Selective Drainage Toolbox—Lidar-based Elevation Rasters

Video Transcript
Download Video
Right-click and save to download

Detailed Description

National Hydrography Advisory Call:

This episode of the features speakers Ryan Thompson (Dakota Water Science Center – Huron, South Dakota) and Curtis Price (Dakota Water Science Center – Rapid City, South Dakota). The presentation is an overview of the Selective Drainage Toolbox. The discussion focuses on a study area in South Dakota that identified the need for the tools and the methodology that went into development of the tools. Q & A follows with both presenters.

Hydrography for the Nation:

High-quality hydrography data are critical to a broad range of government and private applications. Resource management, infrastructure planning, environmental monitoring, fisheries management, and disaster mitigation all depend on up-to-date, accurate, and high-quality hydrographic data. The U.S. Geological Survey National Geospatial Program National Hydrography Advisory Call has initiated a series of virtual seminars to highlight the uses of hydrographic data. These presentations are intended to share success stories from users who have solved real world problems using hydrography data, provide information about the National Hydrography Dataset and related products. The USGS manages surface water and hydrologic unit mapping for the Nation as geospatial datasets. These include the National Hydrography Dataset (NHD), Watershed Boundary Dataset (WBD), and NHDPlus High Resolution (NHDPlus HR). Hydrography data are integral to a myriad of mission critical activities undertaken and managed by government entities (Federal, State, regional, county, local, Tribal), nonprofit organizations, and private companies.

For more detailed information on national hydrography products visit https://usgs.gov/NHD
 

Details

Date Taken:

Length: 00:52:18

Location Taken: Reston, VA, US

Transcript

Al Rea: All right. So hi, everybody. This is Al Rea, and this is the USGS Hydrography Advisory Call, and we'll ... Today, our guest speaker is going to be Ryan Thompson, and he's going to talk about Selective Drainage Toolbox that he's been using, and I think we've got the main developer, Curtis Price, is also on as well to talk about this, so this is a tool that's used with lidar data to make a good, well-behaved lidar surface, and I think ... Did we have any announcements, Becci or anyone else at the USGS crew? Did we have any announcements we needed to make today?
Becci AndersonNot that I can think of, but if I come up with anything, I'll add it at the end.
Al Rea: Yeah, okay, so, Ryan, go ahead and take it away.
Ryan Thompson: Okay, thanks, Al. As Al said, I'm Ryan Thompson, and Curtis Price is also on. We're in the Dakota Water Science Center. I'm in an office in Huron, and Curtis is in Rapid City, and we'll just have a presentation today that goes over a little bit of a project-specific background for the Selective Drainage Toolbox and then talk about using it sort of in a generic sense, and hopefully be able to answer questions as to if it might work for your application. Let's see, okay. There we go, so I guess what I'll be talking about is the study that we had in South Dakota that sort of identified the need for the tools, a little bit about the study area for that and then also the methodology that went into development of the tools. Go through some of the processing using the Selective Drainage Toolbox and how we reviewed the potential culvert locations that are identified and then what we wanted to be able to get out of it as a science center was to be able to refine the watershed boundaries and subdivide to 14 and 16-digit units where appropriate. I'll touch a little bit on what some of the value was for our cooperators, and then we'll have a time at the end for questions. So the gist of it was that we needed to identify a way to kind of have a standard methodology for utilizing the lidar data that we already had in place in South Dakota, and we didn't have complete culvert inventories, so we needed to be able to identify locations where the culverts were and then hydro-enforce them to get a culvert-aware lidar-based DEM that could be used for resource management, and then we'd refine our watershed boundaries with that and develop a densified drainage network, a few more lines than what you can see in the NHD, and, as I mentioned earlier, to update with 14 and 16-digit unit lines where appropriate and then also explore seeing if there's areas where the NHD lines, flowlines, might need to be updated as well. We had kind of a long list of cooperators on this study, and I won't go a whole lot into it, but the state and local cooperators agreed to do the ground truthing of the culvert locations that we came up with through this Toolbox. Some of them, you could verify that there was a culvert there through imagery. There was a lot of other locations that needed to either have field verification of some sort or use of a different imagery set, and I'll talk a little bit about