NCCC 2022 — Emerging Technologies for Coastal Change
good nice um and Ryan's going to talk to us about lidar scanners for continuous based continuous Beach morphology observations um so can I just get a confirmation that you can hear me over there we can hear you yes okay fantastic so thank you in front of you what you I think we're looking at though sometimes video does not work so well and PowerPoint was a sort of time lapse of the field site where we did a deployment earlier this year uh on the Outer Banks to test out some new lidar scanning technology during storm impact um and I don't think the video is playing but uh let's see if I can at least Advance here we are it probably doesn't look as good um over Zoom I'm sure it's quite jumpy but rest assured uh what we're seeing here is a wave run up during high tide which is impacting a a dune or what I like to call the air quotes Dune Dyke system um in front of this house here so I'll just talk today about a deployment we did on this particular house and in this particular region as sort of a miniature small scale test of some pretty cool new 3D lidar scanning technology um so the reason we chose this particular field site which is denoted here in the red circle along the Outer Banks but because this stretch of coast is eroding quite rapidly uh in in this circle area the field site was located in Rodanthe North Carolina the average latest average yearly annual erosion rate is about five and a half feet per year but over the past couple decades that has been accelerating from about 20 years ago being about two feet per year uh as most of you probably know OverWatch is a common thing that occurs along this stretch of a highway particularly the s-curves on nc-12 and so recently The Jug Handle Bridge as it's called right has been open to bypass this over washing region so that may have more significant impacts down the beach so there was a storm uh that was forecast to approach and pass near the Outer Banks of North Carolina as indicated by this forecast screenshot showing a forecast predicted simulated radar um and so it was going to be a pretty large low pressure system so the near shore extreme event reconnaissance or near Association decided to deploy a field mission in The Weather Channel termed this winter storm Kenan our report for this and our preliminary data report and uploaded data it can be found on the natural hazards engineering research infrastructure online portal this is a picture a few days prior to storm impact here we can see a pretty steep beach and a scarp here as well as a dune or Dyke scarp which existed from a prior storm about two months before this study this is again the region showing here some forecast water levels so there was about two days predicted of higher water levels not quite necessarily up to what was predicted to be the elevation up here in that area so it was going to be an interesting storm to investigate because it was not expected to be major overwatch but perhaps there would still be lots of coastal erosion uh we also looked at a lot of the data from a lot of sensors around the region to get sort of a larger macro scale view of just how the entire region was being impacted so these are some water level observations I just want to put into context here sort of what this region of the coast right had been experiencing throughout January and February of this year so these are plots of water levels from three different stations along the coast one is Oceanside and two are more sound side so to speak um and uh here are three events uh where there was a pretty significant say offshore storm uh impacts in the month of January and then you have highlighted winter storm Keenan so we see over the month of January particularly on the South Side locations the water levels were almost always higher than the forecasted water levels uh being blue measured being green if we take a look also at some of the wave data at this buoy and the wind and meteorological data from this station near Oregon Inlet where our field site is sort of right between those two stations we see that these three storm events consisted of about 30 to 40 mile per hour gusts uh knots of gusts and about 25 to 35 not sustained wind speed um and it's notable that these two in particular occurred right before winter storm Keenan so that this area was already really vulnerable because there was a large waves almost 20 feet in significant wave height um just about a week or two prior to winter storm Keenan uh impacting this area I also just want to point out the wind direction here so here during the large events in the middle of January with the largest wave heights we see the wind direction coming around that time out of the East which would be onshore um whereas during winter storm Keenan for the most part the uh wind speeds were from the north so sort of from north to south in the Long Shore Direction so here is our deployment schematic it was quite an extensive deployment I'm only going to focus today mostly on lidar data but we installed two lidar scanners on this house here some pressure gauges on the pier to measure waves and water levels we also had cameras one two and three monitoring um in real time the impacts of the area and the water level sensor here installed on this house here's a picture of the wave gauge installed on the uh up here and then this is one of the two lidar scanners that was mounted on the deck on the house South adjacent to the pier um this is just the wave and