Lidar-Based NHD and WBD Delineation in Western Oregon

Video Transcript
Download Video
Right-click and save to download

Detailed Description

This episode features Jay Stevens (GIS Specialist BLM – Oregon/Washington) Jay presents BLM’s collaboration efforts with the Forest Service, the Oregon Department of Forestry, Oregon Department of Geology and Mineral Industries, NRCS to delineate Western Oregon with new lidar-derived NHD and WBD.

Hydrography for the Nation: 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.

Hydrography data make it possible for these groups to, for example:

  • Manage water such as stream flow and stormwater
  • Monitor, manage, and report water quality
  • Assess water availability and water rights
  • Model and map flood risk
  • Conserve terrestrial and aquatic habitats
  • Manage fisheries, rangeland, timberlands, and agricultural lands
  • Assess coastal hazards
  • Plan for future land development activities and infrastructure development
  • Manage riverine and coastal navigation and safety
  • Provide recreational opportunities for citizens

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

Details

Date Taken:

Length: 00:57:38

Location Taken: OR, US

Transcript

Al Rea... get started. This is Al Rea, and this is the National Hydrography Advisory Call. Today we've got Jay Stevens with us from the Bureau of Land Management in the Oregon-Washington office, and before we get started, I do have a quick announcement to make regarding our staffing. Our co-lead, my co-lead, Becci Anderson, is back. She has decided to come back to USGS, and she's back in her old job. So I'm really happy about that to have her and have her back and really appreciate all the hard work that Steve Aichele put in as acting in that position while she was gone. So we're really, really happy to have Becci back. I think Steve is happiest of all to have her back, but I'm pretty happy too. So anyway, just wanted to give you that quick announcement, and now I think we'll go ahead and turn it over here to Jay Stevens, and he can take it away. Go ahead, Jay.
Jay Stevens Okay. Well, thanks for inviting me to talk with you all today. So I'm Jay Stevens. I'm a GIS Specialist in Portland, Oregon, working for the Bureau of Land Management, and we've been experimenting with delineating NHD and WBD from our good lidar that we have in Western Oregon for probably about 4 or 5 years now, but we're definitely ramping up our efforts right now to really try to get most of Western Oregon covered with new lidar-derived NHD and WBD, and we're working with other agencies. We're not the only agency that has an interest in getting this work done. We're working with the Forest Service, the Oregon Department of Forestry, Oregon Department of Geology and Mineral Industries, NRCS, but I'm going to kind of give you guys the BLM view of things and kind of go over how we got into this business in the first place, what our business requirements are and sort of how our process has evolved and how we're doing our interagency collaboration on this. So let me start by, I'll show this. This is a map of Oregon showing the ownership pattern, and the boundary that you're looking at is the boundary of the BLM's Western Oregon plan. Most of the activity that's happening in Western Oregon as far as land management goes, it's really centered around timber harvest. So there's some of the most productive timbered lands in the world. They're in Western Oregon. It's a pretty wet place in the coast range. The precipitation covered by this plan area ranges from somewhere around 10 inches of rain down here to about 200 inches of rain in parts of the coast range here. So it's quite a varied terrain, but our basic goals for our plan are, besides producing a sustained yield of timber to counties that depend on timber receipt, we have quite a few species of endangered salmon here. So if you look at the distribution of specifically listed species of salmon, it covers quite a bit of the area of our Western Oregon plan, and our management prescriptions here are pretty restrictive and very specific. Our primary concerns as far as the health of salmon goes is going to be stream temperature and turbidity, and the way that we preserve the health of these aquatic ecosystems is to make sure that we have a good riparian buffer around the stream networks. So there's an area where we really don't do anything. We don't harvest any timber within 120 feet from a stream and in the outer edges where