that, but the cooperators agreed to do that on their own, with their own staff, which made the project a lot more affordable, given that we didn't have to go out and do ground truthing in the field. So a little bit of background on the project specific to South Dakota, I'll go over the location of the study area and kind of physiography there, sort of a hydrography there, the basis of the methodology and then what we had going into it for data. You can see the study area is that little red polygon that's in the kind of the ... near the southeast corner of South Dakota, and that pink-shaded area is the Prairie Pothole Region, and so you can see out study area is kind of on the edge of the Prairie Pothole Region, and that's characterized by poorly developed drainage. There's not a well-defined stream network in a lot of those locations, so a lot of wetlands and sloughs, and that was the area that was looking at contour lines on a 1:24,000 map. It's not necessarily always very evident where or what direction water is going to flow when those splatter areas, so this is kind of a concern of some of the co-operators. They needed to worry about expanding the urban areas and developing into the areas that are currently ag now, and this was just going to be a tool that made that possible for them. Actually, let me back up here. I guess you can see part of the study area does have a fairly defined stream network. It's mainly in South Dakota and a few of the different counties there. Part of it did spill over into Minnesota and Iowa. As sort of the southwestern edge of the study area, that's where it gets into that Prairie Pothole kind of a topographic setting, and where it was a little bit more critical to have the lidar data there. The basis for the tool, it builds on the Selective Drainage methods that were pioneered by Sandra Poppenga, Bruce Worstell, Jason Stoker and Susan Greenlee at EROS, and they published that in the SIR there, and then Curtis kind of expanded on that in building the Toolbox. The gist of it is that most of the time when we are delivered lidar-based digital elevation model, it's typically a bare-earth type of product, and from what my understanding is, a lot of the contracts are written such that bridges, or at least bridges larger than a certain size, need to be hydro-enforced, so the bridge deck will be removed and the surface will actually reflect the channel, but culverts oftentimes are not hydro-enforced, or at least not all culverts will be hydro-enforced, so if you model run-off over a surface that does not have hydro-enforced culverts, basically the roadbeds, or railroad grades or other built infrastructure is going to act as a dam. The lidar returns are going to be off the crown of the road, in the ditch slope, but the DEM has no way to know that water can pass through that culvert until you hydro-enforce it. So the ... Kind of putting it into a little flow chart, we start out with an elevation model. You can fill that to get a ... generate a flow direction, flow-accumulation grids. Once you filled it, you can get a fill-difference grid that kind of comes into play. From the fill-difference grid, you can come up with the filled areas, and when you intersect the filled areas with the highest flow accumulation cells, that will tell you where the pour point is for when that fill area begins to spill and contribute drainage to the adjacent topography again. From the filled area, you can also go ... Using this tool, you can sort of use the Selective Drainage methods and identify areas where culverts need to be burned in, then you can actually burn those in and feedback through the beginning of the loop and do this in an iterative manner, and if you do know some of your culvert locations prior to this, you can certainly burn those in at the top, and that will get you part of the way there, and the tools should help you identify the remaining culvert location. Though, here's just an example from our study area. You can see in the aerial image on the left, we have a stream that flows roughly north to south, or top to bottom. You can see the culvert going underneath the ... It's a gravel road there, a township road, and it's obvious in imagery, but the ... In the DEM, it does not know that's there, so it's going to just act as a dam on the left here. In this image, the pale blue kind of transparent area represents the fill area where the water would basically pile up on the upstream side of the road until it hits an elevation such that it can finally overtop the road. And the low point in the crown of the road, in this case, is a little bit to the left of the culvert, but it's not too far off, so in this case, it didn't change the flow path a great deal, but if you digitize a line where that culvert is and higher enforce it or, effectively, just trench through the road at the location of the culvert, you can rerun your flow accumulation flow direction and generate your streamlines, then you're going to get streamlines that match up a lot better with what you actually have going on in the