water level of data that was recorded by the rbr uh solo D wave gauges um and so in the future we will process this data to sort of uh time correlate the impacts in culsa erosion uh from the Morpho Dynamics and tie that into the hydrodynamics however for the sake of time in this presentation I will not cover any of those details further we captured the morphod Dynamics of uh the beach in this particular area using two compact low-cost 3D lidar scanners this is the Blick felt Cube Cube one model they're about five thousand dollars each I just mentioned that because that's relatively inexpensive compared to some of the lidar scanners in the hundreds of thousands of dollar range uh they're also really cool because they can log be logged autonomously and so we just run python code or some other other API scripts uh to log the data and control the sensors continuously so we captured high resolution scans every four minutes for about 3.5 days which consisted of before during and after the storm uh we would have had two weeks of data but I made a silly mistake and bought cheap USB drives that I thought were two terabytes and turned out they were about eight gigabytes uh so we ran out of data after about three days so this is just a picture of the setup that we had on the thing of course there was a lid over this um and so I'm going to focus on analysis just on L1 today which was on this corner of of the house but this just gives you a quick sort of overview or picture of what the data kind of looked like before I jump into Data analysis um and so here is the pool wall looking out over the deck and then in the lidar scans here if you look closely perhaps you can sort of see the outline of this pool wall there and so this is what we're capturing with those two lidar scanners and so we get this high resolution field of view every four minutes we get a scan like this but by the way it only takes about uh 10 seconds to complete this entire scan uh and then this is what the beach looked like after the storm so uh we need to geo-rectify the data uh so in order to do that we installed or laid out six Ground Control Point targets which look like these little white cylinders here uh and across and throughout the you know the view of the lidar scanners and then we survey those with rdk GPS and then we can georectify the uh XYZ Point cloud data into uh stapling coordinate systems and then we need to validate um whether or not our georectification was actually done appropriately um and so we also took after the storm seven post storm beach profile transects we ran out of time to do those pre-storm um but we still have the Prestone lidar data and so then what I'm going to show you today is a sort of just an overall comparison of how that survey data and georectified lidar data compare with each other by the way it was very cold as you can see by these images here during this rapid response um so we carry out this sort of sequence of filtering so we filter the lidar point files we have to remove the waves water non-land objects like people fence posts and birds and so while you did see this very nice large triangular field of view from the previous lives of the lidar point clouds we end up removing a lot of that data to just give us the actual dry beach we also removed the scar from beach profile surveys as it was not so reliable and then we removed the Dune as I call it the air quote "Dune" from the lidar point clouds as well as anything landward of the dune then we just interpolate the lidar data and the beach profile survey data onto the same grid with about a half meter grid spacing dxy and then we use inverse distance squared method to give us an interpolated grid so this just shows you sort of a 3D view of the beach profile transects that we measured um and then the L1 lidar field of view with the magenta dots are the filtered point cloud data and then here is our lidar surface the interpolated survey surface is not pictured here but would overlap across this entire span so then we're going to look at transect four here to see how well the lidar data compared with the surveyed uh grid interpolated grid so on the top what you're looking at here is on the x-axis you have cross short coordinate and on the y-axis is the longshot coordinate so we're looking from above and here is the region where the lidar and survey data overlap with each other and this is transect 4 again just to orient you so offshore is over here to the right so this is just a difference map of the overlapping survey in lidar interpolated grid and across this entire region we have an average mean square error of about seven centimeters and a mean bias of about negative five so that's just the difference between the lidar and the survey now if we look just at transect 4 which is plotted here on the bottom now in profile view and we evaluate the statistics just in the region or the span where we have lidar data overlapping with survey data we have an improved average root mean an improved root mean square error of about four centimeters and a mean bias of around three and a half negative centimeters so I think that's relatively good given that the overall error between the two different survey methods uh vertical error is in the range of a couple of centimeters and I just want to wrap it up and end here uh we've been now working on analyzing the filter data but prior to that it did make an animation of the unfiltered uh gridded surfaces so here you do see a bunch of noise from the waves