Posts

Toward a complete SAR base map

Mapping in search and rescue is both a key part of a successful mission and a mess. Truly, it’s an absolute mess.

Ideally, everyone would be looking at the same map. The map would be complete — all area trails, roads, former roads, points of interest, etc. Whatever device is used by ground teams to view this map would be capable of relaying its position, track and waypoints back to the incident command post in real time. And whatever mapping is done at the ICP — search areas, clues, cell-phone forensics, assignments — would be relayed back to the field units in real time, too. Situational awareness would just ooze out of every device and flood the operation with efficiency and effectiveness.

That all seems simple enough, but it is not. For one thing, SAR teams deploy a few dozen or hundred times a year, depending on the group, and only a subset of those missions call for any serious mapping. We just don’t get good at it — or we haven’t yet, anyway.

Pick a platform. Any platform.

In the front country with cell service, our needs could be met by everyone using a team account on SARTopo, an ATAK server or ArgGIS Online. Each of these has its advantages, drawbacks and costs. SARTopo, its team functions and its mobile app get closest to meeting the requirements at minimal cost and learning curve, but it is not a true GIS capable of being fed new shapefiles as needed. ArcGIS provides the most flexibility and functionality, and a SAR template and utilities exist, but it can be pretty inaccessible to the person at the ICP doing mapping (and the SAR world is not exactly full of GIS pros). ATAK is feature-rich and can be fed a customized set of GIS layers, but it is Android hardware-specific for now.

Away from cell service, another set of limitations immediately arises. Garmin InReach devices are the best thing going for position reporting and messaging, though some teams use amateur-radio-based APRS to accomplish much the same thing without expensive subscriptions. InReach with a business account enables a couple of key features that do not exist in personal accounts: device-to-device position reporting and an ability to feed location data to another program (such as SARTopo); APRS can also perform these functions.

All of the technology choices aside, what literally provides the base of the whole thing is a good base map. Each of the above options has a somewhat suitable base map except ArcGIS, where the GIS tech needs to build one. But a full set of the ideal data for front- and backcountry SAR in any region of the U.S. is found in a variety of datasets from federal, state and local agencies. Each of the available maps gets close but currently misses the mark.

Can we at least agree on a map?

This brings me to the topic at hand. There’s no perfect and affordable technology available at present to meet every SAR need, but surely we can at least compile the layers of data from various agencies necessary to make a solid base map. Everyone has REST services posted, and these are easily added to an ArcGIS Pro map and symbolized, resulting in a base map. Right?

It’s actually harder than it should be. Let’s start with trails. They cross agency boundaries. They exist in cities, on private timber land, in national forests and in municipal parks. Far too many agencies own trails, and too many trails are not maintained or owned by any agency, to compile a comprehensive set of trail layer. Thankfully for my area of the country, the Northwest Trails data for Garmin devices is actually pretty close, but it is a private project and does not necessarily track trail creation, modifications and closures. I found that to be close enough and added some obvious newer trails from our county’s GIS.

Next, roads. This is far worse. SAR teams want to know about every road, from the highway and county roads needed to get to a scene to the forest road that was closed 40 years ago but might provide easier overland travel than crashing through the brush. State and local GIS departments provide the authoritative data on their road networks. Federal land-management agencies provide the data on park-service and forest-service roads, even roads that are now closed to motor vehicles. That just leaves all of those abandoned logging roads and some other gaps.

In this state, the Department of Natural Resources comes to the rescue to fill that gap. Unfortunately, they fill it like someone who’s never used expanding foam insulation from a can. City streets and forest roads with geometry slightly different from the authoritative sources are in the same layer as those all-important logging roads. The names of roads not under DNR’s authority are often incorrect and the remaining attributes nearly always woefully incomplete, offering little hope of filtering by owner.

What to do? I messed around with the data for a few days then just decided to keep the DNR roads I could not filter out without losing other key roads. I attempted to bury them below the authoritative roads, which worked only when the geometry matched.

Thorough road data in this section of the Olympic National Forest required both the correctly named road segments in the USFS data (white roads) and the underlying DNR road data (double dotted lines), which is visible along the right side of this map where its geometry differs from the USFS data.

Ugliness ensued. I can’t believe the best choice is to tolerate multiple overlapping layers attempting to show the same roads. But here is an illustration of why it’s so essential to show both. This is from the front country, a large swath of mostly private timberland adjacent to a residential community.