field. As far as the data that was available in our study area, quite a bit of Eastern South Dakota was available. We had lidar-based elevation in the NED for South Dakota. Across the border in Iowa and Minnesota, there was similar vintage lidar data, but it was in a state-maintained product and not in the NED. Additionally, Sioux Falls, for those of you not familiar, Sioux Falls is the biggest city within our state that was within the study area, and they had a 2012 lidar acquisition to kind of update some of the areas that they're expanding into, and they wanted us to be able to use the 2012 data where it was available and then use the 2008 data surrounding that where it was not. Additionally, there is Pictometry imagery available for Lincoln and Minnehaha counties in South Dakota, and Pictometry is an oblique aerial imagery that, in this case, had a 3-inch pixel size, very high resolution. Because it's oblique, you can look at things from different angles, and that proved really, really helpful in confirming culverts at different locations, and we'll see some of that imagery in slides coming up. Also, we had some culvert data available going into the project. There was an inventory that contained culverts, but only if they were 30 inches and larger on state and city highways. The city of Sioux Falls had a GIS of their storm water system, and culverts were in there. They weren't necessarily in a feature data set by themselves, so we kind of had to pick those out from the rest, but we were able to use that information. Also on county highways in Minnehaha county, there was inventory for culverts 30 inches and larger, and then throughout the project, culverts on county roads within Lincoln county were inventoried, and then also some of the culverts on local and township roads in Minnehaha County were inventoried, but our tool was able to find a lot more culverts that couldn't be identified in the field because it got late enough in the summer that the grass grew up, and it was just very difficult to see well enough to find them all. So the ... I guess we mentioned earlier, the Toolbox was developed by Curtis Price. It was released on ScienceBase last September. At that time, we were at ArcGIS 10.4. I wrote up a little bit of a user guide that's available to go with it, and then currently, we're in the process of updating to make the toolbox compatible with ArcGIS 10.6 and also Pro, and we're hoping to be able to release that here within the next maybe month or two. And if you search for Selective Drainage Toolbox on ScienceBase, I'm pretty sure that will come up. Although, I think there's a link for the ScienceBase page in the meeting announcement today, or the call announcement. So once you've downloaded the Selective Drainage Toolbox, it's going to look kind of like that screen capture there on the right. There's a little bit of a utility tool in there, in case if you need to start with, let's say, a LAS file, and you're working with points rather than a raster product already. Tools one, two and three are included in the global tool as well. I guess in the workflow that I've kind of settled in on, I tend to run those tools consecutively more often than using the global tool, and then tools four and five are used to enforce the culverts and then evaluate the identified culverts. If you download this and some of the tools are showing up for you, it may be because an extension that they need has not been enabled in your settings within ArcMap, so that's something to keep in mind. We've run into that before. There's kind of some terminology that might be helpful for you to know going into using it. Drain line is the name that we've given to potential culvert locations that the tool identifies. There's also a search distance that the user can specify, and that's basically the maximum distance that you want to allow between the minimum point associated with a particular fill polygon and the end point of its potential drain line or culvert that the tool will find. You can also specify maximum and minimum culvert-fill-zone areas, and that's useful for if you're in a setting that has natural lakes, you can use the height, the max filter, to make sure the tool isn't wasting time trying to drain your lake with a culvert if it's actually a lake on the landscape, and, likewise, you can specify a minimum size that the tool doesn't need to be concerned with, and that way you don't identify a culvert for every little puddle on the sidewalk. There's also a ... pour-point distance is a term that is kind of the maximum length that you want to allow for the drain line or the culvert features, and then the tool was set up to run iteratively, and that was so that ... In some cases, there's going to be more than one culvert that fall within the extent of a given fill-area polygon. Sometimes there is a culvert and a relief culvert nearby, or that kind of a situation, and the iterations help identify multiples. So this is ... We'll kind of go through a sequence of slides that just shows what the menus are going to look like