Coding and AI classes in Glenrothes

The learner requests heights for a 40 by 40 grid covering 27.39 square kilometres around Glenrothes from the OpenTopoData service, which serves the European EU-DEM elevation model. Each cell is about 159 m by 113 m, and heights range from 49.3 m to 174.3 m. Nineteen cells turn out to be pits, lower than all eight neighbours, where the model's water would be stuck; a filling step raises 33 cells by up to 5.98 m so that every cell can drain to the edge. Then each cell sends its water to its steepest downhill neighbour, and the program counts how many cells drain through each one. The largest single flow gathers 944 cells, about 17.0 square kilometres.

To test the model, the learner downloads the waterways mapped on OpenStreetMap in the same area: 46 of them, including the River Leven, Lothrie Burn, Camby Burn, Back Burn, Bighty Burn and Balbirnie Burn. A predicted stream cell counts as a hit if it lies within 150 m of a mapped waterway. By chance alone, 18.5% of all cells are that close.

The bigger the predicted stream, the more often it matches a real one: two in three of the largest flow lines sit beside a mapped waterway, more than three times the chance rate. Smaller predicted streams match less often. Some of the gap is the coarse grid, some is that town drainage runs through pipes and culverts the height data cannot see, and some may be small burns that nobody has mapped. The model is useful and imperfect, and the check tells you how imperfect.

Testing a model against the real world keeps Glenrothes teenagers in charge of the AI they use, and that is worth learning to code for in 2026. The longer argument is in why teenagers should still learn to code in 2026.