do commercial thinning in the very outer edge, but even on intermittent streams, we have quite a bit of protection here. So when we're planning a potential timber sale, like this shows an example of ... The red area on the map would be areas where we could be harvesting timber, and, of course, the blue is showing the riparian buffers. The black lines are revised NHD after deriving flowlines from lidar, doing fieldwork. That really changes the picture of what's actually accessible in terms of timber harvest. So that has a major impact on our planning. This is our definition of intermittent stream. Really evidence of annual scour and deposition with well-defined bed and banks within a continuous channel network. So that covers quite a bit and particularly in the coast range. I'll show an example of what the NHD looked like on top of the yellow 10-meter DEM, and the kind of changes that are possible, the kind of changes that become evident when you can see through that heavy conifer canopy and see what's actually on the ground. So this can have a really profound impact on our ability to estimate what we have available in terms of timber and what we need to be protecting. So I'm showing here, this is the availability of high-quality lidar-derived DEMs that's available in Western Oregon, and this covers most of the BLM's ownership in this area. So obviously it's a lot of ground to cover. It's an expensive proposition to get this work done, so we had to go to management and explain that, "Hey, we have a wide variety in density here that needs to be corrected. Obviously most of our stream channels that are mapped are not in the right spot, so we really need to get to work on this, and rather be doing it incrementally through the spotty field work that's done, it would be much more efficient to just go ahead and get this done by delineating entire 12-digit units." So we were able to get some funding to kind of ramp things up and hire some contractors who were able to do modeling work and editing work, and so we can have a goal now to get all the work done where we have BLM ownership, where we at least have some BLM ownership in a HUC in Western Oregon and then work with the other agencies to really get all of this area done regardless of the ownership and do that by engaging the hydrologists that we have working in our various field offices and resource areas to make sure we get review of the work before it goes into the NHD. So as I mentioned, we've been working on our techniques to do this delineation for quite a while, but this is a workflow that we're basically using right now. We are resampling the DEM to 3 meter just to make the processing faster and applying the filter, the fancy filter that Paola Passalacqua advocates in her research. So we've borrowed pretty heavily from Passalacqua's research and some of the open-source Python-based code that's published. So we're using both curvature and flow accumulation as kind of a basic filter to identify the channel skeleton, and so we have a rough draft of the linework from curvature and flow accumulation. Then we do quite a bit of work with clean up and hydro-enforcement, incorporating whatever field data we can get from hydrologists that have done work out in the field, and then actually have the hydrologists review and mark up our draft delineation and do our final delineation based on that, estimate the periodicity, go over the imagery and the lidar and straighten out all the waterbodies and areas, do the actual NHD update, deal with whatever data we had that was attached to the NHD, which in our case we have a lot of fish distribution data that's referenced to the NHD that we actively use the HEM tools to manage our events. So when the NHD changes under that, we have to go back and synchronize all that data to the new line work, and finally we also need to do the WBD. So generally the WBD is easier to do once we already have the flowlines in place. So here, I'm just showing an example of what happens when you resample and filter the 1-meter DEM. You don't lose a lot. We're finding that we're not losing a significant amount of detail in terms of what it takes to get channel location accurate, and we do benefit from the filtering in that it's a little bit easier to get the streams to get across the roads without having to do hydro-enforcement, so it saves us a little time on that, but mostly processing time. So with this resampled DEM, resampled and filtered DEM, we can look at geometric