21 september 2016

From cdsm-wg

Participants :

Chuck Bulmer (CB), Julien Béguin (JB), Bob Macmillan (BM), Asim Biswas (AB), Brandon Heung (BH), Scott Smith (SS), Glenn Lelik (GL), Margaret Schmidt (MS - observer).


1. Update from the main working group (Scott)

* Main outcomes. Notice to participants to join the working group has gone out to several organisation. To date about a dozen people have responded over a couple of days. 
* Training session with Alex Mcbratney in next winter. It will probably of last week in January or first week of February. To be hosted by Simon Fraser University. Fundamentals of predictive mapping and multivariate techniques. R-tools. 
* Some issues on privacy policies of the group have been initiated by Brian Titus. 
* Brian suggested that the working group rotate membership and responsibilities periodically.
* Website will be on the web site  "soils of Canada". University of Saskatchewan is going to take over the server for that particular site. The whole site will have a new look and feel and will be modernized.
* Communications tools are all link to the web site.     

2. Use of, and progress on wiki and google group (JD, Chuck)

* Bob asked for a better wiki-help. JD will provide some quick examples on the issues wiki-FAQ.
* Idea may be best discussed on the google groups forum and final materials can appear on the wiki page which is more public (Asim)
* We should use the chat tool more to discuss ideas and then provide.

3. Review of ST objective statements Our purpose(s) as a scientific and technical group:

  a) encourage, coordinate and guide production of a 100 meter resolution soil map for Canada (global soil map specs).
  b) provide guidelines or protocol and go ahead with a common flow chart that will help to compare validation.
  c) create and improve platform, datasets and procedures for DSM in Canada
  d) What should we do, discuss single map, provincial maps. Once is part of the guidelines how do we merge in a composite maps. 

3.1 Guidelines notes

Guidelines should suggest ways people can produce comparable maps. Guidelines will help to legitimize the work by a fixed criteria (e.g. Mean Square Error). This will help to make the process inclusive and not exclusive. People will work on what they have to do (JD). Same data sets, flowchart should be used, as it will let us a baselines to compare work (AB). CDSM-ST should also stimulate the production of tools to make DSM going forward. Go back to CDSM and follow the same procedure. Guidelines should provide an adequate projection to work with (JD). How to deal with boundary effect between Provinces (J. Beguin). Guidelines should provide a way to merge multiple maps (ensemble predictions). We suggest (at least initially) merging the maps using the weight according to the MSE. Obvious discrepancies may exist but a weighting average may help to reduce the border effect between Provinces. Who is producing the maps or should we encourage to do the map (BM). 

4. What next ? (continue reviewing responses to initial query for ST committee input)

"Please describe a general approach that you think would be effective for producing a Canada-wide soil map at 100m resolution"

5. Also please comment on the updated flow chart (attached) that might inform our efforts.

6. FAO request for C map (SS)

A request of the FAO to Canada to prepare a carbon grid at 1 km. They provided guidelines in an official document (section 5.2). There is some work that is done with Cindy Shaw to complement the database of carbon. Actually the precision of the maps is not sufficient (JB). This could be our first results as a group. can be use to test our coordination, skill and motivation (SS). This might be a case where the group can produce the maps. Please take look at the document (JB). BM think it can be done. Cindy Shaw's database excludes organic soil and work only on mineral soils.  

7. SLC decomposition (GL)

GSM specifications evolves and specifications. 6 fixed primary soil properties. Secondary properties 13-14 that were used to create a DSM. Soil name file and soil layer file and create a product of the slc 3.2.2 by creating a 90m grid cell for the whole country. It was pretty close to the 3 arc seconds and SRTM-DEM. Overlay the SLC polygons and DEM and propose a SLC-decomposition. SLC product (range of components 1-12, heterogeneous unit) were converted to soil horizon file to fixed depth. Each raster is a weighted product of all components of the polygon maps that represent the SLC of Canada (Agricultural area only). It was delivered on the official site of Agriculture and Agri-Food Canada. Accuracy of maps have been recorded based at polygon scale and attributes, which leads to an accuracy of 85% (GL). Not a statictical accuracy as it can be done using continuous value (SS).

8. Proposal (JD) - We should try to limit the time allowed for each meeting to 1:00 hour. Members often left before the end of meeting. The meeting can be extended for interested participants if and as needed. We can also planned extra "urgent" meetings if needed.