County roads are white with blue borders, private roads solid gray and logging roads again in double dotted lines.

Approximately half of those old logging roads are shown in SARTopo’s base map. Garmin’s Explore maps (part of the InReach system purchased with Delorme several years ago) show even fewer roads and are pretty much unrecognizable as the same swath of ground.

It’s a start.

The annoying artifacts of overlapping road networks aside, I managed so far to assemble a workable base map for SAR operations in this county. I included most of the same data from three surrounding counties, too, but subjected it to a bit less scrutiny. In addition to the Northwest Trails data, county roads and DNR roads, it includes state-park roads, National Park Service roads, U.S. Forest Service roads, parcel boundaries and owners, structures, contours, hillshade, tree canopy, water channels, water bodies, public-land boundaries and county boundaries.

It’s not incomplete, yet it is fairly unobtrusive and looks like a base map on top of which a mess of SAR data can be drawn. I would love to have some group knowledge mapped — the helicopter landing zones, radio-relay points, way trails and climbing routes we have used in the past.

And just like that, I notice that my new base map is missing a key trail — not a primary route, but one we need to know about. SARTopo has it. So does OpenStreetMap, another potential source of a thorough trails layer.

Assessing hydrant coverage within a fire district

For the final project for a class, I took a look at how to delineate the area of a fire district that can reasonably be serviced by existing fire hydrants. The topic came up because firefighters are interested in seeing how far they can stretch an engine’s large-diameter hose from the nearest hydrant. They could use such maps to train new people unfamiliar with the reach of the hydrant network into rural areas. And the map data could be used to cross-check CAD data that automatically dispatches tenders to fires at certain addresses.

Another practical application of this work would be the prioritization of installing additional hydrants. One could easily see where can a hydrant could be added on or near existing water mains to serve the highest number of previously unserved structures.

“For effective firefighting it is of crucial importance that there are sufficient fire hydrants and that they are properly maintained” (Nisanci 2010). A fire engine arrives on scene with perhaps two to four minutes of water supply on board — connection to a hydrant or service from a team of slow-moving water tenders (tanker trucks) is therefore essential in the case of a working structure fire.

Methods

Hydrant coverage was calculated as 1000′ along the road network from each hydrant to account for the ability of a standard fire engine to deploy up to 1200′ of large-diameter hose from a hydrant to a fire scene (leaving 200′ for driveway length and hose routing). This type of buffer was calculated using the Service Area feature of Network Analysis in ArcGIS, because large-diameter hose can only practically be deployed along roadways.

In addition, a standard buffer was added to the coverage area defining a 400′ radius around each hydrant. This accounts for coverage from some hydrants that are not located on roadways or are at the end of roads but ostensibly could be utilized in the immediate vicinity (i.e. at the airport or a school).

These buffers were combined into a single layer, which was subtracted from the district’s boundary polygon, yielding a layer showing the areas not effectively served by hydrants.

Results

Sample detail of the map to ensure no further use of this unvetted analysis. Hydrants are circles colored by max water flow. The muted red shading is the area not reachable either within 1000′ by road (shown on the right) or 400′ in a direct line (shown on the left) from a hydrant.

The map above shows the un-hydranted areas of the district shaded in red. Of the 16,384 structures located within the boundaries of the district, 4,285 are located beyond hydrant coverage. Most hydrants are concentrated within the city.

Discussion

This fundamental output of the analysis can be used both in fire response (adaptation to the problem) and in adding hydrants (mitigation of the problem).

First, the map above can be used in greater detail to provide firefighters a concise expectation of where they will have effective hydrant coverage versus where they will need a tender-based water supply. The layer file can also be imported into the 911 center’s computer-aided dispatching system to automate tender response where it will be required.

Again, a detail of the map showing in blue-to-yellow heat map the high concentrations of structures that lie in the area unserved by hydrants.

Second, we can use this analysis to respond to the problem by adding fire hydrants where they are especially needed. For this purpose we turn to the map above, which is a heat map showing relative concentration of structures that are beyond hydrant coverage. The areas of high concentrations of significant size would, provided water lines are located nearby, be logical neighborhoods in which to add fire hydrants.