curvature and flow accumulation and come up with basically what we call a skeleton of potential channels. The green is the flow accumulation, and the red is the curvature. So both of those conditions have to be met to have a stream initiation. So just having, say, 2 or 3 acres of curvature up here, that's not enough to initiate a channel. There actually has to be some evidence of a channel based on the curvature for us to delineate. So from that skeleton, we identify that ... We thin the skeleton and locate the end points. So these end points, even if it's just a couple of pixels that meet that criteria, that could be an initiation point for a channel and used on the least-cost path, we would trace to the pour point and create our initial network. But of course that still results in a number of spurious channels, and it still requires some visual interpretation to adjust the initiation points, take out the channels that we don't think should be there, particularly in these flatter areas. There's quite a bit to clean up. So generally using slope, hill shade. Also we've been making extensive use of topographic position index to do this visualization work and cleanup work, but we've also started using TPI more as a filter to sort of automate that cleanup. So I'm going to show an example of how our process is evolving a little bit here. So we're looking at sort of a problem area where there's local curvature here in this flood plain, and there's adequate flow accumulation for channels to be delineated there, but really these are kind of spurious channels that we would take out. But experimenting with the topographic position index, we can apply this as an additional filter, and you see that some of those spurious channels aren't there because we're requiring that the topographic position index threshold also be met for the channel to initiate. So once we've got our basic network, we still have a bunch of hydro-enforcement to do. This example shows cases where despite the filtering, we still are having problems with channels running along the roadside ditch before they get to where they need to go. So we're just doing basic drawing enforcement lines and just burning these into DEM and saving our breaklines. So these are pretty straightforward. As we get to working in areas where we have more anthropogenic disturbance, particularly the agricultural areas we're having to do a lot more work to get these streams across farmers' fields following imagery and evidence of ditch features and the DEM and doing quite a bit of this kind of hydro-enforcement along the way. Now, sometimes, when we're lucky, we actually have a decent amount of field-collected data from our hydrologists that are out there planning timber sales and restoration work and getting out there and locating these initiation points or intermittent streams, and this is just an example that's showing where we had quite a bit of that data, and in that case we'll just go ahead and use that information to adjust the network before we even give it to the hydrologists to review. So this example is showing the initiation points in green, and then these are locations or culverts in red, and this information can also be used for us to adjust, basically looking at a few examples, head water locations, we can adjust our curvature threshold that we use or flow accumulation threshold that we use for general initiation to get the model more calibrated to the field data that we have. So once we have our final ...
Al Rea: Say, Jay.
Jay Stevens: Sure.
Al Rea: Yeah, I got a question on that last slide.
Jay Stevens: Mm-hmm.
Al Rea: So you have quite a few culverts there, the red marks, where there's not a channel indicated.
Jay Stevens: Right.
Al Rea:You don't require there to be a channel for a culvert, or just how do you handle that?
Jay Stevens: Well, it's kind of tricky business with the culverts because they are ... Let's look at an example here. These are going to be cross-drains. So they don't necessarily indicate a stream crossing, but they might just be for basically handling storm drainage along the ditch. So it's kind of a tough call. Some of them do represent stream crossings, and some of them don't, and usually there is some information that the engineer that went out and drove the road will indicate whether it's a cross-drain or if it's a stream crossing. And sometimes they even have some notes about periodicity or whether there's water or not. So we do use that information.
Al Rea Okay. Thanks.
