GRIDSCAN can now tell a utility which spans will breach clearance in eleven weeks, not which spans breached last month. This is the single most requested capability in the product's history and it took ten years to build, almost none of which was spent on the model.
The model is not sophisticated. It is species-aware growth curves, conditioned on local climate and aspect, fitted to observed change. A competent postgraduate could build it in a fortnight given the data.
The data is the hard part, and specifically the comparability of the data. To fit a growth curve you need measurements of the same trees, from the same viewpoint, corrected the same way, over many years. One methodology change in year six and you have two short time series instead of one long one.
Stadtwerke Oberpfalz has been surveyed annually since 2014 with a consistent methodology because we made an early decision that changes to the processing chain would be versioned and reprocessable rather than simply applied. That decision cost real engineering effort every year for a decade and it is the entire reason this feature exists.
Growth forecasting requires at least two prior survey cycles before predictions are issued. There is no shortcut and we will not pretend otherwise.