00:02Welcome to Map Makeovers - How to Make Your Map Great.

00:03My name is Charlie Frye, I'm the chief cartographer at Esri, and with me is one of my colleagues, Jim Herries...

00:08...who is a project manager, geographer, cartographer, data expert on my team.

00:13And a few other things that you'll see come out in today's talk.

00:18So first, I'd like to welcome you to the second half of the Technical Users Conference.

00:21As you've gone so far, we're now - we just passed the halfway point...

00:27...and so hopefully everything has been going well for you...

00:31...and then we'll try to keep things on that same track throughout this session.

00:37So what we were thinking with this session is that we wanted to take some maps that we had either...

00:42...worked on or intercepted kind of halfway to getting finished and share some of how we thought about them.

00:51Make sure my PowerPoint's working here; there we go.

00:59And the idea was is that what we do in-house is that we offer a critique...

01:04...and the idea if somebody is making a map and they're not a cartographer...

01:09...in terms of their background, we want our team to take a look at them...

01:12...and get a sense for what could we do in five minutes, ten minutes, couple days to fix the map.

01:20And we want to give you kind of an insider's look under the thinking we use when we do that.

01:25And so what I've started this out with was a little section on how we think cartographically.

01:31And this is more opinion and experience rather than kind of textbook...

01:34...knowledge and the kind of the academic distillation of cartographic thinking.

01:40The purpose is to give you all some confidence to do this same kind of thing...

01:44...or build your confidence if you are already doing this same sort of thing.

01:50So when I'm doing some of this sort of critique is that I look at maps holistically.

01:57I'm looking for a geographic rationale for why you're making the map...

02:01...in other words, as a geographer I'm looking for, is there something that's being informed by geographic thinking?

02:08So you know why is what where, what's the rationale, is it something that's insightful that you're seeing in the map.

02:18As a designer, my background was architecture before I got into geography...

02:23...and so one of the things we learned in there was, as we learned geography, was that...

02:28...design by pulling things out until the design breaks.

02:31And then put whatever it was back in, and that's a good place to start thinking about a minimalist design.

02:38For cartographers, we think of what we do in terms of communicating graphically.

02:44So putting, we call, graphic marks on a map and those are the things that communicate to people.

02:49So the idea is every graphic mark should do something on the map...

02:53...and if it's on there and it's not doing anything useful or contributing, it could probably be dropped off.

02:59Readability or legibility is a big thing for what we look at in our maps, particularly for web maps.

03:05The idea is you should be able to look at that map and immediately see the most important thing...

03:10...read it, understand it, there's no extra work in trying to move the map in or out...

03:15...or spend time reading the legend or trying to translate the title to what's actually going on.

03:23So all of those things should just happen for you.

03:27One of the things that's kind of interesting I've put down here is that, and we talk about this...

03:31...in terms of getting maps reviewed, is having good judgment about what needs to be on the map...

03:35...what the concept of the map is.

03:38We started to talk about how we do our maps in terms of specifications as if the map is a product.

03:45And what we would do is literally have a top-down summary of what the map's about...

03:48...like an executive summary, like you'd put together for any other product specification.

03:52So you know, what does the map do, who is it for, and get very specific about those things.

03:57What that does then is it allows us as we're going through the mapmaking processing to go back to that...

04:03...and ensure we're well grounded as we go along.

04:08It's also good when somebody else sees that map for the first time and says something similar about that map.

04:13So if you're making a map of - we have a map of radiation that we'll show you in a little bit...

04:19...that if somebody looks at that map and says something pretty close to what's in the executive summary...

04:25...after about 30 seconds of looking at that map.

04:29One of the things that's a little harder to do, and it's a skill we try to encourage on, for folks on our team to do...

04:35...is to be blatantly objective, so not just take a step back, you know, like let something sit for a few days...

04:40...but actually be ruthless about it. We'll show you some examples of that.

04:46Clarity is another thing. Last year, for the 60 percent of you who were here maybe last year...

04:52... we had a keynote speaker name Saul Wurman, who's - I guess he's a designer and an information architect...

04:58...and about a half dozen or a dozen other things, and his big thing was clarity.

05:03And he distinguishes clarity from simplicity in a nice way...

05:07...where he looks at simplicity as something that might be dumbing something down...

05:12...so moving material to a less educated audience...

05:15...whereas clarity is making something easy for the audience that you intend to communicate to, to understand.

05:25And kind of the bottom quote down here is that I don't like to think about some of this as cartography so much as...

05:32...and one of the things I've done in the last six months or so is stop using the word map so much around Esri.

05:38What we've found or I find and a few others who have been doing the same thing...

05:41...is that maps are cheap and free and easy and people don't have a high opinion of them...

05:47...but when you talk about an information product...

05:51...just the idea of that sounding more important engages the rest of the organization in an interesting way.

05:57Our marketing people feel like they have a vested interest then, so there's information product...

06:01...okay, well, I could care about the information, and I have a job because it's a product.

06:06So just turning that kind of thing around and getting people out of their usual pattern...

06:11...helps everybody, I think, do a better job of producing maps.

06:19So what all of this really boils down to is that your map should tell a story that your audience can easily interpret...

06:27...and relate to, so it's really about kind of accessibility to the message...

06:32...and that the message works in as an efficient way as possible.

06:43I guess where I learned cartography, I was at Kansas State...

06:47...and one of the things that they really emphasized was communication first.

06:52In other words, whatever you're doing, you're doing it for a reason, to pass some information along.

06:59With web maps in particular, overtly communicating is important.