As with the first map, the muted red shades areas unserved by hydrants, and hydrants are shown as colored dots. The black squares are structures located in the unserved area.

Above is an area from the heat map in greater detail. It features a concentration of structures located adjacent to hydrant coverage but in the uncovered, shaded area. Given the adjacent water supply, areas such as this present opportunities for near-term addition of relatively few hydrants to better protect high concentrations of structures.

Suggestions for future study include performing fire-flow analysis based on building construction type, occupancy type and size and comparing required water flow to available water flow for each structure. Kaufman and Rosencrants (2014) describe a GIS-based method for doing and provide the tables necessary to make the calculations. Given the necessary inputs for structures, GIS would make such an analysis practical.

References and Data Sources

Road centerline, trail. park and district boundary data from Jefferson County GIS. Hydrant data from City of Port Townsend GIS and Jefferson Public Utility District GIS. Hydrology and structure data from WA Department of Natural Resources.

Kaufman, M. M., & Rosencrants, T. (2015). GIS method for characterizing fire flow capacity. Fire Safety Journal, 72, 25-32. doi:10.1016/j.firesaf.2015.02.001

Nisanci, R. (2010). GIS based fire analysis and production of fire-risk maps: The Trabzon experience. Scientific Research and Essays, 5(9), 970-977. Retrieved December 13, 2020, from https://academicjournals.org/journal/SRE/article-abstract/DDFB12518871

Draft fire template with FAA elevation

I spent some time during a fire toward the end of the Pacific Northwest season working on a draft template to replace the standard one used for printed map products on wildland fires. There is certainly precedent for spending more time making the template than on making a good map — just add lots of logos, lots of names, lots of other information irrelevant to fighting fire. I do not want to perpetuate such things by handing another team my projects based on a non-standard template.

But I did want to take a shot, even just for fun, at upgrading the standard template. I want to improve a few notable issues:

  • The painfully slow performance of text updates due to its use of complex, dynamic text boxes in ArcGIS Pro;
  • The need to move individual boxes around inside the neat line depending on map content, sometimes triggering that slowness in Pro;
  • The use of fonts that make it look relatively boring and staid; and
  • The sometimes inability to see a rolled map’s title and timestamp.

The draft above (a fairly small view of the upper portion of a map intended to be plotted at 36″x48″) shows the header outside the neat line. It is intended to be horizontal on portrait maps and rotated 270 degrees on landscape maps, thereby keeping the header on the outside of a tightly rolled map.

The text boxes are more numerous so contain fewer dynamic fields per box. Only the legend remains to be relocated within the map, plus the coordinate table on airops maps.

The large title is a sans-serif font, while the remainder is a serif font. This alone brings the overall look of the document forward a decade or two. The title is the most important piece of information, followed by the operational period for which the map is current. Other information like the GIS Specialist responsible and even the name of the fire are far less important data to everyone involved, from the firefighters and pilots to the incident commander.

Along with the template change, this map includes two additional upgrades I have been playing with. As with most of our airops maps this season, it uses Pro’s table feature to draft point data directly from the relevant table in the open white box at upper left; it takes some work to get the coordinates to display effectively, but it seems better than exporting to Excel, formatting and importing as a table.

Also, the DEM is set to the FAA standard color table for navigational charts by elevation; this works in relatively mountainous areas but would be unimpressive in flatter areas. Here are the colors and elevations in feet:

Approximate color (hex)Minimum elevation (feet)Maximum elevation (feet)
D0DEAFn/a0
DEE6C311000
D7E4C610012000
C2CBB620013000
FDF0CE30015000
FDECA750017000
D6C26C70019000
C2903B900112000
6C5B4B12001n/a

Glacier-covered area is supposed to be white. Given the rare need for this in wildland fire, I have not yet played with that. However, it should be simple to take the state hydrology data, filter it for glaciers and use that to cover the DEM in those areas.

Spread thin, will GIS specialization diminish?

The “GIS Technician” job is not long for this world, though I had a difficult time finding published thoughts to support this theory. Given the proliferation of GIS within university programs for various professional and academic fields, reliance on a GIS technician isolated in an office separate from professional and field staff will diminish over the next decade for three reasons: GIS-related workload will increase faster than GIS-specific staff can be funded and hired; staff already familiar will GIS will be frustrated with having to request work from an overloaded GIS office; and geographical intelligence is too essential to decision-making to keep it locked away.

