the street . Then come the tricky bits—clouds, convection, and ocean eddies—that are smaller than those grid boxes and must be handled by “parameterisations,” which are statistical stand-ins, not the real gears; you can sense the difference like vinyl versus live drums .
That gap matters—more than the smooth texture of a fancy forecast lets on. Miss or fudge the wrong processes, and the local picture bends away from reality, even if the global tune sounds right . The future asks for realism across intertwined processes; anything less leaves us listening to a beautiful symphony with a warped violin.
If we’re building policies on this, we need to check the wood for knots. Tomorrow’s work starts with honest materials.
Why Chasing Resolution Isn’t Enough
There’s a loud push—think Berlin, July 2023, a room humming with projectors and the dry smell of heated plastic—for Earth Virtualisation Engines (EVE): a $15 billion, 10-year bet on 1 km grids and more detailed physics, aiming to resolve stuff like atmospheric convection rather than faking it with statistics . That’s a bold, gleaming server hall vision, chilled air biting your fingertips, and it would sharpen some processes we care about .
Still, the promise that resolution alone will unlock reliable, local, multi‑decadal predictions is a step too far. Many important processes would remain missing, we’d still lean on parameterisations for others, and we don’t even know how “good” is good enough for this kind of forecast—the silence in that answer rattles like an empty lift shaft . Pouring everything into high-res risks starving the work that actually maps uncertainty—and that invites overconfidence, the kind that feels smooth until you hit black ice .
Let’s spend for clarity, not just clarity’s costume. Our future budgets deserve the right kind of noise.
Two Ways To Get Wiser, Not Just Louder
First, stories—with spines. The storyline approach starts with a system we understand—the Indian summer monsoon, cyclone drivers, the weather patterns behind central European floods—and builds credible futures consistent with physics, each crafted with the felt grain of expert knowledge . These stories come in packs, like a deck you can riffle through and smell the ink, each card a different but plausible path, and scaling them needs investment in people, not only machines; even experts tend to anchor to model behaviour, and it takes careful work to pull them back to process understanding (been there, persuaded that) .
Second, diversity—by design. With perturbed physics ensembles, we tweak model assumptions to produce many credible versions, each responding differently to greenhouse gases, like a choir where every voice has grit and air in it . The CPDN project, powered by volunteers’ computers whirring in living rooms and warm offices, showed consistent signals—global warming in all 6,203 versions, wetter winters up north, drier Mediterranean summers—but the scale of change stretched wide, which is the point, not a flaw .
When we focus only on high-res, we leave no room to explore uncertainty—and that tempts us into crisp, wrong answers. A wiser path smells like rain on dust: uncertainty front-and-centre, quantified and communicated, guiding adaptation and the value we put on policy and the energy transition .
What To Do Monday Morning
If you’re a planner, stop hunting for the single “most likely” number the way you hunt for your keys by the door. Name a range that passes the sniff test of physics and expert judgement, then pre‑commit to moves when signals snap into place—heatwave shelters that convert to cooling centres in a day, pumps sized for now with room to bolt on more later, parks that sponge water and give shade you can feel under your fingertips on a burning July afternoon. You’ll hear the city breathe easier when extreme days hit.
If you’re a business owner, test your plan against both the soggy year and the bone‑dry one—smell the damp boxes and the hot dust, then choose investments that survive both. Your resilience should feel like a well‑oiled hinge, not a welded door.
If you’re a citizen, ask for decisions explained in ranges, not in immaculate singulars. The sound of good governance is candid and a little rough.
A Personal Note
As a guy in a wheelchair, I plan my routes with uncertainty baked in—the tactile bump of cobbles, the squeak of a too‑steep ramp, the sweet relief of a smooth curb cut that wasn’t on the map. I don’t wait for the “perfect” path; I carry two or three that work under different conditions and choose on the fly when the wind stings with rain. That’s how I think about climate planning too.
We all live with imperfect maps. The trick is learning the texture of the ground.
Looking Ahead
Uncertainty isn’t a bug—it’s a compass you can grip, even with cold fingers. When models differ from reality in the places we call home, the honest move is to widen our lens, build rigorous storylines, and cultivate model diversity until the future we’re planning for smells like real air and not just conditioned data . I’ll keep pushing for investments that map ranges, not just pixels, and for decisions that flex without snapping when the weather shouts.
We’ll meet the next storm with open eyes and good boots. And yes, the concepts here are simplified—intentionally—so you can use them on your street tomorrow morning.
Uncertainty, held tightly, is courage.
Post a Comment