07:03It's okay to make 28-point text if you need to to get somebody to read the two most important words about a map...

07:09...and put it front and center and then they'll figure the rest out.

07:18What I feel like is that when we do a lot of our maps and we iterate on them is that...

07:22...the more time we spend doing that and having people just look at what is their first impression...

07:30...and evaluating that kind of thing that we do a lot better in the long run...

07:34...for how well that map does when we finally do give it to the intended audience.

07:44I can't remember where I was, I was trying to remember who was telling me about self-evidence in communication...

07:47...but basically what it boils down to is that obvious stuff should be obviously right...

07:52...so blue for water, green for vegetation, and so on.

07:57Sometimes we get kind of lost in the process of being creative and trying to create something new and innovative...

08:03...when making sure it works first is most important.

08:09And I guess back to the middle bullets.

08:12One of the things that is interesting about making maps or doing anything creative is it's hard to force it.

08:19The bit about inspiration there is that we do a lot of maps after maybe years of sitting them on the table...

08:26...because we didn't quite figure out what we needed or we didn't have the data that we needed...

08:29...the right way rather than trying to force something out the door because somebody was asking for it...

08:35...or some other external influence.

08:37But at the same time there's days when you do need to make your inspiration happen...

08:41...or at least some portion of it to get a project done, and that's where the real work in mapping ends up being.

08:49So to do these makeovers, we're going to go through three different maps.

08:54I'm going to talk about the first couple, then Jim's going to talk about the food deserts map.

09:04So the history behind this is that shortly after the earthquake and tsunami in Japan...

09:09...we had a group of people at Esri who were asked to work on a map to show what was potentially...

09:15...happening with the radiation that was reported as being emanating from the nuclear power plants...

09:24...and in particular, they were looking at the Fukushima Daiichi plant.

09:28That team was also - had kind of an ulterior motive, which was that they had been creating...

09:34...a new analysis method in ArcGIS for empirical Bayesian kriging.

09:39And so this on the screen is what they initially sent me as can you please take a look at this.

09:44We know the cartography is not too great, you know, give some suggestions to help us out.

09:50And so what I've initially looked at was, what's the story of this map?

09:54I mean if you look at it, I know where it is now thanks to the title, but I don't know what I'm looking at.

10:03What are we empirically Bayesian kriging on this map? I don't understand that.

10:09I know there's a date on there, and the legend gets into a little bit of the story...

10:13...but at the same time I'm also looking at the colors, I'm trying to figure out, what am I supposed to feel about this map?

10:21Basically, it doesn't work very well, but you know, they're not cartographers producing it, they're analysts...

10:25...and so we actually started way back in the early part of what they were thinking about as a result.

10:35So the map turned into something that looks a bit like this.

10:40One of the things that we realized is that the map did need to tell a story about what we were analyzing.

10:45As it turned out, we were analyzing predicted ionizing radiation levels, so this is a certain level of radiation being measured.

10:54You can even see some of, if you go back to the before here, look at the shape of the analysis.

11:04The analysts were actually learning a bit about the data as we went through some of the process.

11:07You could see the shape of what was the hot spot in this area so to speak; it took on a different character.

11:16Now let me go through a bit of what we did to get from the beginning to this point.

11:23So the big things that changed are the colors, then the map elements and a lot of the wording...

11:28...and what you would read about this map.

11:29So in other words, how we told the story.

11:34So the original colors, I'm looking at this, I think it's...

11:39...I don't know how to say maybe it's popular to choose a full spectrum color ramp...

11:44...these days with heat maps and so on, and at the same time that's a bit loud in a lot of cases...

11:49...so the team was trying to do something like make them transparent so at least it's not loud.

11:54But I'm not understanding what's going on, and actually you can't see it very well here, but the colors of the points...

12:00...were also given the same color scheme, and what was happening...

12:03...is that points were different colors than what the color bands were showing.

12:08And so that's one of the things that I pushed back, like, okay, I'm not buying your analysis model here...

12:13...because there's clearly points in different classes and different color bands; what's going on with that?

12:18So you don't want to present something that's confusing to people.

12:21So that gets into that kind of legibility and clarity. It's like, okay, if you look at a map...

12:25... and you see immediately conflicting information, that's the sort of thing that it's easy to be kind of blatantly objective about.

12:33And you can't let that kind of stuff slip by as you review maps.

12:39Alright, so the final color scheme, this is one of the earlier suggestions...

12:42...was, okay, this is something that's measuring intensity.

12:44And one of the basic cartographic principles for showing this kind of information is that intensity is best understood...

12:50...through a value change, not a color change.

12:53So they started using a single color scheme and then they also started to add a few other things to the map.

13:00One of them was these distance bands, so that you could see distance...

13:05...and make sure that you weren't confusing distance with color change.

13:09And this represents, this map overall represents kind of a middle ground in the process...

13:16...so we wanted to see what else was going on with this.

13:20And there are a few other things that happened on this map.

13:22One is that they started adding some city points, they added some political boundaries to give better context...

13:29...we added an error map instead of just a legend for the error data.

13:36A good title makes a big difference.

13:38I see a lot of titles where it's either the place-name or it's the data provider.

13:45Both of those have different places to go on the map than the title in most cases.

13:50In this case we're mixing a place-name with an analysis method.

13:54The only thing that was really good about the title was the date, and that's an important thing to have on a map like this.

14:01And one of the things about this sort of map is that we're presenting actually some pretty complicated information...

14:06...to an audience that, odds are, didn't have a college degree that included the word nuclear.