coming in but there was the erosion of the beach uh uh profile during storm impact and then during the next high tide we see some accretion here and then the next higher high tide we see additional erosion and landward transgression of the beach profile those were some people walking through the stand that we had not filtered out yet and then if we take trans X4 and look at the time stack what we can start to see is with this kind of data sets and with this technology uh of course this is just three and a half days you can imagine here we see one two three four five six high tides coming into our field of view we also see this land where rapid landward transgression and erosion of the beach profile and the start and then during the next higher high tide again we see little bit more erosion and this grid each grid is half a meter so there was only one meter remaining of uh this dune before the pool will at that particular home was going to be impacted by waves so this is uh uh a three-dimensional view of that same time set just to give you a sense of what we're able to resolve and if you look here after these uh second higher high tide where the additional erosion occurred you can see in subsequent high tides slight accretion on the beach just uh at the landward edge of the beach of course this was sort of the beginning of what became a accretion over time uh and if we look at images that I took in the end of February uh versus the end of August both photos were taken at low tide um here we see people way way out in the ocean so it's quite shallow that far out versus here which was a couple of weeks after the first house of what became three houses that collapsed um and so there was quite a significant difference it doesn't look like this today however um and so I would love to move in in the future try to apply this technology to more continuously monitor Beach morphordynamics over more than just three days because as we all know the beach is a very dynamic system so thank you and I will take any questions [Applause] Ryan you did an excellent job on your time and you've got seven minutes for questions so um we have a mic set up back here but I'm going to bring this to Barbara Doll so that she can answer her question ask her question Ryan I'm wondering if it would be good to just maybe have like a summer snapshot and then a winter snapshot on a non-storm day and kind of look at this over time because it seems like you're going to generate a lot massive amount of data so like if you have continuous over time I'm just kind of thinking of what the use application of this would would be and how to do it like practically that's a great question and I think uh part of the initial goal of this was really just to prove the concept uh and approve the sort of methodology and analysis methods uh but but moving forward um the fantastic thing about this particular scanner model is that it's very very configurable uh and so depending on what you're interested in if you're interested in just a single transect you can program it just to conduct a scan along a single transect uh you can also program it to do a scan once a day once a week once per hour um or you can even remotely talk to it and say it set it to do a scan every hour or once a day at low tide but if this storm is coming then you can reprogram it to to start scanning once per hour once every 30 minutes and so there's a ton of flexibility because sure if you're scanning every four minutes for an entire year it does become quite an extensive data set but um yeah maybe hopefully kind kind of answer your question so I I think there is tons of room for like practical applications it just it comes down to having partners who have power supply she said that did answer her question we have plenty of time for question two we have to wait until um and we get to like the 20 minute mark before our next speaker goes yeah right uh this is Ling I'm from Coastal Studies Institute and we are also using lidar and we have grand prix lidar from regal is called basic 400i so first um for those lighter sensors you use what's the range limit I say it's a short profile so what's the range limit for your lidar sensor? Ah that's always a loaded question right uh so the manufacturer states a range of uh 250 meters but that's rarely the case particularly when you're outdoors in sand and water uh we've uh a student of mine has also that this past summer at the FRF tested the entire summer different configurations um on one of their lidar towers and I I think during lower low tides there was a range out to about 50 to 75 meters um it's it's a little bit less dense at that point but uh the actual point cloud on this particular slide um uh would kind of come back out and then sort of re connect over here um but it just doesn't get any return off of the non-white water that far away so here it's around 25 to 30 meters in the crossword but that's what the reason you're only seeing that is because the water at this particular location was only 25 meters away from this uh person's house uh but I think up to maybe 75 meters is what we've seen at the FRF okay yeah thank you and you can even go uh to the website of blickfeld um and there are a couple other graphics and figures on a case study that I did with them um which shows some graphics uh from the FRF which will give you maybe a better visual sense yes uh thank you but uh I have a lot of questions okay because let's see if there's another are there any other questions