when you open it up. There is, if you show help, there's going to be a little bit of help on the sidebar on the right there, and then as you click into each one of these fields, there will be a description that hopefully will clear up questions you might have there, but for tool one, basically, you enter the elevation raster that you're starting your work with, and then populate the name and path for the output fill raster, the flow direction and flow accumulation. You can also give a name, assign a name, to a different raster, and fill-line rasters and specify a minimum area there. The minimum area can be specified either in cells, depending on what your resolution is, or you can specify it in terms of just an area. Tool two, basically, it takes the fill zones that are going to be built in the first step, and it comes up with some extra information for them that's needed for subsequent steps, so you have your elevation raster, and fill raster and flow direction, flow accumulation. From the fill-zone raster, it's going to make polygons of those fill areas. I'd mentioned earlier that it intersects the fill areas with the flow accumulation grid to identify the pour-point cell, the cell with the highest flow accumulation value, and then also the minimum elevation point within that fill zone, and here, again, you can specify a minimum area that you want it to be concerned with. Tool number three is creating the selective drainage flowlines, so the input point is the minimum point that we found in the earlier step, and the fill-zone raster, elevation raster, and then the output flowpath line is just what you want it to name the file that it's still going into these potential culvert locations, and here's where you can specify the search distance for how far you want to allow it to look for the ... Basically, the cell on the other side of the road, that is, is what we're looking for there, so if you have ... If you're working in an area that primarily is two-lane roads, that's going to be a ... You'll be able to use a shorter distance than if you're in an area that has multi-lane roads, or for us, it's the interstate, and you could do this in multiple steps if you wanted to. Once you've gone through ... And we'll talk more about this in slides coming up, but once you've identified your locations of your culverts, this is a tool that basically burns those in. You can also specify lines where you want it to be walled up as well as just having culverts burned in, and that might be for if you have a situation where there are either berms or levees or some kind of a feature that was either added after your lidar data was collected, or if it's a ... We have a few locations where there's kind of a concrete flood wall that's narrow enough that it may not get picked up in lidar data, but it could be an important feature that you want included in your elevation raster. There's also a field there where you can specify sink points, and you might find that useful if there are something like a quarry, or gravel pits is something that we run into in our study areas, where, basically, it's sort of like a non-contributing area. A terminal lake is what it's going to act like, and that just allows ... If you put a sink in there, then it doesn't try and find a culvert to drain that out because, in a lot of cases, there's ... That'll be the lowest elevation area around, and you won't be able to get it to drain. And offset z-value is basically how deeply you want your culverts trenched, in and how high you want your walls built up there. Tool number five is once you've burned in your culverts, then you'll want to rerun the elevation derivatives to get your updated flow accumulation, flow direction grids. Then this takes those layers again and builds a synthetic drainage network, and you can see from that little snip of the table at the bottom, Curtis has this set up so that each stream reach is populated. There's an attribute table that's populated with the number of cells that that particular reach drains, and also the area, so once you get this layer out, you can ... If you're going to display it in ArcMap, you can change your symbology depending on if you want to show a very detailed, zoomed-in sort of drainage network, or if you want this more of a broad-scale, not-quite-so-much detailed. And then I mentioned also, there's the Global Selective Drainage tool, and this kind of is set up so that it can run the other tools consecutively. Allows you to specify the same kind of limits we talked about earlier, and then this is where you can specify if you want it to do one iteration or three. Let's see, and I think the next ... Yeah, we also put together kind of some tips for processing, and depending on what your DEM resolution is and the size of your study area, we found it really helpful to run the tool on a subset of the total study area, rather than turning it loose on it, and having it crank away for 3 days or something like that. It's a little bit more helpful to have it just ... if you specify an extent ... and the tools