SPADE family of AI models

The SPADE family of models has many different incarnations

SPADE stands for SParse DEnse. The models, which belong to the general family of geospatial generative models, leverage the power of AI to smooth out and improve data which is too sparse to be practically useful.

The first application of SPADE has been to the reanalysis, nowcasting and forecasting of okadaic acid in mussels, with very interesing results. We then moved to test it for water depth in the Port of Rotterdam, for oxygen level in inland (and coastal) waters, and on the reconstruction of flows in sewers networks (the family of models). In all these applications the model gave very interesting results. We are now preparing to apply a similar approach to an innovative geospatial project, thanks to a grant from ESA.

In parallel, now that the technology is robust enough, we moved to the engineering of a sustainable business model. Estimates produced with AI are a new thing in most contexts, and even sophisticated corporates sometimes have difficulty “believing” that such a model can provide a step improvement over traditional approaches. As always for startups, business model definition is more difficult than the technology itself!