14:12So the difficulty here is, you know, how do we measure radiation?

14:15Turns out that's actually really complicated too.

14:18There're several mainstream methods, or measurement types, and even those vary quite a bit.

14:30So the first attempt was, okay, make it clearer by removing the Bayesian kriging from the title; that didn't help much either.

14:39We finally had a geographer get involved on the analysis team...

14:42...and she took this from the standpoint of, well, what was she actually analyzing?

14:49And that turns into a much stronger title, and this is, I think, close to the final title.

14:53I think we may have picked a different day of values to analyze.

15:01Actually, let me go back here.

15:02So one other piece to this that we'll cover a little bit later is, looking at this title, you know...

15:08...how many of you would actually know what that means?

15:15So maybe 15 or 20 percent of you are feeling somewhat comfortable; nobody like, I know what that means.

15:23So keep that in mind for a little while.

15:26So back to the color schemes, the nice piece of art was a comment of somebody made looking at the map...

15:33... it's like, yeah, it looks pretty, but I don't understand it, and the dots didn't match.

15:38I think that was something I brought up earlier.

15:41Then there were some things that as we looked at the next kind of level of changes that the team made...

15:48...does all of this make sense? So one of the things that we were looking at was the hillshade.

15:51And it turns out that for whatever reasons, whether you're measuring values, just made the hillshade...

15:57...look like it was showing the values at being out of context.

16:02So we were pretty sure that the government was actually and the people doing the measurement...

16:05...were actually recording accurate measurements.

16:07But they had the hillshade there making it seem a little out of place, 'cause sometimes you get different measurements...

16:14Actually added some confusion to the story.

16:16So by taking the hillshade out, that kept people from having to think about one additional piece of information.

16:24The rings actually ended up getting labeled after a while...

16:26...that was these were five-kilometer rings coming out from around the Fukushima Daiichi reactor.

16:40And so we changed things to this kind of arrangement...

16:44...so now each of the rings was labeled like I was saying.

16:47We stopped showing the additional reactor that wasn't having a problem. It was really a story about this reactor.

16:53So again, just focusing the map on what was actually needed.

16:59Alright, so back to the original map elements, so this was the original legend, it doesn't tell you a whole lot.

17:07One of my first comments was that there's - in parentheses right up over here, here in the legend there is a unit...

17:20...and it's USV per H, and so that was not very helpful to me, being the educated layperson without a nuclear degree...

17:32...and so I asked what that was. Nobody on the team actually knew a lot about it so we started doing some investigating...

17:40...and the team actually did a credible job of educating themselves, so these are ultra-sieverts.

17:46It's one of the most common measuring systems for radiation.

17:50The difficulty with ultra-sieverts is there's also millisieverts, and then you'll often see variables expressed...

17:56...per second, per hour, per week, per month, and per year...

17:59...in giving you a sense for how much radiation happens with different things.

18:07And what I wanted to see in this legend was, okay, well, tell me what's normal radiation.

18:12Tell me what radiation I should worry about for, you know...

18:17...if I have a predilection for cancer in my family, what should I be worried about?

18:24And it turned out that that was even debatable; I had a friend of mine who was a nuclear engineer...

18:30...he said don't worry about it, radiation coming over from Japan. Well, fine.

18:34Another person on the team, actually his father was a, and still is actually...

18:39...a doctor who specializes in nuclear medicine, and he had been studying in Japan for decades.

18:45And so he was actually quite helpful in giving us some additional perspective.

18:48So the lesson there is be honest with yourself and that kind of blatant objectivity.

18:55When you don't know something, don't keep going; go back and learn what you need to know.

19:04So part of the analytical method, and actually it's something that's useful for understanding the map, is the error map...

19:12...or the error that can be produced from this Bayesian kriging method.

19:19And so what we ended up deciding to do is just put a map of that on as an inset on here.

19:24So that way, you could easily compare to know where we were most confident of our analysis.

19:33Another question we had, and we decided we weren't going to worry so much about it was that we didn't make a locator map...

19:39...well, yes, we did a much smaller locator map of the island of Japan...

19:45...just to show where we were looking. Even though we were kind of debating, most people were seeing this area in the news.

19:49We're now familiar with this shape of the coastline, and we could've said, okay, well, maybe we drop that.

19:55But in the end, we decided that more people than were watching the news might see this map and it'd be useful to have it.

20:05You can see that small locator map there. And we also added a lot of text to the map.

20:11Part of it was that we wanted to explain what we were doing...

20:16...so how did we analyze it, and then we also needed to explain what we analyzed.

20:21So back to that ionizing radiation thing.

20:25And you can also see the legend took on a lot more text in terms of its explanatory capabilities.

20:33So one of the things about this map is it is actually a complicated topic, so it's okay to put enough text on it.

20:41However, this was kind of a first attempt at doing that, and this gets into that legibility and readability thing.

20:48If you're looking at this block of text, it's really pretty difficult to follow.

20:54It's a little bit wide for what I would want to see on a map.

20:58So what I draw from that is, look at most newspapers, you get a narrow column of information.

21:03And the idea is that you're able to scan down that and read. It's sort of like accelerated speed reading.

21:10In speed reading, you're scanning down the page in a book...

21:11...and your eyes are just pulling off to the sides a little bit to pick up extra words.

21:16This is a little too much to do that with, so what we want to do is just clean that up so it's got a little bit more hierarchy to it.

21:22And you'd want to know, depending what you're interested in, where you could focus.

21:31So we changed that a bit, and also - but before...go back here once.

