NACIS 2016 Presentation
Eben Dennis, Icon Engineering, Inc.
Robert Soden, University of Colorado
Online, interactive web-maps are rapidly becoming important tools in the communication of flood hazards and the visualization of risk information more generally. Although there is a wealth of study and information available guiding the design of traditional cartographic products, there is comparatively little research available that is solely focused on web-mapping formats. This presentation will discuss initial findings of ongoing research being conducted in partnership between Icon Engineering and the University of Colorado, Boulder. Outcomes of the work will lead towards the development of design heuristics and best practices for visualizing flood data through interactive web-mapping products.
2. Who We Are
Eben Dennis
ICON Engineering, GIS Coordinator
Civil Engineering firm
FEMA Mapping Partner
Primarily Public Sector projects (UDFCD,
Boulder, Denver, Greeley, Ft Collins)
Certified Floodplain Managers
UC Denver Geography Program Graduate
Robert Soden
University of Colorado, Boulder, PhD Student
Co-Risk Labs, Principal
California-based research and design
company
Previous Background
World Bank Global Facility for Disaster
Reduction and Recovery
3. Why Are We Doing This?
Web-based flood maps are mostly hot, flaming piles of
Web-based flood maps need to be
improved.
(NACIS is nicest NACIS is nicest NACIS is nicest…)
4. User Centered Design
Specify the context of use
Specify requirements
Create design solutions
Evaluate designs
From usability.gov
5. Cartographic
Interaction
Roth & Cartographic Interaction
“How maps are manipulated by the
map user”
Contrasts with the classic
Communication Model
Map & user influence each other
Speed of application can affect
productivity and focus
8. Public Facing Applications
Typical Problems:
Too much focus on the map itself
Information Overload
Solutions:
Treat your data more like a search engine,
encourage text-based exploration.
Focused, single-topic applications; accomodate
the small number of power users by making
data available for download; provide custom
basemaps with some data already built in.
9. Competitor Analysis
Analyzed 25 different flood mapping
sites
Nearly all government
owned/operated
What interaction operations are
supported?
How is flood risk represented?
Clear Statement of Purpose
Clear Branding/Ownership
Disclaimers
Address Lookup
Dictionary/Links to Further Info
Guidance for Property in the Floodplain
Base Map Options
10. Competitor Analysis
Almost everyone using ESRI stacks
Good news is that most sites had address lookups
Bad news is that almost none of the sites tell you what to do if, in fact, your
home is in the floodplain.
Who do I actually talk to about buying it? What if my house is elevated? What
the heck is an elevation certificate? These are all important pieces of
information that users want to know in this context.
12. Competitor
Analysis
GIS on the internet
Too many buttons that have no
intuitive meaning to non-GIS
users
No clear purpose
Slow load times, near impossible
to keep user interest
14. Competitor
Analysis
Single purpose application
Limited interactions
Plenty of opportunity for self-
directed learning by users
Not too text heavy, but does a
good job of providing links to
users who are interested in
learning more
15. Interviews
Conducted interviews with floodplain engineers, clients, and end users to help
determine the utility of the applications.
Two audiences were identified:
Public - typical homeowner, wants to know if they are in the floodplain
Expert - engineers and clients utilizing maps as a data immersion tool
Decisions made with this application will potentially affect the public user
16. Walkthrough
Created web-map application for a
public audience
Single purpose of the application is
to identify flood hazards in Boulder,
CO
Had public users think aloud while
they used the application
Built on Mapbox GL and Turf.js (fast!)
Consists of search bar, map panel, info panel
Limited interactions
Flood information is provided for users in and
out of the floodplain
Additional flooding information is provided in a
custom basemap
18. Walkthrough - Findings
People used the search bar, overwhelmingly - though some were confused by
autocomplete.
Almost as overwhelmingly, they don’t read text
Simple, actionable instructions - what is the key information the public needs at
each stage of the project?
19. Walkthrough - Findings
The NFIP 100-year floodplain presents a limited portrayals of flood hazard.
People were genuinely surprised when they realized that areas outside the
floodplain were affected by the 2013 Colorado floods.