Jay Stevens Yeah. So once we have our final stream network ... Or actually, I shouldn't say final. We've done our hydro-enforcement. We're ready for the hydrologist to do his final markup on it. In the cases where we don't have field data for periodicity, we are using Shreve stream order to estimate the periodicity. So the Shreve stream order is different from Strahler. It's just a straight additive stream order. So I think in this case it's going to be a stream order of five. It just adds up the tribs. So it's kind of a better proxy for flow, and we didn't want the periodicity breaks to be in the middle of a flowline unless we actually had field data indicating that. So that's what we're using for our basic estimate of periodicity. So we developed a web app for hydrologists to review line work, and we've tried to make it as simple as possible for them to do that. If they want to bring the inception point down, they'll just put a yellow point where they want the adjusted initiation point to be. If they want to get rid of the trib entirely, they'll put a red X where they can add an inception point or a periodicity transition. If they want to actually reroute something, they can draw lines to add and delete. So that's working pretty well. We support this either in an MXD or the web app, and we also can make this available for external partners to do the same review. So once this review is done, then we can take these adjusted initiation points and basically just by pushing the points around and redelineating using least-cost path again, now we have our final stream network that's ready for the NHD. So we'd pass this on to our editors to go ahead and start getting this into the NHD, but the editors have some other work to do. Doing any modeling from a DEM doesn't include diversions. We're modeling a dendritic network, so if there are features, particularly when we get into the dryer areas and we're dealing with irrigation, these have to be added in by hand. So we're trying to definitely make sure that important irrigation features are included. Other work that the editors are doing is, they're scanning over the imagery and making sure that all the waterbodies are digitized correctly and also reshaping the NHD area features. So this is showing an example of an NHD area feature that's been reshaped, probably looking at contours and looking at the lidar, looking at the imagery and coming up with a better double-line stream to represent this river. One problem that we've had here is that if the DEM hasn't been hydro-flattened, then you're going to have a pretty erratic path that gets modeled through this wider channel, and really, it's pretty meaningless information because this isn't the green lidar. It doesn't penetrate water, so really, the meandering within this channel is just pretty much random. So what we're starting to do is if we really see this, and it's messy, we're just starting to do some basic hydro-enforcement along these wider rivers. So that's what we've done in this case. The yellow line is showing the hydro-enforced line. So one of the major difficulties in doing the NHD editing here is that we need to preserve the reachcodes, and we need to preserve, basically, reachcodes and GNIS names, and we need to preserve reachcodes in the cases where we have fish distribution information because we have multiple agencies that are referencing these reachcodes just to attach linear events to the NHD. So when we're going in here to do these edits, we're making sure that we do a geometry replacement while saving all the attributes on these reaches. And for the major streams and major fish-bearing streams, it's usually pretty straightforward. There's not super radical differences in the geometry here, but we do have the occasional difficult situation where it's debatable whether to save the reachcode because like in this example, this stream basically doesn't exist. It actually goes this way, and there's a bunch of conflicting reachcodes. So that's kind of part of the difficulty of editing is deciding what to save and what to throw out. For all the minor tribs, we're not saving the reachcodes, and that's mostly because with the lidar-based delineating, there's really not enough spatial coincidence that I feel like it's really worth the effort to save those reachcodes. So some of the subbasins in Western Oregon have a lot of flowline. So we have subbasins that have 100,000 flowlines in them, and running the QC on the regular NHD update tools has been a really time-consuming task, and it's really kind of extended the time that it takes for us to do edits. I wish there was a way to do kind of isolate the editing just down to a six-field HUC, but at any rate, the tools are working pretty well, and they've been getting good support from our POC, and it is working, but it is a pretty slow process. So now I'll touch a little bit on the WBD. Now, our WBD boundaries are in pretty decent shape, so the change has generally been pretty small, but we do occasionally run into situations like this where obviously this drainage belongs in this HUC, not in this one. You can see where the error originates here, and I'd like to be able to just say, "Let's just model this and use this line," but as with the NHD modeling, with the WBD modeling you also have interference from roads, et cetera. So you can see how this kind of got messed up by this road. So we're going to have to go back and stick to the contours to really straighten this out, but I think the modeling is pretty effective for ridgelines where there aren't any roads. So in this case, we'd probably just go ahead and use that