Despite what I believe is an obvious need for ArcGIS to become as commonly deployed and used as Microsoft Access, if not Microsoft Word, GIS applications continue to present a steep learning curve to those unfamiliar. Devotion to formal cartography and accuracy and precision in data and process require a level of training beyond simply dabbling in GIS. Licensing costs for software are high, or licenses controlled by a GIS office. GIS professionals may have both deserved and unreasonable protective attitudes regarding their earned training, skills and status as the keepers of geographical data, its analysis and its cartographic output. How would the industry get beyond these barriers?

Gao and Wang (2020) mention one approach in their discussion about university librarians supporting the GIS needs of students and staff. While a GIS-trained librarian is for some schools a specialty occupation, funding priorities and staff turnover make this arrangement less than resilient. The authors describe meeting the need instead by having GIS training become more accessible to a wider group of librarians (who may already study Python coding as part of library science, perhaps giving them a leg up to get beyond basic GIS operations).

Gelfert (2019) outlines some of the more specialized roles that will continue to be done by GIS professionals. While that article does not acknowledge my main point of wider GIS accessibility, it does reinforce my point that additional specialization will be required for anyone hoping to remain in a GIS-specific role.

Yes, some “Data Entry Clerk” and “Word Processor” positions have survived mobile apps and RFID tags for data collection, the desktop computer and email for written communication, desktop databases and cloud-based reporting systems for data analysis. Yet most of the skills that formerly required special training have been rolled into the requirements or assumed skillset of other positions. These relics of a previous era may provide a steady income and good working environment but are not high-paying, nor do they offer much in the way of advancement opportunities absent the higher-level information-technology training and experience required to . If my theory about the future of GIS is correct, the basic GIS technician role will be the same in the near future. This leaves the open two routes for persons with GIS training: Fit that ability into a job that was historically only a consumer of GIS product rather than a GIS user, or obtain the database-administration, coding or other higher-level IT training necessary to remain in a GIS-specific role when everyone else in the office is able to do the basics.

Jessup and Lenzi (2007) illustrate my point for my current industry: maintenance of county-road networks. Since that paper was written, the state agency at the root of their study has purchased a new asset-management system based on ArcGIS for maintenance of the official records of county roads throughout the state. The system is set for deployment in 2021 and will replace a dated, text-only database. The manager of that system at each of the state’s counties will be forced from a textual environment into a GIS-based environment, one more way in which GIS escapes isolation and pervades operations and management.

Perhaps a Microsoft Excel level of penetration should be the goal. Some people cannot understand a formula but use a single worksheet to make something they need. Others understand just enough about how a spreadsheet is supposed to function to mess it all up, but at least they do it on their own machine and their own copy of the data. Some think they know a lot about it and end up creating something so complex it should be in a database, not a spreadsheet. Others can do robust analysis in multiple worksheets with complex formulae for which the spreadsheet is the best tool. Regardless, everyone has it on their desk.

Gao, W., & Wang, Y. (2020). The Provision and Sustainability of GIS Services: How an Academic Library without a GIS Specialist Provides GIS Services. International Journal of Librarianship, 5 (1), 53-60. https://doi.org/10.23974/ijol.2020.vol5.1.160

Gelfert, A. (2019, February 25). GIS Job Titles of the Future… Retrieved December 07, 2020, from https://community.esri.com/t5/arcgis-enterprise-questions/gis-job-titles-of-the-future/m-p/305819

Jessup, E., & Lenzi, J. (2007, March). Washington State All-Weather Road GIS Mapping: Improving Statewide Freight Flows and Connectivity (Tech.). doi:10.22004/ag.econ.207827

Noise Pollution in Bishop City Park

Introduction

Bishop City Park is a small property minimally developed for recreational use. It consists of several parcels owned by the City of Port Townsend, Washington, as well as portions of adjacent public right of way. The use of public right of way as multipurpose trails where they remain unopened and undeveloped is common throughout the City.

My investigation was as to whether this property, which could generously be measured at 5 acres including those portions of the adjacent right of way through which trails run, provides any significant respite from nearby streets and developed parcels. The property’s primary topographic feature is a small ravine that is usually free of running water; this is indistinguishable on topographic maps due to its small size but does offer a small measure of apparent seclusion due to its depth below the surrounding roads and properties. Nearly all of the property is wooded with second-growth timber except where utility lines have been constructed in the various right-of-way parcels.