21:35This was all about the methodology, so there was nothing else.

21:39In this case, we introduced it with what predicted ionizing radiation was...

21:45...then cut the text in about half for just explaining the method...

21:49...and not worrying so much about the technology so much as what the value of the method was to this particular analysis...

21:54...and a little bit of what we did.

21:58We also toned the language down from PhD level to kind of a smart high school student or entry level college student level.

22:07So there's still some, as my drill sergeant in the Army once said, "20-dollar words" in there.

22:13But the idea is that most people can follow this and get something valuable out of it.

22:22So the legend shifted quite a bit, so we actually dropped the observed values for just being confusing and not contributing.

22:31We also had a shorter definition of ionizing radiation here. And we explained what other symbols were on the map.

22:38In both cases for the map's legend as well as the standard error, it turns out they were using ultra-sieverts per hour.

22:45So it was just making sure that everybody knew that that was what was going on.

22:51What was kind of interesting about this - and this just gets back to the story for this map...

22:57...is that, okay, what did we want to say when we showed this information?

23:00Should people be worried? Should they be scared? Should they be reassured?

23:07One of the things we learned about ionizing radiation was that...

23:11...the first two classes in this legend are actually normal values for everyday Japan; it just depends on where you are.

23:21And actually well into the second class is normal in Sweden. Or excuse me, into the third class is normal in Sweden.

23:26So it just depends on where you are in the world as to what normal radiation levels are.

23:32So that meant just really it was just the last three classes of radiation were things for people to be concerned about.

23:39So what we included in those descriptions was actually mandated levels of concern that the Japanese government had.

23:45So these were things that when you measured this level of radiation, that some kind of action was required.

23:53So we actually went out and found what those were.

23:55So that way it's not us just making up our classification or taking the default natural breaks value.

24:01It was actually going in and finding something meaningful, and making sure that we could communicate that to the audience.

24:16Alright, so the final result, the major things changed...

24:20...the colors went from a multihue scheme to a single hue scheme that showed intensity.

24:25I think what was interesting here is that only these three darkest values were areas of concern.

24:32And really just within 10 kilometers. And so you remember there was a 25-kilometer kind of cordon around the area.

24:43So that kind of showed you the second band here. So that got out to about there.

24:50So it showed that, okay, it was well intended, may not have been perfect, as you can see.

24:55There was probably ionizing radiation getting well beyond that, but not at a high level.

25:00So this wasn't going to be at a level that was going to be worrisome for medical problems.

25:05But at the same time, you know, you don't want people staying in that area for prolonged amounts of time.

25:12And actually just to give you an idea, if you flew here from Japan, you got almost as much radiation from that fourth class.

25:17So that's how little radiation actually was out...

25:22...but again, you don't fly for weeks and months at a time, so you're stacking up a lot of flight hours by living there.

25:34Alright, so this next map is kind of a mystery map, with the idea that - Do you have a question?

25:40[Audience question] Is there a reason why you chose not to put the north arrow on there? Is there [inaudible]?

25:45That's a good question. So he is asking if there was a reason why we chose not to put the north arrow on here.

25:50And for the most part, we were expecting that because we had the locator map on with Japan...

25:55...people would be able to orient themselves.

25:58I'm not a big fan of putting north arrows on maps when it should be obvious.

26:02Particularly when you get to a small enough scale, there's an assumption on the audience's part...

26:07...in this case where they are fairly well educated, that they could figure that out.

26:11So this wasn't intended to be a map that was going to go into, say, a fifth grade textbook...

26:16...which maybe I would say, maybe that's where a north arrow would have been helpful.

26:19This was something that, I call it a higher order news agency or magazine might put out...

26:25...so it wasn't as necessary to add some extra piece of graphics to the map. It's a good question though.

26:33[Audience question] Is there anyone else who would like to see the distance rings be concentric circles?

26:41That's a good question, which was, Is anyone else out here who would as you're looking at this...

26:47...feel like distance rings should be concentric circles rather than just arcs on the land?

26:55Nobody brought that up; that's a good thought. 'Cause if you're out looking...

26:59...for instance, if you're looking for people who were washed offshore from the tsunami, that's a consideration in that area.

27:10It's interesting, 'cause that depends on, you know, now - we were looking at historical data in this case...

27:16...but definitely if we were thinking about making this map within a day or two of getting these values...

27:21...that would have been an important consideration.

27:23[Audience question] [Inaudible] projection, you see a lot of maps that are [inaudible]. Should be.

27:31Right, and I'm not - So the concern that's being brought up is that are those rings circles which aren't equidistant...

27:39...or are they actually equidistant buffers?

27:42And I don't remember the answer. I can't remember if the projection supported having circular looking buffers.

27:52Alright, so the next map we're going to look at is a little bit of a mystery map to start with.

27:57And I'm doing that on purpose 'cause what I want to do is to see what your impression of the map is first.

28:04So this is before, and it goes back to that initial comment of a nice piece of art.

28:12And I'm thinking it does look a little bit like something Jackson Pollock might have done in his early days or something.

28:18But, so how many people have any idea what this might be?

28:26[Audience comment] Remote sensing map?

28:27Yeah, so I'm hearing things that we said around our hallway - land cover, remote sensing.

28:32[Audience comment] [Inaudible] map of rainfall or...

28:34Yeah, so I'm hearing like rainfall, and so far everybody is wrong.

28:42[Inaudible audience comment]

28:45So what we were looking at, what we're actually looking at here is soil moisture potential.

28:51So this is how much water, what the soil in these areas contains.