that the Toolbox uses will honor a processing extent, but not a processing mask, so that's something to keep in mind, but if you can chunk it out in sections, that will get you progress faster than if you give it one massive area and make it process all the way through that. You can kind of do a little bit of playing around to find out what your optimal processing area is, and while you're doing that, I'd suggest having it just do one iteration until you've kind of honed in on what size is going to complete in a reasonable amount of time. I'd also suggest having it process, including in your DEM, and area beyond what you expect the study-area boundary to be, so, for us, we had the WBD lines that were based on the 1:24,000 contours, and then we included a buffer beyond that just because we thought there was a fairly high likelihood that those boundaries could change, one way or the other, based on culvert locations with the high-resolution lidar data, so we included a buffer there. And as you develop your layer of culverts, I found it helpful to attribute them to just kind of give ... In our case, it was for our own use and also for the cooperators, what we used to say, "Yes or no, there is or isn't a culvert there," so this is an example for that. Basically, we went through and I added a couple fields for if ... Is it a good location? Typically, you're going to find a culvert going under a road, but not in all cases, we discovered, but ... so if you have a ... if the tool finds ... has drawn a culvert line, but it's in a depression out in the middle of a field and it's not even adjacent to a road, then you could say, "No, that's not a good location." We also did kind of a preliminary check within just ESRI basemap imagery to see if there was a culvert visible within the imagery, and if there was, then we populated that in that column as well. And then I ... Well, I guess we have another slide coming up for Pictometry. There's better imagery out there now. I've found that Google Maps is actually pretty good in a lot of areas, and that works well for just being able to look and see if there's a culvert in that location.
Curtis PriceYeah, I mentioned, Ryan, we were using Street Map.
Ryan ThompsonYes, yeah, that's also very helpful. I'm not sure if all states are that way, but in South Dakota, there's, at least on state highways, there will even be a delineator post in the ditch to mark where the end of the culvert is for when the ditch gets mowed and things like that, and even if you can't see the culvert itself, if you can see the post or there's kind of little cues that you can tune in to to help you decide if there's something there or not. So the tool will come up with these assumed culvert locations or potential culvert locations, and this is just kind of going to go through a couple examples of those. Again, this is a stream that flows roughly from the top of the photo down to the bottom. Went up against a gravel road again, so we have that pale blue area is the fill zone, and then the pale blue line is the stream line before the culvert was hydro-enforced, so you can see we have a lot of the straight and diagonal stream lines going on there in that fill area, and in this case, the low point in the crown of the road was almost directly over the location of where that culvert is, so the fact that it spilled over the road didn't change a lot of the path of the main channel, but once we did hydro-enforce the culvert and rerun things, we got a lot better set of flowlines for the stream once that fill area was not there, impacting it. We had other situations where this is a similar deal. This is actually a kind of a draining canal, following from north to south again. In this case, the point where water would ... It didn't overtop the load. Instead, it overtopped kind of a berm adjacent to that drainage canal, and then it flowed off to the upper left in that corner of the image there, so in this case, if we would not have been able to ... If we didn't know that this culvert was there and hydro-enforce it, our stream network would have been very much impacted. This flowline actually went over, kept going to the west in the section, and then it ended up dumping into a tributary adjacent to the west, so it can potentially make a big difference if ... That's really one of the ... The pour point does not coincide with where the culvert is in the road. We also had some things that I didn't really expect to see, but the tool was able to find. If you look at that aerial image on the left, you can see that there's a fill area that's kind of coinciding with the old oxbow channel of the Big Sioux River, and there's definitely not an obvious road there, but it was identifying potential culvert locations on the road, and if you look to the DEM, you could definitely see, in fact, there was something build across that oxbow, and let's see, I think it's the next slide. This is that high-resolution oblique imagery that I mentioned, and I'm not sure ... If I mouse over the picture, are you guys seeing my mouse at all?
