EnvirometriX Predictive Mapping Services

EnvirometriX is a science-based company at the Agro Business Park in Wageningen that produces predictive environmental maps for national and international organisations. Its co-founder Tom Hengl describes the gap it addresses: there is more and more satellite data, but less and less capacity to turn it into decisions.

The company combines ground samples with Earth observation covariates using ensemble machine learning and high-performance computing to predict soil properties and nutrients, vegetation and land potential, and to supply decision-ready data for scoping and designing landscape restoration. For iSDAsoil, a soil information service for Africa mapped at 30 m for two depth intervals (0–20 cm and 20–50 cm), EnvirometriX, jointly with GiLAB and MultiOne, helped produce the predictions and build the back-end and front-end of the system. The models used 2-scale ensemble machine learning on Sentinel-2, Landsat, DEM derivatives and coarse-resolution MODIS and PROBA-V covariates, trained on over 100,000 soil sampling points.

Clients named by the company include the UNCCD (country reporting for Land Degradation Neutrality), the Government of British Columbia, InnoTech Alberta (predictive soil mapping for the Alberta soil quality mapping project), ISDA Africa (soil property and nutrient mapping and web services for agriculture) and The Nature Conservancy (the mangrove restoration potential map). It also designs customised sampling campaigns and hosts and distributes large spatial datasets through customised web solutions. Its methods are published openly, including the landmap R package for ensemble machine-learning spatial prediction and the open-access book Predictive Soil Mapping with R.