28:55The reporting units are soil unit polygons; the location is in the Corvallis, Oregon, area.

29:02It kind of points out in the US when we do soil surveys, we do them on a county basis...

29:07...where what can happen is that soil scientists in one county will do one thing...

29:11...and then the adjacent county will have another thing happen.

29:15And you can see that fairly clearly here. So one of the things we wanted to do is change the map a little bit...

29:22...so that we would go from something that had a little more hue change than what was, I think, useful.

29:29And what was happening is that because you had all that hue change it was looking...

29:33...most people in our hallway said, "It looks like a land cover map."

29:36Just 'cause they were expecting a lot of different classes of something going on, and they were certainly...

29:40...not seeing a sequence of colors indicating how well drained soils were.

29:45And so the idea is that we wanted two things with this color ramp, not just show you how well drained the soils were...

29:51...but we also wanted to show a little bit of which soils were going to be useful for agricultural production or holding vegetation.

29:58So the notion there is that the soils that are best drained are in green, and it doesn't come out real well on this projector...

30:05...so you're not seeing some of the yellows that are on top of hills...

30:07...that are going to be more aired where it's either running off or it's well drained.

30:12But it does look a lot better, and people, when they look at the new map, see that there is a sequence.

30:17So that was one of the fundamental things that was needing to be changed here. So the color scheme was a big deal.

30:24Initially, we didn't have any labels on the map.

30:26So there was no way to tell where it was or what - for instance, in Corvallis there is a lot of urbanized area as well...

30:33...which will change your interpretation of what's going on.

30:37What we were finding out is that this was done by a cartographer on my team, who was cranking out maps of soils...

30:43...and does a great job of picking colors most days.

30:45It's just he picked colors quickly on this one and nobody else got to see it until after the fact.

30:50And when we looked at it, was like, well, we should've reviewed that, so we went back and fixed it.

31:00So this is the original color scheme, and it looks fairly qualitative. You can see a lot of different hues...

31:06...and that's something that, like I said, people thought it was a land cover map looking at it.

31:12Even when they were familiar with the area, they still thought it was a land cover map.

31:19So by using a more sequential scheme, we're essentially using three hues, going from green to yellow to blue.

31:27And what begins to happen, particularly when you get out of this particular area...

31:30...where the soil scientists weren't in so much agreement...

31:34...is it does take on a pretty obvious sequential look to the map as well, not just to the color scheme in the legend.

31:51So one of the things we were thinking too was that, okay, in this map it's, you know...

31:53...without any other features on top of it, it's really difficult to know the significance of what you're seeing.

32:02And so by adding the additional information...

32:05...so for instance, what you're seeing here is we put our hydrographic reference overlay map on top of this.

32:11So what it's got in there are a little bit of cultural place-names and information...

32:15...but it's also emphasizing the rivers and streams as a network.

32:19And then the dashed and dotted - or excuse me - the dotted lines are watershed boundaries.

32:23So what you expect to see is that where the watershed boundaries are...

32:27...you generally have lighter colors where things are going to be more drained so there is less water in the soil...

32:33...and then as you get towards the rivers, you're going to see more water in the soil, and it's going to be less well drained.

32:45So one of the things that I'd say not enough people know...

32:48...but Esri does require that we do a review of all content we put out to the public.

32:53It includes maps. So like, for instance, on our blogs, we have somebody - there're two sets of eyes looking at everything.

32:59And the same thing is true for our maps.

33:01And that's one thing I'd encourage every organization to make sure they are doing is...

33:06...as you put things out, you could have a well-intended person produce a product that's not quite right.

33:13And you don't want to have to deal with the ramifications of having to do that.

33:22Alright, so there is the final result for that map with the colors on it, and then in general I think that it turned out quite well...

33:31...and it fit actually better with the family of soils maps that were being produced online as well at the time.

33:37So I'm going to go ahead and switch over to having Jim present the work he's done on the food desert maps now.

33:47Great, thank you, Charlie. Thank you, everyone, for coming.

33:52How many of you live in a food desert?

33:56How many of you know definitively what a food desert is?

34:01How many of you do not know definitively what a - there you go. Come on.

34:05Alright, good. The first step is acknowledgment.

34:09So in the Japan example, we had a very scientifically studied locational problem.

34:15Things emanating, people are here; get out of this area. Okay.

34:20And with soils we have a very interesting set of data, very detailed.

34:26This is something that touches on human geography, upon commerce, upon how we travel...

34:36...how we move about; access, if you will.

34:39This is the popular definition that you'll find if you do a literature review.

34:44Food desert is defined by areas anywhere in the world or in a city with limited access to affordable...

34:50...and nutritious food; it's frequently associated with health issues, like obesity or diabetes.

34:58Often it's associated with demographic characteristics, such as income or access to transportation.

35:06Not everyone owns a car. Sometimes competing destinations are included in the definition of what a food desert is.

35:12So it's not just that there isn't good food in a given area, it's that there's a concentration of other types of food.

35:21Clearly there is a behavioral choice.

35:22You can put me a thousand miles away from an In-N-Out and I'll find my way back there.

35:29It's not a problem, for me. I know what I'm choosing, I've passed 40, I've had the talk, I got it.

35:37I'm still going to have my In-N-Out, once in a while. Is my wife going to - no, she won't get access to this.

35:45Finally, access does not automatically result in good choices; we recognize that, this is not about that.

35:51So as you can tell, there's lots of room to define what a food desert is.

35:54So in the year and a half that I've been involved with this at Esri, I've been looking at, well, how important is it...

