Field Notes · Island
A week on Socotra, simulated and then lived
We ran the Journey Simulator for Socotra in March with eight days and the comfort band, then went. Here is what the model said, what the island said back, and where the two disagreed.

The Journey Simulator said: eight days, highs of thirty, lows of twenty-two, two rain days in the month, sunrise at six, sunset a little after six, no altitude to speak of, a comfort-band cost of about two hundred euros a day, and two nervous flags: the weekly flight and the wind. It laid the sample plan across the eight days and gave each an effort score, with the dune at Arher and the canyon at Wadi Dirhur as the hard ones. Then we went, in the second week of March.
What it got right
The temperatures, within two degrees on the highs. The sunrise, to the minute; we watched it from the dune at Arher at 06:04 and the model had said 06:02. The flight as the frame of the trip: the whole island runs on the days between two Wednesdays, and every itinerary from every operator is the same length for that reason. The cost, within ten percent in the comfort band, once we added the charter flight as a fixed item. The effort scores for the dune and the canyon, both of which the group felt in the legs.
What it got optimistic
The nights. The normals say twenty-two; on the beach at Arher the wind from the sea made it feel like eighteen and the light sleeping bags were suddenly worth having. On the Diksam plateau at seven hundred metres the night was cooler still, and the cook produced blankets from somewhere. The model has a wind adjustment in the ice realm and did not have one for a tropical island; it does now, and it is small.
The water. Three litres a person was the model’s allowance on the walking days, and the guide’s Land Cruiser carried it. Nobody ran short, but nobody had a bottle left at the end of the canyon day either, and the model’s allowance is now three and a half above thirty degrees on the island’s plateau days.
What it got wrong
The hours in the vehicle. The sample plan lists the east coast as a day; the road from Hadibo to Arher is three hours of sometimes beach, and the return is the same, and a day of sights plus six hours of driving is a long day that the effort score did not carry. The simulator now has a transfer factor for Socotra that it did not have before, and the Journey Simulator’s timeline shows the driving hours as their own line. We would not have found that at home.
What no model holds
The man at Detwah lagoon who lives in a cave and catches octopus with his hands. The dragon’s blood forest at Firmihin in the last light, red sap on the bark, the trees like umbrellas in a storm that stopped a thousand years ago. The sky at Shoab beach with no light on any horizon. The model got us there in the right month with the right shoes and enough water. The rest was the island.
Questions people ask
Was the model accurate on temperature?
Within two degrees on the highs and a little optimistic on the nights, which were warmer than the normals on the coast and cooler on the plateau. The wind was the thing the normals could not hold.
What did it get wrong?
The number of hours in a vehicle. The sample plan says a day; the road says a day and a half. The Journey Simulator now carries a transfer factor for Socotra that it did not have before we went.


