Deep GIS: Mapping What You Touch In the Subways

I’ve recently been communicating with Ebrahim Afshinnekoo who is Project Director for the PathoMap project based at the Weill Cornell Medicine Mason Laboratory in New York City.  Launched in the summer 2013, PathoMap was the first project of its kind, with the intent to comprehensively map and investigate the presence of bacteria and DNA on the surfaces of large urban, metropolitan environments such as New York City. And of course what better venue to collect bacteria samples in NYC than the subway system – the large subterranean behemoth home to 5.5 million riders on an average weekday.

I was drawn to the project in that it involves several common geospatial components the traditional GIS community is routinely involved with such as  data collection/data validation, data analysis, mobile apps, web mapping and visualization. To date, discussion on this geospatial research effort has focused mainly within the Cell Systems (scholarly journal) community, though with little exposure within the traditional NYS GIS community. While both the Wall Street Journal and the New York Times published articles on PathoMap in 2015 we’ve seen little work of this nature at statewide conferences or how it can promote similar geospatial analysis across the Empire State. With this in mind, eSpatiallyNewYork initiated this blog entry with the purpose of exposing the PathoMap project, and its subsequent global expansion (MetaSUB) to the larger statewide GIS community.

Data Collection

The molecular profiling initiative launched in the summer of 2013 with the help of undergraduates from Cornell University and Macaulay Honors College – which were soon to be given the appropriate moniker “Swab Squad”.  To create a city-wide profile, the research team first built an Android/iOS  mobile application in collaboration with GIS Cloud to enable real-time entry and loading of sample metadata directly into a database (Figure 1).

Figure 1: Data collection from the project included the “swabbing” of sites and subsequent analysis and data entry of the findings into a mobile app which are dynamically uploaded to the Cloud GIS database. Source: Afshinnekoo et al., 2015

Continue reading

Relative vs. Absolute Accuracy Revisited

I recently attended a presentation by Dr. Wende Mix, Associate Professor, Geography and Planning Department at SUNY Buffalo State entitled “Field Data Collection Using Smart Phones, Tablets, and GPS Devices:  A Case Study Though the presentation focused on using mobile devices for field data collection, augmented with high resolution aerial imagery,   Dr. Mix inadvertently helped revisit a debate on the long standing geospatial issue of relative accuracy vs. absolute accuracy. While relative mapping accuracy issues are certainly pertinent as part of the emerging Volunteered Geographic Information (VGI) and crowd sourcing data collection movements, Dr. Mix’s presentation highlights street feature data collection which was once the mapping domain reserved only for surveying and engineering disciplines.

So how do mobile devices intersect with spatial accuracy?  If at all?   Tons of geospatial data being collected with mobile devices (particularly by the growing Smart Phone market), data of varying scales and accuracies by personnel with varying degrees of training and expertise.  But at the end of the day, with all the disparate data combined, the data mash-up stills supports most decision making needs.  Quite a difference from the efforts of New York State Association of Professional Land Surveyors (NYSAPLS) which for years lobbied that similar street feature mapping across the state could legally only be done by licensed surveyors.  So does spatial data accuracy matter anymore?

Of course it does, though an easier answer is that data is normally collected of sufficient accuracy to support specific business needs.    But perhaps the best way to illustrate how this new market of mobile devices plays into the relative accuracy vs. absolute accuracy discussion, one first needs to consider the body of geospatial data development since the late 1970s/early 1980s.

Early Data Development:  With many early government GIS programs getting started with public domain U.S. Geological Survey  1:100000 (+/- 166’)  Digital Line Graph (DLG) or 1:24000 (+/- 40’)  digital files, widespread use of the technology, particularly within the engineering communities and urban environments, was slow to take hold because the data was considered too generalized and “not accurate” enough. Beyond the human resources needed to manually digitize and convert hardcopy manuscripts, much of the first generation of geospatial data was cheap to acquire and develop.  The trade-off was that the geospatial data was of limited accuracy and content due largely to the generalized nature of the source documentation.

However, as data accuracy improved through photogrammetric projects creating many urban and metropolitan land bases at larger and more accurate scales (1”=200’  & 1”=100’), including a wide range of planimetric datasets such as building footprints, edge of pavement, hydrology, bridges and even stone walls – so did the associated data development costs.   However, the increased accuracy and completeness of the data resulted in a much broader acceptance and use within the engineering community.   And certainly some of this increased user acceptance was also a result of the growing inner-operability between GIS and AutoCAD software packages.  Improved  GPS technology (as well as with “Selective Availability” being discontinued in 2000) also gave government and industry additional tools to further push the limits of high accuracy feature mapping, though as a whole, industry mapping costs remained high.  And cadastral programs continued to mature making large scale digital tax map datasets available providing even more reference and content to both hardcopy and online mapping efforts.  Overall, particularly in the urban environment, higher accuracy datasets with features being mapped a higher degree of positional accuracy,  were slowly replacing the more generalized first generation land bases.

Referencing Data Collected in the Field:   With many urban and even rural land bases now created and available online as a service and augmented by a variety of high resolution aerial imagery services, a large portion of data collected by mobile devices can now be easily referenced and spatially edited to its right relative location. (And as an added benefit, normally at a lower data development cost.)  Most mobile devices now include cameras, so including a picture of the selected feature adds even greater context to its relative location.  Using desktop tools, fire hydrants can be moved to their right relative X,Y location in front of the proper house.  Catch basins can be spatially adjusted to register in their right relative locations on street corners, street signs in their right relative location in the right-of-way, or the locations of underground storage tanks or septic fields moved to their right relative location on the proper tax parcel.  Overall, an industry witnessing an increased body of geospatial data that is not absolute accurate, but relatively accurate and ultimately more useful to a larger community of users – including the public works and engineering disciplines.   (It is noted that some workflows and business models may limit or not include resources for the editing of data; thus requiring high accuracy data capture in the field).

Dr. Mix’s presentation unintentionally illustrated how far we’ve come in context of building and using relative accurate geospatial datasets.  The content of her presentation was both typical and timely as much of the work across the state with mobile devices is being used with public infrastructure and street feature mapping.  While non-survey grade GPS units initially introduced some of these very same issues, the new mobile devices, and in particular Smart Phones, are game changers in context of affordability and ease-of-use.   It is to be seen long term how Smart Phone data collection will impact the low-end GPS hardware market.  (Any Google search on “GPS vs. Smart Phone Data Collection” will provide a long list of opinions on the matter.)

There is no question absolute (or near) accuracy – and its high price tag in data acquisition – is still mandatory for engineering and design/build projects.  But for nearly all other business needs, relative accurate and complete datasets will continue to augment design/build projects and support government and industry decision making.    All said “everything happens somewhere – and it is increasingly being mapped in its right relative location”.