AttendeeYep.
Ryan ThompsonOkay, so where that red arrow is, this is actually the path that we were seeing in the DEM, where the farmer was ... had that road to cross that oxbow lake to be able to get to fields on the other side. It's a little bit settled, but if you have enough evidence, you can find it. Yeah, there has to be culverts there. There's a few other interesting things we were able to see in looking through this. Here is an example of something that is not a culvert. It's a depression that's not adjacent to a road. The red drain lines the tool came up with were not in locations likely to be a culvert, but in looking at that little X as the minimum point for that fill area, we can see that ... You can tell there was something going on with ... in the field. The farmer basically had to farm around some object there, and so what I'm thinking this object is, is the inlet for an underground ag drainage system, which I don't want to oversell this, but there's going to be times when you can find things like that, too, that are helpful to know, so if you were going to model this, you could actually put a sink in there and have that behave better than running off. And then as I mentioned, one of the things we wanted to be able to do with the tool is update our WBD boundaries, so in this slide, you can see in green the 12-digit HUC boundary based on 1:24,000 contours, and that's compared to sort of the blue polygon with the red boundary of what we came up with based on this lidar data. For the most part, there's pretty good agreement. We can see one area right there, where there's a little bit of divergence. Then when we looked at that closer, you could see that there was actually the watershed flooded up against a road there, and so we went back to the imagery and looked at that again, and within ... This is ESRI basemap imagery. There's nothing obviously ... nothing in there that screams culvert to me, but if you go back and look at it with this Pictometry imagery that I mentioned, in here it becomes very obviously that there's the end of the culvert in the ditch, so this was a spot where we needed to hydro-enforce one more culvert, and then the boundary ... rerun the elevation derivatives, and the boundary will adjust itself. And then I guess you could go through and subdivide your 12-digit HUCs into 14s and 16s, if that's something that is helpful to your cooperator. So these slides were ones that, when I originally did this presentation, the cooperator talked on, so I'm kind of trying to go from memory here, but I know one of the ... Some of the applications that they used it for was there were some areas within their municipality that had undergone some development prior to implementation of their drainage ordinance, so this helped them to kind of know what was going on there. They were able to ...for areas that were slated for development, they were able to identify spots that they could expect nuisance ponding, and maybe even specifically suggest that, "Hey, this is an area that could be a good stormwater retention area. It's already low. It's going to hold water during certain times." Just to help let developers know where culverts were likely going to be needed, and what the drainage areas were going to be for culverts in new areas. And I guess this was an example they had put together of ... You can kind of see the gray crosshatch polygon there, and that was area that was identified in a FEMA floodplain FIRM map, but a landowner wanted to be able to put a storage building on part of their lot, even though it was in this floodplain-designated area, but with use of the selective drainage tools, they were able to see that ... Let me see if I can get the mouse there ... that a corner of that lot was actually high enough that they could add the storage building on there, and it made the landowner happy, and made the cooperator happy to be able to help him as well. I guess that's kind of all I had, but we definitely have time for some questions if anybody had some.
AttendeeI have a question.
Ryan ThompsonAnd Curtis is on too, so if you have questions specifically to the nuts and bolts of authoring the tool itself, he will be able to field those as well.
Al ReaSo thanks a lot, Ryan. I'll start off and ask ... I know you've alluded to this a couple of times, but could you give us some examples of what the ... first of all, what your DEM resolution was, and then what size of processing units you ended up using and what were your run times?