36:01...that whatever maps that get produced are relatable and they connect the user to a definition that has meaning?

36:10I like to use real-world examples to help create some understanding.

36:13So the map on the left you see...if I can get this mouse rolling.

36:18Let's say, theoretically, I live here. And let's say my two kids love going to this park.

36:23And, sorry, the type's a little small; I see I should have made that bigger.

36:26But the road we take to walk to this park is lined to here, and it says its 1.2 miles.

36:34And the crow fly distance you might get from drawing a ring using a popular software package like, I don't know, ArcMap...

36:42...is .6 miles. So let's think about this for a second - .6 miles is a very easy way to measure that distance...

36:49...and 1.2 miles is what I actually experience as a human being, if I want to get access to that park.

36:56So what's the error rate on that? .6 versus 1.2, it's a 100 percent error rate.

37:02So if I were to analyze every single house in this neighbor, calculating its access to that park as measured by distance...

37:11...I did it with crow fly distance, or I did it with network distance...

37:15...do you think distance might enter into a regression equation if I was trying to analyze some dependent variable...

37:21...like do people in this area use this park?

37:26It might behave a little differently if I'm using something that's a little more reality based.

37:31I don't get to the park, my kids would love to shoot me through a cannon to the park; I might sign up for it.

37:38And so in the case of food deserts, what I started looking at about a year ago is what's out there?

37:43And there's a lot of good small-scale maps out there showing you state or county level information...

37:49...the results of good analysis done by folks who are really interested in human geography and the factors that drive this topic.

38:00And the advantage of small-scale maps is they're easy to work with.

38:03There is a lot of good county data out there from CDC and others.

38:06Medium-scale maps start to give you a little bit more detail at a level where you live.

38:13When I asked you all who lives in a food desert, I've never been in a crowd where...

38:18...I've had more than about a handful of people raise their hands, because it's not a well-defined thing.

38:23There is a whole political argument out there; we're not going to go there today.

38:28What I'm here to talk about is when you start to zoom in a little bit, you start to see different things going on.

38:36Things people have published where on the top left...

38:39...there's a nice grid that's representing various demographic information in context to food deserts.

38:45The food desert locator from USDA just came out recently showing tract boundaries...

38:50...where the results of the analysis done over here is represented as a binary decision; tracts in pink are food deserts.

38:59And similar, Mari Gallagher on the top right has done this map of Chicago...

39:04...where she's translated census tracts into things we can relate to - neighborhoods.

39:10That long list of items on the left, those are all neighborhoods that Chicagoans would recognize.

39:16So that was one of my first clues - oh, so if you make this topic localizable and relatable to the person, okay, they start to get it.

39:25As a colleague, Randy Jackson at Ohio State when I was there, said, "I've never seen a census tract get on a bus."

39:34And so that kind of inspired me a year ago to start looking at large-scale maps.

39:38Kirk Goldsberry out of Michigan State is doing some very interesting things in the Lansing area...

39:43...showing here are the locations of various sources of food...

39:46...and here are the households that participated in my survey, and here is my analysis.

39:50So really good local study, and the map helps you develop kind of confidence about, okay, this is your methodology...

39:58...but, you know, I don't live in Lansing, so I can't quite relate to that.

40:03So one of my inspirations I guess last year was to look at - I'm sorry, let me make sure.

40:09Oh, one of my inspirations I guess last year was, we all have this mental map of food.

40:14I know where I shop, and when I see a walkability index for Redlands, the first thing I do is look at, well, here is my house...

40:21...can I walk to that Albertson's that I go to in the distance they specify?

40:27Oh, they're using straight line distance, I can tell.

40:30And I don't walk to a grocery store.

40:31It's about one and a half miles away, and there's some topography that's unfavorable to my 40-year-old knees.

40:38So I think that relatable aspect, how can I relate to this?

40:42We all have a mental map of our own areas.

40:44And like when you travel, you go to San Diego, for example...

40:47...how many of you know where the nearest grocery store is from you where you sit?

40:50Where is it, what's its name? Shout it out. Ralph's.

40:53Name the second closest? Yep, yep. Similar to how gas prices work, right?

41:03When you need gas, doesn't matter if the cheap gas is one mile further down the road...

41:07...what's convenient to you, what your eye can see.

41:09Probably saw the Ralph's sign; maybe some friendly Esri person pointed it out to you...

41:13...or a colleague, somebody made a map.

41:15But when you are in a condition where you don't have all the options you'd like to have to get the optimal...

41:23...access to what you want, that's where you're sort of at the mercy of the market, the market conditions. What's here?

41:30There's one supermarket that 500 of you just named near you. That's kind of interesting. What if there were 10?

41:37We'd say, wow, there's a lot of choice in this area.

41:39I can go to Ralph's, I can go to Albertson's. It turns out there is just one.

41:42So it's just that mental map we all have in our home, how does that translate to areas you don't know about?

41:50How many of you are from Kansas City?

41:54We have a very nice image on-screen of Kansas City right there; you probably recognized it.

42:00And square in the middle is a proposed supermarket.

42:04If you put a supermarket there in downtown Kansas City...

42:07...that map's telling you you would serve 4,400 additional people living in poverty.

42:11[Audience comment] It's there.

42:12Is it there now? Oh, good. Well, this is a year old, don't tell anybody.

42:17My point is that scale matters; all scales have their purpose. Not all scales can serve the same purpose.

42:24Large-scale maps definitely, showing neighborhood level information...

42:28...tend to be something the public can relate to a little bit more.