Methods

I began my investigation by mapping the Bishop City Park property and surrounding parcels in ArcGIS Pro. Most map layers were sourced from Jefferson County GIS via their REST Services. I used (but did not ultimately include on the map due to a lack of data within the small study area) hydrology, utility-line, contour-line and shaded-relief layers obtained from Washington Department of Natural Resources. Structure polygons were obtained from Microsoft Building Footprints data. Creating the base map in advance of field work allowed me to export the map as a georeferenced PDF. I opened that PDF in the Avenza app on my smartphone so I could use it to plot noise measurements.

On the same smartphone, I installed the Sound Meter app. Lacking the equipment necessary to verify the calibration and performance of this app, I tested it indoors and verified that ambient indoor noise approximated the app’s rating in decibels (dB) for “quiet library” and that my speaking approximated the app’s rating for “conversation.”

A Garmin eTrex 20x handheld GPS receiver functioned as my source of coordinates for each noise measurement. While the Avenza app recorded similar information, I judged the handheld device to be a more accurate source of location data than the smartphone’s built-in GPS receiver. The GPS receiver recorded a track log as well as the individual waypoints I created; I later used this track log to add supplemental trails to the base map where the county’s trail data fell short (I did not, however, attempt to reshape the trails included in the county’s data that are quite obviously based on poor sketches).

On foot, I traveled each of the primary trails within the greater park. Using the base map in Avenza to approximate equal spacing and focusing on key points within the small trail network, I paused at 14 locations, creating a waypoint in the GPS receiver and an identically named marker in Avenza. In the Avenza marker, I recorded the average noise rating in decibels from approximately 30 seconds of observation with the Sound Meter app.

Data from the GPS receiver was extracted using DNR GPS. Avenza data were exported from Avenza. The waypoints and track were imported into ArcGIS, the two sets of waypoints joined based on their name fields to form a single table, and the track edited to show supplemental trails as described above. GPS coordinates in WGS84 were projected on the fly by ArcGIS Pro 2.5 onto the base map, which used a NAD 1983 HARN state-plane projection.

Results

The average noise rating recorded at the measurement sites varied from 42dB to 57dB. The highest reading was taken nearest the state highway that borders the park property to the south. This was not unexpected due to the apparently higher traffic volume on that road than on any other surrounding street. The second nearest point to the highway average only 46dB, likely due to its location at the bottom of the small ravine. The third nearest averaged 50dB, as its exposure above the bottom of the ravine provided a more direct line of sight and sound to the highway.

The mapped results show the decibel reading at each measurement point.
The mapped results show the decibel reading at each measurement point.

The remaining measurement points showed generally decreasing noise readings further north and west, away from the highway. It should be noted that no traffic was present on 9th Street at the time of the closest readings to that northern border of the greater park, though the street is used by some through traffic, though at lower volumes and generally lower speeds than the highway on the south end of the park.

No major sources of noise were noted during any one measurement that were not noted during others. Some noise from passing airplanes was present at times, though distant. Birds and other wildlife were noisier lower in the ravine than in most other places, but the measurement app did not seem sensitive enough to include this in the average rating. All measurements were made just before and just after 5 PM on a Thursday evening, timed for the evening commute in an attempt to maximize the amount of noise on surrounding roads and, thus, to highlight any attenuation of the park property.

Conclusion

While this small, little used neighborhood park provides some visual respite from the surrounding streets, its benefit on the ears is limited. The sound of traffic is a constant companion along the trails, especially in areas close to the state highway and approaching it, where the ravine appears to channel noise from the highway through the forested park. However, the short arms of the trail system that reach north and west from the main park parcel are somewhat sheltered from this effect and provide something more peaceful, if not silence.

Increasing ethics awareness in GIS

As with other nascent fields of study and more established scientific endeavors forced to confront a history of unethical application, geospatial science has had to reckon with its potential for misuse. As a technical field wherein one can obtain skill, training and qualification through various routes other than a undergraduate or graduate degree, there is a high potential for GIS professionals to enter the industry never having considered or even learned to recognize the ethical and privacy challenges they will face.