Ryan ThompsonSure, so the DEM that we had to ... I guess we went back to the 1-meter source data, but we actually did a little bit of kind of a sensitivity analysis, and we took a small subset of the area. We went through all the steps, and we ran and got sort of a ... the synthetic flow network when it was at that 1-meter resolution. Then we also redid it when it was resampled to 2-meter and 3-meter resolutions and compared. We actually went up to 5-meter as well, and sort of did a comparison to see how much can we ... how big a cell could we have and still not lose accuracy? And that was the 3-meter cell size is what we kind of settled in on, or yeah. Yeah, 3-meter cell size. It was actually the city wanted to work in English units, so it was 10 feet is what our cell size was. I think initially we started working ... We kind of tried to keep our processing units to 12-digit HUCs or smaller, but because it's working on an extent box and not necessarily an analysis mask, sometimes we'd cheat the boundaries to ... just to get another ... squeeze another one in there and have the processing time not take as long. I think it was, the steps that took the longest to run were: Flow accumulation was one, and then the step where it's identifying the potential culvert locations was also one that it cranked away at for quite a while, and those often would run overnight and would be ready the next morning, or partway through the next morning, if I started those in the afternoon the day before. Did I catch all those out, the questions you had?
Al ReaYeah. That was good, so those were kind of overnight runs, and everything else you could do in ...
Ryan ThompsonYeah.
Ryan Thompson... sort of ...
Curtis PriceScalability is, of course, a real issue with these things, and ... But the cooperator originally is going, "Well, we have this 1-meter data," and I guess my point is, is when you average that out to 3-meter data, you get really good 3-meter data because you've averaged out all the artifacts, so it's ... There's a lot of stuff in there you see in the full resolution that isn't really real for this kind of analysis.
AttendeeHey, guys. This is Sean Vaughn from Minnesota.
Ryan ThompsonMm-hmm?
AttendeeSo I understand your methodology well. It reflects a lot of what we've done here. I've been doing this type of work for well over two decades. Question specifically is directed towards your search tolerances. In much of our experience, when you're allowing the algorithm to search and create that line, it ... We sometimes get what we would call here a digital dam bridge line that really isn't in exact alignment with the culvert. Did I miss something in the presentation? As I'm trying to think back to it, did you realign things, or how did you maintain more of a perpendicular directionality for your lines?
>> So what I did in my workflow is, I would go through and review these potential culvert locations, and rather than use the drain line that the tool came up with, I used that as the basis for, "Okay, that there's a culvert somewhere here, but I'm going to digitize a line of my own that I'll call the culvert," and then I could make it ... Depending on if you had evidence in imagery, you could make it perpendicular or a little bit diagonal, but then you could get the length of that culvert a little bit more realistic. The tool, you kind of have to give it a long enough distance so that it can get across the road embankment, in cases where the culvert is not perpendicular, but that kind of has the side effect that it may continue the culvert farther down the slope than what an actual one would, so I digitized my own layer of the culverts just to have a little bit more control over that, and then also attributed those culverts as to, "Is this one we could see in the imagery, or is it one that we just knew it was there because of looking at the" ... If you can see a channel on the upstream or the downstream side of the road, but you can't see a culvert, you can oftentimes be assured that, well, there must be a culvert there, or sometimes you can see evidence in imagery, or sometimes you just have to assume there's a culvert there because there's not water ponded up, or you know that, well, the township is not going to put up with water running over the road every time it rains, so there must be a culvert there. So yeah, I digitize my own culverts, and then attribute it as to whether there was evidence in imagery, or if it was something that was just kind of a, "I presume there's a culvert here."
AttendeeSure. So I apologize if I missed that level on that detail earlier. One of the reasons to flush that out is because going back, manually digitizing them certainly does add to the time commitment of this work, and one thing a lot of us in this meeting tend to be aware of is just this is very time-intensive work, and you just really can't ... It's always so hard to get away from the human involvement. You need human eyes on this in reviewing it so that we can ensure that our flow paths exist with a lot of convenient features.
Ryan ThompsonTrue, and I guess we were coming at this kind of from the flatland extreme. As I mentioned, we had kind of a part of our study area was sort of a Prairie Pothole environment, so there we knew there was going to be culverts that were going to be really critical. If you're in a higher relief environment, I ... We haven't worked in a super high-flow area yet, but I presume that more of your culvert lines that the tool comes up with are going to be found basically right over ... They're going to plot ... The pour point is going to be right over at the location of the actual culvert location, and it may be in an environment like that, you could get by with less oversight. If you ratcheted down your culvert length, you may be able to get that to work without having to digitize things.