42:30Anyone here use Zillow.com? Yeah, Redfin, when you're looking for a house?

42:36So what if Zillow looked like this map on the right?

42:39That's an actual screen shot of a friend's house in San Bernardino County...

42:44...and the price of that house, I think he's buying it like 300,000 or something.

42:48And the median price for San Bernardino County is lower than that. So he's, I don't know, is he getting a good deal?

42:57The large-scale map would give him better context, right, about...

43:00...well, does he have access to other homes for sale in the area...

43:03...that are of similar characteristics and quality and price and availability and terms? No.

43:08So the scale definitely matters.

43:12So Charlie's indicated that the Japan map, someone else made it, brought it to him for review.

43:18The soil water retention map, someone in his own group brought it to him for review.

43:24My name's Jim, and this is my map.

43:26I got it out last year with - it was reviewed, but we didn't get it to where we wanted it to be.

43:32Let me explain what we were going for in this map.

43:35Because the definition of food desert is pretty murky, complex...

43:39...we don't quite have agreement about it, similar to the radiation situation...

43:42...I targeted one main topic with a little something slipped in on the side.

43:49I just wanted to show access, so in blue, the little Ss are supermarkets.

43:54And that's your typical dots on the map.

43:58And in green are all the areas that are within I think it's a one-mile walk to that supermarket.

44:05So all the areas in green one could argue, well, you can drive to it, you can walk to it.

44:08If you're the "typical" person, you're probably well served by this.

44:13So we're going to call the areas in green not a food desert, just based on access.

44:19But the red dots, what's that about?

44:22Well, not only did we draw a half-mile walk time around each or one-mile walk time around each supermarket...

44:35...we also did a 10-minute drive time.

44:37Those dots in red are people who...

44:41The dots represent census block groups with more than 30 percent of the population living below poverty.

44:50So if I know this block group 30 percent of X is below poverty, what's X?

44:56Oh, well, X is a dot. Each dot represents 50 people or 100 people or 25 people.

45:03And so you get both the density of the pattern, as well as just the visual.

45:08Okay, so these areas are not walkable, and because they live in poverty...

45:12...could be the case they don't have a car or reliable car transportation; they have to rely on something else.

45:17So what we wanted to do is just kind of illustrate access.

45:22And so if that's the before, just focusing on access...

45:26...knowing that the food desert topic needs a little bit more to it than just access.

45:32Many of the comments I received about this map were, hey, don't you know there's more to this than supermarkets? I sure do.

45:41You know, farmer's markets, convenience stores, fast food, fast food carrying good food...

45:46...nice restaurants, diners; again, back to personal choice.

45:49There's food everywhere, so how can we have a problem?

45:52How can - who raised their hand that said they live in a food...

45:54How can you possibly raise your hand that you live in a food desert?

45:57So to me that's terribly interesting; why can't this map explain it?

46:01So we were trying at this iteration to blend many of those socioeconomic things.

46:07Rather than just show raw socioeconomic data as dots on a map...

46:12...which it shocks people but it doesn't really grab them the right way, let's try to actually do some modeling a little bit differently...

46:19...to blend that into the underlying surface and create a little bit of a 3D view like this.

46:26So what was changed? We changed methodology, and we changed colors.

46:29Our original methodology was, which census blocks have walkable access to a supermarket.

46:35We used Network Analyst with NAVTEQ data, you can get it in StreetMap Premium...

46:40...and we analyzed from this block, how many supermarkets are within X miles or X minutes.

46:46You know, I ran a half mile, I ran one kilometer, I ran one mile.

46:50Looked at all those results and scored each census block.

46:54From this block representing all these households in this block, there are...

46:58...I mean if we talk about this, I bet the Ralph's is within a half mile.

47:02So that would be a score of one.

47:05So that's what is represented on the map. You do the same thing with restaurants.

47:11In New York City, there's a census block in Manhattan that has access to 611 restaurants within a half mile.

47:20I'm going to try not to let that drive my legend too much.

47:24The goal here was to put distance measurements in full view...

47:27...but again acknowledging it's only a partial view of how people access food.

47:32My purpose in this map was simply to get people to stop using rings and start looking at access.

47:38From where you live, how would you get to these, this network of dots on the map?

47:44'Cause after all, when you look at that map, with all these supermarkets...

47:46...there's 26,000 supermarkets that I used in my nationwide study.

47:51How can you say there's a food - look at all of these supermarkets. How can you say there's a problem?

47:55Well, Kansas since 2007, the state of Kansas lost, I think it was, 84 out of about 200 supermarkets...

48:03...more than a third; since 2007, they're gone, they don't exist.

48:08And the article I read said, the reason that's interesting among other reasons...

48:12...is that the hardware store right next to it is now gone too.

48:15So it's an economic development. It's, you know, this is part of the bad cycle, so it is what it is.

48:22And if we want to talk about food deserts effectively, we need to be able to talk about the components effectively.

48:29So our goal with this revised methodology was to start to put some of those components into the map and see...

48:36...if it rings true for people.

48:38'Cause what rings true in New Orleans might be for different causes than what rings true in Philadelphia.

48:44So we worked in Philadelphia in the time available before User Conference for this project.

48:50Our goal is to, again, create a relatable measure of the food environment where access is just part of the story.

48:56We wanted to create a number of socioeconomic conditions and make them explicitly part of the map...

49:02...poverty, car ownership, percent income spent on food at home...

49:06...from the Consumer Expenditure Survey that the US government puts out...

49:10...measure of crime, rural population, distance to and from restaurants, distance to and number of supermarkets.