Most of the written material on the topic focus on the most basic component of the problem and its solution: Introducing GIS students to issues of privacy and ethics. DiBase et al. (2009) suggest that the “…rich literature in GIS and Society and Critical GIS is more useful to students and instructors in academic programs than those in professional programs” which “produce practitioners rather than scholars.” One could count certificate programs such as the I in which I am a student among educational programs with the potential to yield such professionals.

The authors present a solution in The GIS Professional Ethics Project. They propose a series of GIS-related case studies for use in ethics education for GIS professionals. The case studies may be found currently at www.e-education.psu.edu/research/projects/gisethicsproducts. Each case study is a one-page summary of the case, based closely on actual events, followed by references or related reading.

The authors describe the seven-step model by Davis (1999) for ethical decision making and propose it be used by GIS professionals or students to consider and discuss each case study.  Using the first case study posted above, they present an example analysis using the seven-step model. The facts and implications and options are systematically described, and while a “right” answer is suggested, the focus is less on that ultimate choice than on the process of evaluating the problem. They suggest using this method in GIS-related education with the remaining cases to prepare “students to analyze ethical problems rationally and to respond with integrity.”

I agree with the problem as they describe it. It is quite possible to learn the mechanics of a technical field without ever considering the ethical, legal, moral and privacy-related issues inherent in the field. This problem is likely to be especially profound among those completing non-degree programs that not make space for consideration of these issues (Elwood and Wilson, 2017),  second to those who stumble into working as a GIS professional without the benefit of any guidance from professional educators.

I also agree with the solution proposed here. It has become evident in my career in local government in large and small agencies that people do not learn ethical behavior from ethics class; they learn by reading in the newspaper about the mistakes made by coworkers and colleagues. Similarly, GIS professionals will best learn to handle ethical conundrums in their work by studying the real-world decisions faced by other GIS professionals. This process can help a GIS professional first recognize such situations and secondly analyze and respond to them. The format appears well suited for GIS students in degree and certificate programs as well as working professionals.

I find little with which to disagree but am somewhat disheartened by the facts that the original domain name obtained for the project has been allowed to languish and that the published cases number only sixteen given the preeminence with which I believe the issue should be treated.

Davis, M. (1999) Ethics and the University. London: Routledge.

DiBiase, D., Goranson, C., Harvey, F., & Wright, D. (2009). The GIS Professional Ethics Project: Practical Ethics Education for GIS Pros. Proceedings of the 24th International Cartography Conference. Retrieved August 30, 2020, from https://www.e-education.psu.edu/sites/default/files/ethics/DiBiase_et_al_GIS_Pro_Ethics_ICC2009.pdf

Elwood, S., & Wilson, M. (2017). Critical GIS pedagogies beyond ‘Week 10: Ethics’. International Journal of Geographical Information Science, 31(10), 2098-2116. doi:10.1080/13658816.2017.1334892

Mapping voters for political campaigns

Political campaigns, especially on the local level, rely on personal outreach to voters. This can enhance a candidate’s reputation with a voter, give the voter a chance to get answers to questions otherwise left unanswered and remind the voter to participate in an upcoming election. One could hope that, at its best, such personal contact would also result in the candidate learning something of a voter’s hopes, problems and needs, better informing the candidate’s work for the people if elected.

Due to the ongoing pandemic, 2020 was not an ideal election season for such a personal connection with voters. The ground game continued, though it required masks and involved more quiet deposition of printed campaign materials than door knocking.

To aid in such an effort, I produced a map book that gave a candidate and campaign volunteers printed access to current address and voter data. Publicly provided GIS data from local agencies provided the basics: roads, parcels and addresses. I added structure data published by Microsoft to make visible the position of the residence within each parcel, important to help find the home on large rural parcels and multiple homes on parcels with ADUs.

I joined the state’s latest published voter registration data to the address data to highlight those addresses with registered voters. This could be used in an area full of part-time residents to focus efforts on registered voters. However, it could just as easily be used to encourage others to register in time for the election.

The printed product met the immediate need in a short period of time, but in the future I rather present such data using a mobile app such as Esri Field Maps. This would eliminate issues of access to a tabloid printer, loss of detail during reproduction and the desire for different scales to capture adequate detail in both urban (large scale required to include closely packed homes and condos) and rural (only a small scale needed) areas. Such an app could also help a campaign track its coverage of each residence or voter.