[OpenCitySmart] [GeoForAll-UrbanScience] Ideas on spatio-temporal analysis of air quality point data

[Tom Roche]

Apologies if the following is too obvious, but since you're a "policy guy," I thought I'd point these out:

Charles Schweik http://lists.osgeo.org/pipermail/opencitysmart/2015-November/000051.html
> [We have AQ] samples (5 second intervals) over 2 weeks in six locations in a city in Asia and measured two public health contaminants. I have a student trying to use QGIS to analyze these data. We want to understand the spatial-temporal patterns in this data (where/when there are particularly dangerous levels, and where/when there are not). 

Your group seems only interested in the spatiotemporality of concentrations, not what drives those concentrations. There are a lotta folks who do the latter, since in many jurisdictions the "fate and transport" of criteria pollutants[1] (or whatever they're called in your Asian city) is important to municipal finance. (You may be familiar, in the US, with the interaction of AQ modeling (driven by the Clean Air Act et al) and transportation planning (driven by programs including STP, MAP-21 née SAFETEA, etc) regionally administered by MPOs[2] and states.) When "real money" is at stake, ya wanna know from {where, what} the pollutants are coming, not just where they are; and for that, you need an AQ model (e.g., CAMx, CMAQ). So I would

1. Try to discover who does AQ modeling for your area of interest. The municipal government or relevant MPO-like consortium should know that; alternatively a brief lit search should turn up pubs from relevant academics and quasi-academics (e.g., the fine folks from EPA AMAD and OAQPS).

2. Ping the AQ modelers regarding what they're using, since every AQ modeler I know does concentration visualizations. Very often such folks are willing to disclose (if they haven't already in their Supporting Materials or whatever their journal calls them) and their tools are open (though there is still IMHO waaay too much MATLAB--worse yet, the egregiousness that is IDL).

I would also ping the fine folks @ openair[3] and PyAOS[4], depending on whether you are more R- or Python-inclined (respectively). The AQ folks I have known worked mostly programmatically (esp with large datasets), so you might find GRASS (combined with AQ-specific code) more useful than QGIS for this usecase.

but ICBW, Tom Roche <Tom_Roche at pobox.com>

[1]: https://en.wikipedia.org/wiki/Criteria_air_contaminants
[2]: https://en.wikipedia.org/wiki/Metropolitan_planning_organization
[3]: http://www.openair-project.org/
[4]: http://pyaos.johnny-lin.com/