okay yeah so um since you was so curious about the accuracy of those um uh lidar sensors I'm I'm also interested in like we can bring our Regal visit 400 I understand the same area and let's see how does it work what's the uh what what does the result look like because we also have a very interesting site in your density now in the it's in most part of the United States there is a very uh frequently overwatched area we bring our scanner to scan those overwatch area again again so I I again I think so over this summer at the FRF um maybe you're aware that they have uh I don't remember the model of Regal lidar scanner that does a full scan across the entire FRF they have multiple and so part of that study was also to look at how accurate this model was in georectification versus the previous model we're still improving the georectification methods right now we're just using a sort of camera based uh rectification approach uh and skipping the intrinsics so we just calibrate for intrinsics but I'm sure there are better approaches to improve the accuracy as far as you know Point Cloud registration and all this other kind of stuff um but certainly more comparison data sets between this model and other models in all sorts of different environments would be a good idea all right thank you so much yes hi I am Logan Howard I am a senior undergraduate at the University of Nebraska in Lincoln um but this was a research project I conducted as a part of my Noah Hollings internship at the weather forecasting office in Newport Morehead City North Carolina so first to start with a little bit of background um so for weather models currently they are ingesting sea surface temperature grids that are a very coarse resolution and that is of particular importance here in North Carolina because um just off our Coast we have the Confluence of the cold Labrador current and the warm Gulf Stream current that creates a very sharp sea surface temperature gradient that the models are not capturing effectively this gradient is particularly enhanced in the cold season and here I have an example from February where we can see that just off our Coast in the Marine zones here that there is a nearly 30 degree Fahrenheit temperature difference across a distance of just 60 miles and this is pretty significant and another thing to note is that this is of particular uniqueness to the Newport region which I have here in the the black squares um so right in our region is where we have the confluence of these two ocean currents so the objectives of this um this this project was the overall goal was to help improve our marine wind forecasting and of course that will help improve our marine hazards like small craft advisories and Gale warnings as well as eventually improving our wave and rip current forecasts and the way this will be done is we built a tool that will incorporate winds and sea surface temperature observations to help correct and make better Marine wind forecasts what this means scientifically is that we are going to determine how well winds are mixing in the Marine boundary layer and just for context the boundary layer is the lowest approximately one to two kilometers in the atmosphere and this is the region where the winds are under the influence of the friction by the surface so in it we have mixing nuts occurring throughout the day and we can approximate this based on the lapse rate which is the rate of change of temperature in this layer by using the sea surface temperature and the temperature at the top of the boundary layer so we will catalog all these relationships and any additional variability by other things and put that into a climatological catalog that is put into this tool so quickly going back to the Sea surface temperature maps here on the left I have an example of a sea surface temperature grid that is being input into a weather model and on the right I have observed sea surface temperature which is a blend of buoy observations and remote sensing satellite observations so the first and most important thing that you can see is that the gradient is much larger and spans a much uh larger space than uh what was observed another thing to note is that the sounds are actually 10 degrees warmer in the model than it was in observations and that's going to have a big impact uh one last thing to know is that the um the bay is Onslow Bay um these are really shallow continental shelf Waters and thus these waters are a little more influenced by seasonal variability and thus you can see in the model they're just a touch warmer than what was observed and one last thing to know is that um this one is one example from one weather model but all high-res American models are using a sea surface temperature grid that is of this quality or similar so this problem is not unique to just one weather model so the observations that were collected um we took sea surface temperature and as well as surface wind observations from two buoys diamond shoals which is located just off the coast of Cape Hatteras and also Onslow Bay to the South and this was obtained from the national data buoy Center and um characteristics from Aloft we're taking from a radius on soundings those are occurring twice a day at the Newport Weather Service office which is located in blue and this data is stored in an archive on the Iowa environmental Mesonet website data was collected over a span of four years and if any sounding or buoy data was missing the entire time slot was scrapped just to make this