Creating and Commercializing a Predictive Ecosite Classification Platform for Alberta

Ecological land classification (ELC) is a process for delineating ecologically distinctive areas and expressing relationships between vegetation, physical geography, and soils. Though this topic is vital to many sectors, no provincial scale ecosite information presently exists. Current approaches to producing and delivering ecosite information in Alberta is inefficient and costly, beset by information barriers, results in poor quality interim products and is focused on site level requirements. This project had two phases. Phase 1: (1) pilot project to determine if machine learning techniques can create accurate maps of ecosite information in two pilot areas; and (2) develop a model for such a mapping platform to ensure its continued operation over the next 30+ years. Phase 2: address two questions: (1) can soil layers be mapped predictively/accurately enough to reduce field sampling required under current pre-disturbance assessments (PDAs)?; and (2) can coarser scale predictions be used to streamline PDAs for soil by selecting sampling locations based on the degree of uncertainty from the predictive model?

More detailed conclusions can be found in the 5 reports that were generated from this project. A few key findings include: (1) predictively mapped ecosites and soil attributes could potentially be used in PDAs and conservation and reclamation plans (whether vegetation information can also be mapped predictively remains to be examined); (2) the simulation study showed an adaptive sampling approach targeting areas of higher predictive uncertainty outperformed PDA style sampling; further improvements in adaptive sampling would further increase its efficiency and cost-savings relative to PDA style sampling; (3) little value is being realized from ecosite data collected; it could be used, archived and shared more effectively to support environmental and financial objectives. The regulatory system needs to change to support this, and will need to be monitored closely to ensure critical data and information is available and in a compatible format.