49:19So you can see already what we've added, in addition to supermarkets access...

49:22... now we are thinking about restaurants and farmer's markets.

49:25Our approach was to take census block group data for those various categories and interpolate a surface using natural neighbor...

49:32...and then classify the surface into three categories.

49:36If it's a problem area, because its poverty rate 30 percent or higher in that census block group...

49:44...that's probably - I think most people would agree, that should be an area we look at.

49:48One is likely not a problem area...

49:51...but we just want to have a buffer between total problem area and definitely not a problem area.

49:57I live in the suburbs, I've a couple of cars; if one breaks down I can take the other one.

50:01It's not a problem for me to get to the supermarket anytime I need to.

50:05So I'm in that green area, and that's the difference between the fact that I don't have walkable access to a supermarket.

50:12I have a couple of cars, I can get there, it's not a problem.

50:15In other situations, I have a 25-year-old sister-in-law who lives in Manhattan, she walks 10 minutes.

50:24I loved it when I asked her, "How far do you walk to your grocery store?" "Ten minutes." I just wanted to hug her, thank you.

50:29That's exactly what the literature says is typical.

50:33What we did then is create these surfaces for each of those demographic factors that we talked about earlier...

50:39...and then summed them together.

50:42Whereas our original color scheme, green is good, red is bad...

50:46...and people of a certain socioeconomic condition are represented as dots to show who has access walkable, who does not.

50:54What we actually did in this map, we were showing two measures of access, so at this scale...

50:59...and this map is up on ArcGIS.com - you just type in supermarket and you'll find it.

51:03This is a nationwide map showing access.

51:06And if I could ask the - I don't know who's here.

51:09Yeah, my legend, sorry. I should've moved the legend, I should've made the legend bigger.

51:13The areas in brown, you can't walk to a supermarket, but also you can't drive to it in 10 minutes.

51:18You could drive to it in 20 minutes, but you can't do it in 10 minutes.

51:22And that's just again used as a measure of access.

51:25The areas in gray have drivable access to one supermarket. Why did I choose to show those in gray?

51:35Because, I don't know, if they closed in the worst recession since most of us have been alive...

51:39...then those are areas you would kind of want to focus on.

51:43There're areas of opportunities if they close for whatever reason, somebody else could open.

51:46And then the areas in white at this scale, they are totally drivable.

51:50So as you moved in, notice that almost of this map is white because the whole city is traversable within 10 minutes...

51:57...from each of those individual block groups.

51:59So again, I'm taking each census block, rather, and saying, Can I get to at least two supermarkets within 10 minutes?

52:07The answer is yes, you can drive it. So does anybody in this town not have access to a car?

52:12Yeah, I think that the folks at that end over there in south Coatesville, those are the areas of concern.

52:17So the trick for us was to figure out how to represent that.

52:22Again, we used the classification approach to blend all this together...

52:24...all the socioeconomic conditions that might be of concern, and put it together in a map.

52:29And we had this pre-shaded relief version, this is the point at which Charlie dropped by and said...

52:37...Yeah, I've had this idea for a while. Why don't you try and put some hillshading under it and see what happens.

52:42Okay, that sounds good. That's why it's good to have a cartographer nearby.

52:46So we tried it. And the initial hillshading on the left looked like that, and the one on the right...

52:50...represents what we did with a piece of the software called focal statistics.

52:55We ran what's called a focal min, and we looked at every cell in the blended raster...

53:01...and we said, What's the lowest value nearby? And you get a map result like this. It's very pretty.

53:07What it does for us is it takes a relatively flat surface like this and puts some depth to it.

53:14And that's an effect we wanted to have, right? We wanted to kind of show we as a country, we're doing great.

53:20There are little pockets that we think might be of concern for this set of specific reasons.

53:26I'm not going to publish a paper about what we did in a three- or four-week scramble...

53:30...because the point is not our methodology is magic.

53:33The point is we took a very conservative approach about what we're defining as areas of concern...

53:39...put it together so that we could surface right across the street from or the river from Philadelphia.

53:45It might be pretty challenging to get a supermarket chain to agree to build something there.

53:50So what other options are there?

53:51And again, driving back to if we can't describe the individual components and layers...

53:56...we probably can't solve the food desert problem, whatever it actually is, in city by city...

54:02...unless we can kind of show and visualize and have people who are Tweeting, I live in a food desert...

54:08...raising their hands at a public session, I live in a food desert.

54:10What I want you to do is go type in your address to a map like this, see what it says, give us feedback.

54:15And that's, I think, one of the points of doing this on the web.

54:20The Japan example was a print map.

54:22We hope that maps like this allow us to put together a user experience where you can kind of validate...

54:28...yeah, does this work for you, does this not work for you, similar to the USDA food locator.

54:34In this case we applied a 60 percent transparency on the final result.

54:38And the effect, I'm afraid it probably works well from midrow back.

54:44I think these are the 100-meter cells, I'm not sure. But these, the idea is that those red areas should be kind of recessed.

54:53And that's an effect we're going for to kind of grab ya.

54:59I want to give a special thanks to Clare Steiner - can you stand up? - our intern extraordinaire...

55:05...who did the reclassifications and the blending for us for this project.

55:10Clare's been a good sport putting all this together under pressure.

55:15She graduates in May of next year, so Clare will be up here.

55:20I want to thank you for attending; we're going to take questions now. And I'll remind you to take your - fill out your evaluations.

55:27There's an online; it's at esri.com/evaluations I think.

55:33Thank you.