modeled line and maybe generalize it a tiny bit, but just use that. So I want to focus a little bit on: One of the benefits of doing this work is that for each of these HUCs, we now have a lot of data. We have our DEM clicked out to the watershed. We've got a hydro-enforced DEM and flow accumulation, cost distance, and we've got data that's ready to do other kinds of modeling work. So one of the projects that's come up recently for me is starting to look at prioritizing road restoration or culvert replacement. So it's been kind of exciting because in the past because the flowlines aren't accurately mapped to the DEM, and now we have good flowlines, I can actually identify where the road stream crossings are exactly, and I can delineate catchments and collect statistics about this catchment. I can take the original 1-meter DEM that was not hydro-enforced, and this example is showing depressions that I'm capturing from the 1-meter unenforced DEM, so I can calculate things like fill volume and fill depth behind this culvert. So we can identify places that are at higher risk of culvert failure and road blowout that could deliver large amounts of sediment to this stream here, which happens to contain multiple endangered species of salmon. So this is going to be really helpful for us to further our efforts to protect the fish. So one of the complications I'm going to move on to talk a little bit about, the interagency collaboration piece here: With this kind of complicated ownership pattern, I'm showing an example of ... Let's see, in this HUC, I've got BLM ownership. I've got Oregon Department of Forestry. I've got Forest Service. We got some tribal lands down here. The ownership pattern is crazy in Western Oregon. Private land, private timber land here. So there's a lot of stakeholders within one particular HUC. So in order to coordinate this effort, we've had to develop an application just to track everybody's progress, track what everybody is doing. So I'm going to try to get this thing out of the way here. Okay. So this is showing the status of our delineation efforts, what's complete, what's in-progress, various stages of progress and what's planned for this year. So we've got multiple agencies involved. So I can do ... This app is going to let me do things like look at which ones are BLM, which ones are Forest Service HUCs, Department of Forestry. So a lot of the times, we're working in close proximity, and certainly working within the same subbasin, which is the unit of transaction or checkout. So this is helping a lot because all the agencies that are involved in the effort have the ability to go into this app and edit the HUC polygons and keep everybody else informed as to what's happening there. So we're tracking that both for the NHD and also for the WBD progress. Oh, whoops. I was going to try to go back to my last slide here, but I think I can go ahead and take some questions.
attendee:Is the markup tool being used for this or just a general bulk conflation?
Jay StevensThe markup tool?
attendeeYes, sir. Is that what's being used to edit the NHD, or is it just a bulk conflation?
Jay StevensWe're using the NHD update tool.
attendeeOkay, thank you.
>> The regular editing tool. We have experimented with the conflation tool in the past.
Al ReaSo, Jay, I think he was asking about the USGS markup, the web markup tool that we have. It's kind of a new thing.
Jay StevensRight, right. There has been some use of that here, and I think we're still sorting out how to process updates from that, but the markup tool that we're using is specifically designed to markup the lidar delineation. But to the question of the conflation tool, I think if the conflation tool had the ability to do partial conflation, in other words, if I could say, "Okay, for this list of reachcodes, I want to do conflation," I think the tool would be, for our specific project, it would be more useful to us if we could do that.
Al ReaSo there was a question asked through the chat, "Did you use any kind of automation to delineate waterbodies?"
Jay StevensNo, we haven't done that. Basically we're just scanning through by tile, just looking at the imagery, looking at the existing water bodies, using a combination of imagery and looking at the lidar to identify new waterbodies, but if there were techniques that are being used for that, I'd definitely be curious to know.
Al ReaSo ...
attendeeIs there a direct ... Sorry.
Al ReaNo, go ahead.
attendeeIs there a direct transfer of the Shreve stream order to the NHD feature codes for perennial intermittent ephemeral classification?
Jay StevensYeah. In the places where we're modeling, the Shreve stream order that's less than whatever the threshold is for that HUC is going to go into the NHD as an intermittent stream.
attendeeOh, got you.
Jay StevensYep. Yeah, we felt like it was important to have periodicity on the NHD even if it's not all field verified because I think very little of it is across the country.
Al ReaRight. So you're using different thresholds in different HUCs or different areas?
Jay StevensYep. We are because we have such a variety in precipitation, and then we're also getting feedback from different hydrologists that are working in the areas that they work in. So basically, we're doing a consultation with the hydrologists to determine, what's the best threshold for that particular HUC given the soils and precipitation and so on?
Al ReaSo, Jay, I noticed you talked a web review tool that your hydrologists use that you showed that.
Jay StevensUh-huh.
Al ReaThey use that in the field, like on a handheld unit, or is it just in the office? Or ...