In making these sites simple to use, we need to be careful to not convey
simplistic understanding of science - when we added the 2013 flood footprint
and it created just enough dissonance for people to get engaged and start
asking interesting questions
20. Survey
For the expert audience, we utilized two existing applications built by ICON
Engineering.
All of the questions on the survey were open-ended, with the exception of
difficulty ratings, in order to better simulate a real-world decision making
process.
The audience is wholly comprised of CASFM YMG members, so the
expectation is they will be comfortable with web-map applications and have the
ability to accurately answer technical questions.
23. Survey -
Application 1
Too many clickable elements
Alternatives aren’t easily
distinguished
Flow direction arrows extremely
useful
Speed, clickable information,
layer toggles? All good things.
Missing statement of purpose,
lacking instructions, more
description needed for layers
25. Survey -
Application 2
Transparency on transparency
isn’t the most effective
Layer switching not as intuitive
as toggles/checkboxes
Speed? Still a good thing.
Needs a better description of
what each layer represents
When multiple elements are
clickable, more clarity for what
layer fired the popup
27. Next Steps
Better in-page analytics
Click tracking
Time spent on page
More data!
Gather more user feedback on the two expert systems
How much data can one engineer take?
Iterate!
Web-based flood maps and web-based maps in general are mostly garbage.
The tendency to create a GIS in the browser is a massive pitfall for both the public and experts attempting to utilize these applications.Web-based mapping is still a relatively new field, and cartographic/design standards are not widespread if they exist at all.
The lowered barrier to access for these tools and applications must also be taken into account, and we can’t necessarily expect for either the architect or the user to be trained in visual communication. These tools are great and powerful and have so many upsides to them, but it also opens the door for a lot of junk to be put out for consumption.
These applications have clear potential, but the design standards still need to be developed to provide best practices for effective communication and end-user suitability.
The basis for a lot of this study is the need to incorporate UCD in the application design process, which for many of us that finished our formal education before the real rise in web-based mapping we’ve seen in the last few years, this can be a foreign concept.
Specify the context of use: Identify the people who will use the product, what they will use it for, and under what conditions they will use it. Who is your audience? And how are they using this application? And to what purpose?
Specify requirements: Identify any business requirements or user goals that must be met for the product to be successful.
Create design solutions: This part of the process may be done in stages, building from a rough concept to a complete design.
Evaluate designs: Evaluation - ideally through usability testing with actual users - is as integral as quality testing is to good software development. And then you identify areas for improvement based on this testing and you iterate.
Speed - you have a limited window where an action feels instantaneous. After that, if there’s a bit of a delay, it can be annoying, but the loser won’t lose focus. But if there’s too long of a load time, it can seriously impact workflow, understanding, and usability.
The gulf of execution describes the disconnect between the user’s objectives and the provided cartographic interaction operators.
The gulf of evaluation describes the disconnect between what the user expected to accomplish through the cartographic interaction and the interface’s representation of the result of the cartographic interaction
Communication model where users are more of a passive participant, here they’re an active part of the application - how it gets used, what functionality is expected. They are the ones driving, so we have to take into account their experience and ability. With classic, static-map cartography, you can’t click the image or the print, here, the user can change the state of the application.
Many tools included in out-of-the-box web applications look and act like a desktop GIS, which is great if you’re a trained GIS user, not so much for the other 98% of your audience.If you want a layer to contain information on a click, why would you make a user click on an ‘Info’ button first?
Too much focus on the map? Cartographers are designing this after all.
Make the search and exploration functionality explicit.
Numbers = Science!
Platonic ideal of GIS in the browser
These buttons hold little intuitive meaning to vast majority of users
Added contextual data and a bit of cognitive dissonance really engaged the users, they started asking more questions when confronted with the fact that modelled data didn’t match up with real world evidence.
We’re at the mercy of the engineers a lot of the time when building these applications, even if it’s not the engineer themselves building the application. They’re the internal/external clients here, so they get what they want. So in instances like earlier when Lyzi was asking about ‘how do we get people to accept that there are better ways to do things’, this is one way. Use the applications that they specified, and test them.
Finalize and write up the results of this study
Gather user feedback on two expert systems once they are fully launched
Better analytics - we need data