21 june 2017

From cdsm-wg

present CB BH SS BvdbG JB AB AG DS GL

updates on mapping progress

o. Northern circumpolar carbon dataset

1. British Columbia (Chuck): disaggregated SLC based on parent material and soil development

2. British Columbia (Brandon): l take a look at the BCSIS data when I get back to Vancouver this weekend. I have all the covariates calculated for BC at 100m spatial resolution and will try running the predictions using a Quantile Regression Forest. It's basically a Random Forest that allows for the development of prediction intervals that we need for uncertainty estimation. I'm not sure if it will work, but it's something that still gives us uncertainties and it's still a map.

3. Agriculture regions SLC3.2 gridded 90 m (Glenn) already complete

4. Detailed and semi detailed soil survey areas 30m gridded (Glenn) complete

5. Saskatchewan (Angela): collecting point data from a variety of sources. will run a map product later in the summer.

6. Ontario (Adam, Dan, Assim): collecting point data from a variety of sources. Assim will do a provincial run

7. Forest regions of Canada (Julien): point data is mostly compiled, in contact with Cindy Shaw for carbom database; boreal peatlands dataset from Thompson.

8. Quebec (JD): A first draft of the map for mineral soil has been produced in May at eight depth (0-15, 15-30, 30-60, 60-100 and 100-200cm). The actual maps are covering the meridional part of Quebec at 250 m resolution (below 52°N). The produced map explained 32% of the total variance of the carbon test datasets. We developed all the code needed to train model and produced map. The training of the model can be done in an hour and the the production of the map for this specific area can be done in 20 minutes. We are actually working to increase the number and the quality of covariates. This should lead to a better accuracy and higher spatial coverage of the map for Quebec area (below 56°N).