Jay StevensWell, they're not. I think that theoretically, they could. We do have some applications that we've been rolling out where hydrologists are using a mobile app to collect stream-initiation points, but generally they would do that ... If they have that information, they would give that to us as we begin the delineation.
Al ReaMm-hmm. Other questions?
attendeeAl, I've got a question. You guys have pretty good relief out there, but I'm curious what you do or what happened with the lidar data when you call the roads or whatever is in the way of the streams, digital-type dams. How did you bust through that in the lidar?
Jay StevensJust drawing. Basically the hydro-enforcement technique is very simple. It's just ... I'm trying to get back to the slide here to show an example. Backwards again. So in this case, just to look at an example right here, it would take this line and find the minimum elevation on this line and make all the elevation values equal to the minimum elevation. So it's basically a trench. It's digging a trench in the DEM.
Al ReaSo you put those lines in manually?
Jay StevensWe put those lines in manually.
Al ReaOkay.
Jay StevensWe have experimented with some automation on that. One way to do that is to basically identify the sink, then identify the lowest point in the sink, and then go in a radius out from that and find the next pixel that is a lower value than the center of the sink. We have used that technique, and that works pretty well, but we still have to go through and identify the ones, the sinks that we actually want to do that on. So by the time you do that, you might as well just draw a line.
Al ReaMm-hmm.
Jay StevensSo we're kind of just picking the technique that's the quickest to get the job done the fastest.
attendeeSo a question here. This is Sean Vaughn from Minnesota. Is a culvert there at those locations, or is there not a culvert in those examples just used on-screen with the red lines?
Jay StevensWell, I would speculate that the vast majority of situations, there's a culvert, but we don't know that. We don't always know that. That's what we're assuming. So there could be. There's certainly room for error.
attendeeYeah. It just seems to me like you've relied heavily on a lot of input from hydrologists, which I think is incredibly important in a process like this.
Jay StevensMm-hmm.
attendeeA task I would put upon them is identifying where the culverts are. We do a lot of similar work here in Minnesota.
Jay StevensRight.
attendeeAnd our value is on the culvert, and then putting what we call a digital dam breach line where the culvert exists. And then from there our process is similar to yours ...
Jay StevensRight.
attendee... lowering the lidar-obtained elevation values ...
Jay StevensYep.
attendee... to the minimum value of where that culvert is.
Jay StevensRight.
attendeeWhat we then get is a more in-sync thought as to where our flow paths are now in sync with real-world locations of where culverts are, and it brings reliability and credibility to our work for hydrologists as well as end users of the product.
Jay StevensRight. Well, certainly, in the areas where we actually have culvert data, we'll bring that up and use it, and even in that case, if the culvert data was collected before GPSs were out there or before the GPS accuracy was good, sometimes they're actually pretty far off. So we're having to look at that data and say, "Okay, is this culvert actually here, or should we look at the sink in the DEM and put the culvert there or put the hydro-enforcement line where it actually makes sense on the DEM?" So it really varies a lot, and unfortunately because we're doing all the ownership ... We're not just doing the BLM streams. We're doing the entire HUC, so a lot of times the BLM ownership is only 20 percent of the HUC or less. So we don't have any data for the rest of the HUC.
attendeeI had a question, Jay, about whether you overlaid it with your 1-meter and how well it matches since you're developing the lines off of the 3-meter?
Jay StevensRight. Well, one thing that the 3-meter combined with the filtering that we're doing, it's going to take out some of the small ditches on the edges of the road, and that's actually a good thing for just getting the stream delineation done because you'll have more circumstances where if you had a really small sink here, the stream will actually just go straight across the road, and you don't have to burn it in. So as I was showing ... Let's see. I'm going to show my ... With this slide, what I'm showing is these that I'm showing in red actually have ... I have depressions thematically displayed by still depth. I'm deriving these from the 1-meter DEM that hasn't been hydro-enforced. So I do think that that information is really helpful to have and to keep, especially if you are doing this kind of analysis, which is basically, I'm trying to identify the potential for a backup on the other side of the road. So with the 3-meter DEM, no, I'm not going to get that. I've lost some of that detail, but again, I can go back to the 1-meter DEM and extract that if I want to, which I did want to do for this particular application.
attendeeI had one other question about your reach codes. You said that some of the reach codes you weren't transferring over. Are those just getting lost and assigned new ones when you're doing the update?