easier to work with and while we did have to remove quite a bit of data we um still ended up with between 1500 and 2000 data points and just to clarify this it does span all four seasons and since soundings are taken twice a day we have morning and evening observations so that's twice a day so the first thing we wanted to make sure is um that there is a relationship between the wind speeds at the surface and the wind speeds Aloft at the top of the boundary layer we chose two different um layers to represent different heights of the boundary layer and what we found is that across the board we do have a very nice even normal distribution which is good that means there's a pattern to this and that this can be easily predicted but you will know that there is a very large tail that extends out to the right here on the on the x-axis and you'll see that this ends at 2.5 and that was a deliberate choice um because these wind speed ratios that are very high so 2.5 or greater um these wind speed ratios could be attributed to small scale things that were not representative of the overall flow of the atmosphere these could be a lot of different things these could be fronts this could be related to storms tropical cyclones there are a lot of different things but the point is that these are weather events that another tool is not designed to capture so these um these instances were not used in this tool I have a couple examples here just to illustrate so here we can see that we have a stationary front bisecting the region and what's important here is that the sounding location on the land is on one side of the front while the buoys are on the other side of the front so as you can imagine there are major differences here in the wind speed and direction on either side of this front and so that's going to affect that's going to affect how the mixing is occurring in both of these regions and we uh the tool is just designed to look at the whole region the region as a whole with similar characteristics here I have an example of a nor'easter so you can see there's a lot of boundaries a lot of fronts we have a large cyclone occurring and this is just not um not what the tool is designed to look at so now that we have all of these wind speed ratios we want to tie that back now to temperatures specifically the lapse rate so I'll explain this diagram so on the um on the y-axis we have lapse rate or the the rate of change of temperature in the atmosphere on the top we have instances where we have an inversion or where temperatures aloft are warmer than at the surface and then as we go down we are having the atmosphere increasingly cooling or instead another way we have an increasing lapse rate on the x-axis we have greater wind speed ratios so what you can think of that as is um better mixing in the boundary layer and what you can see here with these diagrams is that with an increasing lapse rate we are seeing that mixing is getting better or more efficient in the boundary layer and we can see that here for both of these these different heights that we chose and now here's just the same thing for the other buoy and we have the same relationship here where we are seeing that mixing within the marine boundary layer is uh becoming more efficient as our lapse rate is increasing so with that we wanted to see if there was any other variables that would otherwise interrupt this relationship we looked at a couple different things here I have season time of day so separating all the morning and evening soundings and then we also wanted to repeat this lapse rate um calculation with the air surface temperature so before we were looking at the water sea surface temperature and this is the air temperature just above the water and what we found is that all of these um all these variables do not greatly affect the relationship that you saw before they were all statistically insignificant and also the air temperature does not change as radically as the sea surface temperature so these relationships were not as meaningful but one thing that we did find was important was wind direction so we divided a wind direction into two different bins uh representing northerly winds or cold air coming into the region and southerly winds or warm air coming into the region and what you can see here on these graphs is that when we have cold air coming to our region we have a greater spread of possible wind speed ratios or what we can think of this is that when we have cold air we have an increasing chance that mixing efficiency will be increased the way that this was incorporated into the database was that the maximum and minimum possible mixing efficiencies are going to be determined by wind direction rather than just having one set for the entire database so now that we have all of this information I want to take a minute to explain just how this tool is going to work and what this looks like so the tool ingests four variables we have sea surface temperature observations we have the temperature at the top of the boundary layer and we also have the wind speed and direction at the top of the boundary layer it's just a simple subtraction to get our elapse rate and then with that and the wind characteristics um the database will take this and look and figure out what wind speed ratios and what range goes with these variables and so actually you don't just get one you actually get five possible wind speed ratios two of which represent our lowest end possibilities which is quantified through our 10th