AttendeeSure. Thanks for your response.
Ryan ThompsonYep.
Al ReaSo, Ryan, you mentioned that your co-operators went out then later, I guess, to verify some of these?
Ryan ThompsonYeah.
Al ReaMaybe the ones that you couldn't see in the photos, or did you work that back into a revision cycle?
Ryan ThompsonThere was kind of a little bit of iteration that way. I guess we came up with our layer of, "Here is where we believe the culverts are. These are the ones that we know are there." Or, well, and I guess in our case, there was some that were already inventoried. They were a given. There were some that were newly found and confirmed through imagery, and then there was some that we just believed to be there, so what they did is they took that layer. They would have their staff check it again with that Pictometry imagery that I mentioned. Depending on the time of the year, that imagery coincided with if the grass was short and you could get an ideal angle. A lot of times, you could see the either the ends of the culvert, or at least a shadow where the opening of the culvert was, and then they could confirm without actually having to drive to the spot, but there was some number of culverts that were a little bit more critical, that they couldn't confirm with imagery, that they did actually do a field visit for, but most of that was a desk exercise, actually.
Al ReaOkay.
Al ReaMore questions from anyone? All right. Well, Ryan, thank you again. Thanks very much for your presentation. It was really interesting, and then you will send out ... Let's see, did you have the URL where this tool is available? We should ...
Ryan ThompsonI believe that's on one of the slides, yeah. Let me back up here.
Al ReaThere's something USGS runs called ScienceBase, and you can probably search for ScienceBase in Selective Drainage and find it.
Ryan ThompsonRight, yup.
AttendeeI have a quick question. This is Karen Adkins. You mentioned that you're upgrading to ArcGIS 10.6 and ArcGIS Pro. Do you have an expected timeline on that at this point? Are you just getting started?
Ryan ThompsonWe're ... I think there's just one of the tools that's not working in Pro, or one of the steps in the Toolbox is not working in Pro. The rest are all working in 10.6, and I might have to defer to Curtis as to how much it's going to take to tease out what's going to work in the Pro version, but I guess I'm hoping we can get that out in ... within a month or two. Does that seem reasonable, Curtis?
Curtis PriceYeah, it just had a few little tools that it had to do. The Python code actually works both ways pretty well. The biggest issue we've had has been the changes in some of the raster tools that ESRI has made in 10.5 and 10.6 that have kind of changed how temporary files have been handled, and it's been kind of a pain for a lot of these applications, and so we're quite willing to hand out if someone wants to work with us and try out one of these tools in the newer version, they're welcome to get a hold of us.
AttendeeThat's excellent.
Al ReaIs the source code available there on ScienceBase as well?
Curtis PriceYeah, when you download it, it's all right there, yeah.
Al ReaIt's all there, so ...
Curtis PriceYeah.
Al ReaOkay.
Curtis PriceYou have toolboxes and script folders, and it's all in a fully available package there.
Al ReaAnd it's Python, right?
Curtis PriceYeah. It's all Python, yeah.
Al ReaOkay.
Curtis PricePython toolboxes, PBX file, and script files and layer files, and then the layer files are set to symbology and all that kind of stuff, so it's all packaged in a toolbox, a folder with toolboxes and stuff, and the development has mostly been done in Arc 10.6, and then we save it back to a previous version, so there's a 10.4.1 version toolbox. You can read from 10.4, or 10.5, or Pro or whatever. You can always read forward. You just can't go back.
Al ReaOkay.
Al ReaGreat, well, thanks again to both Ryan and Curtis. Were there any last-minute announcements from anyone on the USGS team?
Becci AndersonNope, not that I can think of, Al.
Al ReaOkay, well, I think we'll close then.