Jay StevensYep.
attendeeOkay.
Jay StevensYeah. I think, to go back to this kind of radical example here, not all ... This example is in the coast range where the trees are probably 200 feet tall, and you would never see ... The original method of compiling, the original 24K maps, you'd never see the ground. Yeah, I think the problem is that for just with this stream, for example, if you were going to say, "Okay, I want to save the reach code for this." The question is, well, is it this stream or is it this stream? Is there really a point? At this point, I don't feel like there's any benefit to save the reach code there, but for ... Go ahead. I'm sorry.
attendeeI was going to ask if you try and overlay your events and try to save those in particular.
Jay StevensExactly. Yeah. What we would do is say, "Okay, we've got multiple agencies have fish distribution information that's attached to this stream. We want to save that." And we're definitely saving the reach codes where there's a GNIS name, too. So if there's a GNIS name, definitely saving the reach code. If there's fish distribution, definitely saving the reach code, but for the vast majority, no, we're not going to try to figure out which one of these streams belong to this original line. And I think that's why ... Like I said, I think we'd be using the conflation tool if there was ... If the conflation tool could do a partial conflation and say, "Okay, I know that we want to conflate this line, but we're not going to do these." I'd love to figure out a process that would be more efficient to get that work done.
Al ReaSo there was a question from the chat, "Are you working to a particular specification for positional accuracy?"
Jay StevensWell, that's kind of a tough question. I guess not really consciously. I think the level of positional accuracy, looking at this slide again, we're working within pretty ... These are kind of distances that are really important to us. Resampling the DEM from 1 meter to 3 meters while it's not ideal that we're doing that. It's kind of a compromise to improve the performance of all the modeling software and the ability to get the work done, but we don't feel like we're losing a lot of positional accuracy by doing that. So positional accuracy is really important because we really need to know where these buffers are and how this is going to effect our planning for timber sales.
Al ReaBut I think what you're saying is that the positional accuracy you're after is quite a bit larger than 1 meter or 3, even.
Jay StevensWell, I'd say the 3-meter would be where we want to be.
Al ReaMm-hmm. Any more questions? This has been really great, Jay. We definitely will want to pick your brains some more about this as we ... We're in the midst of developing specifications for lidar-derived hydro.
Jay StevensUh-huh.
Al ReaAnd I think we'll definitely want to loop you in at some point to take a look at at we've got.
Jay StevensYep. Well, I think ...
attendeeIs this call being recorded?
Al ReaIt is, yes. Yeah, so if you want the link ... If anyone on the call wants a link to the recording, let me know. I don't know that we'll be able to put it on our website or anything because there are a lot of requirements there, but I'll share it if you send me an e-mail.
attendeeOne last question, Jay. Are you doing anything with the culvert break lines that you digitized?
Jay StevensPardon me?
attendeeAre you doing anything with the culvert break lines that you use to enforce with the hydro-enforcement?
Jay StevensWell, we're definitely saving them so that we can reproduce the stream delineation, and we're saving the hydro-enforced DEM. So, yeah, this is definitely part of our project data that we're going to keep.
Al ReaYeah, actually, Jay, looking at that list, I was very curious. Are you sharing any of this, or you're keeping it on file, or what's the status?
Jay StevensYeah, we're building out some data structures to handle this. We're kind of debating whether to mosaic some of this stuff together. This is a little bit of a work-in-progress, too, but this is basically what we're saving right now along with a little bit of metadata, recording the edits that the editors made to water bodies in the areas and other flowlines that they digitized, the draft initiation points. So ... Whoops. So that we know, what did the model predict versus what was the final based on everybody's review? So we felt like those were important things to keep. But, yes, the breaklines are key.
attendeeAre you finding any trends comparing your model initiation points versus the fielder and the field information that you get from hydrologists?
Jay StevensWe haven't done any large kind of meta-analysis of how we're doing there. We generally eyeball things when we start on a HUC. We just tweak the parameters of the model so that it matches the initiation points if we have them. Otherwise, we're pretty much reliant on the expertise of the hydrologists to make the refinements, but I think we will ... We're actually, now that we're acquiring quite a bit of data, I think it's kind of ripe for that kind of analysis.
attendeeThis is Sean with Minnesota again. What you're touching on now is, I think, one of the greatest values to all of us across the country. This is something we strive for in Minnesota, and what I'm getting at is: When we save point of initiation, I think we really tend to talk about, where is concentrated flow forming on the landscape to a level of ferocity and velocity in which it forms erosion. And lidar is a technology that identifies and captures what I call water-conveyance landforms. In other words: Water formed this erosion and the landforms captured by lidar.
Jay StevensUh-huh.
attendeeI'm always intrigued by all of us in these conversations about where we model that point of initiation and where someone out on the landscape tends to put it, and sometimes they just don't even jive.
Jay StevensRight.
attendeeThey don't match.
Jay StevensRight.
attendeeEarlier in your process or your presentation, you mentioned you use TPI, and that's something we rely on heavily here. I actually call it HPI. In other words, it's using TPI but to really identify hydrography. So hydrographic position index.
Jay StevensRight.
attendeeI would ... In Minnesota, we tend to see a greater ability of our HPI/TPI to identify that point of origin more so than we could model it, and more than where someone could actually be in the landscape at time if it's a very shallow bed and bank definition. So where I'm going with this is to say that, as you write this up, and this is something I sense all of us need to collaborate more on because it's so critical and this role of understanding where does water begin to flow in the landscape and how does that relate to our various definitions of, what is a water course?
Jay StevensRight. Right. Yeah. I think in the case of Western Oregon, we were able to come up with a pretty good consensus, even though this definition comes right out of our plan document. The other agencies involved, the Oregon Department of Forestry, the US Forest Service, because we're all doing the same business in Western Oregon of the timber management and the protection of the salmon, we're all kind of on the same page in terms of our business requirements and therefore on our definition of a stream, but this would change, I think, if we start working in eastern Oregon, it's not going to be the same what people want to see mapped. I think people are going to want to see ephemeral draws that only have water every couple years, for example, but this definition explicitly says annual scour and deposition. So, yeah. I think that's the rub. What are we mapping?
attendeeYeah, I appreciate your definition. I think that works well in many parts of the country, certainly here in Minnesota. It is something that ... Well, I'll just stop there for time's sake. This could go on, or at least I could go on.
Jay StevensSure, yep.
Al ReaYeah, we are kind of out of time today, but we could always have you back, Jay. We really appreciate your time on this and presenting.
Jay StevensSure. Yeah.
Al ReaIt's been very informative. So thank you again, and will you make your presentation available?
Jay StevensSure. Yep. Yeah, I can do that. Do you want me to just e-mail that to you?
Al ReaSure, yeah, and I'll put it somewhere that people can get to.
Jay StevensOkay, and any other questions that didn't get answered, I'm always happy to have a conversation. So feel free to shoot me an e-mail.
Al ReaGreat.
Jay StevensAll right.
Al ReaWell, thanks again, Jay.
Jay StevensYeah, thank you. It's been great. Good questions.
Al ReaYeah. Thanks, everyone, for attending. We'll talk to you again in about a month. Thank you.
Jay StevensAll right. Take care.
attendeeThanks, Al.
attendeeThank you.
attendeeThank you.