and 25th percentile our most likely situation which is also our median and then our highest and reasonable uh situation which is quantified through our 75th and 90th percentiles it's just multiplication to get our wind speed we reincorporate our surface wind direction to get a final uh wind grid a wind forecast grid so with these grids the forecaster is able to pick and combine and blend these grids as they see fit and once you have that then you have your final wind forecast grid through a previously established tool called The Marine Wind Gusts tool you are also able to make a wind gust grid based on the Wind grid that you just made with my tool as well so we'll go through that one more time actually using a couple numbers here's an example from January 16th just a random day and then on here I have the characteristics that are being input into this tool so once again we have our sea surface temperature we have our temperature and wind at the top of the boundary layer we calculate the lapse rate and then with that and the wind we the tool goes into this database and picks a range of wind speed ratios that based on these observations should approximate the amount of mixing that is occurring that day so again we have our two low end scenarios our our most likely scenario and then our two high-end scenarios so you'll see between our 10th 50th and 90th percentiles we have about 10 to 15 knots of variability on either end and while that is a lot we wanted to do this on purpose because as you saw in those histograms it's a pretty decent amount of spread on both ends and so um instead of just giving you a most likely scenario we wanted to also give you a high end and low end scenario so that um based on the conditions that you're seeing you can pick and blend and choose and um based on the current scenario you can make a more educated guess on what you think uh the amount of mixing is going to occur as I said before the surface uh wind speed direction is uh just reincorporated and then you have your final grids so as I was saying before the forecaster can pick blend and combine these as they see fit um to get your final wind grid though in this day actually going with the most likely scenario would would have been a pretty good scenario as uh what was observed at diamond shoals that day was a wind speed of 21 knots and then using the Marine Wind Gusts tool you get your wind gust grid and uh so what that would generate is a 25 knot forecast and the actual observed value was 28 knots so what this looks like in the computers for the forecasters to use it is labeled the empirical Marine wind tool though this is subject to change and when you run this it produces the five wind grids that we have been talking about so um here's an example of a sea surface temperature grid that you can input on the left we have again a model and then on the right we have the satellite derived observations and as you can see the um the big change here is in the northern waters where we had some upwelling and you can see that the temperatures are about 10 degrees cooler than the model thinks is occurring um so what this looks like is you just hit populate and then and then it will uh produce these grids currently it just runs but um as we make changes we want to uh insert a uh a user interface where um one you can pick which model that you are importing as there is many different kinds you can select the height of your boundary layer as we study the two different options and then we want to also be able to include preset edit areas so for example in this choice in this day you might only want to select the northern waters or you might want to only select the sounds where the model is doing okay and then where the model needs a little bit of help foreign so as I said before these five grids will appear based on the tool that we have just discussed and just to compare here I have the 10th and 90th percentile values and um you can really see here where the tool is doing its magic you can see here in the northern Waters where we have slightly lower temperatures you can see here where the tool is adjusting and producing lower wind speeds that the model would not be able to do on its own so just to restate our results we found that there is a normal distribution in the ratio of the wind speeds at the surface and the top of the boundary layer and that there is a relationship with these ratios to the lapse rate in the boundary layer and thus what this means is that we can predict mixing efficiency based on sea surface temperature observations um all these relationships were stored in a database which was then inputted into a tool for forecasters to use in real time to create probabilistic Marine grids which they can use to forecast I would like to thank the entire Moorhead City staff for our journey and the office a particular thank you to my mentors Carl Barnes and Ryan Ellis I want to thank Donnie for helping to code up the tool and also David Glenn for the use of the Marine Wind Gusts tool I'd like to thank the Hollings undergraduate scholarship for providing the funding and resources for this um for This research and I would also like to thank Seagrant North Carolina for hosting This research conference I have all my references here and if we have time I'm available to take a couple questions [Applause] yes we do have time for questions about a little over three minutes great great job with your time great job with the energy um for coming from Nebraska to talk to us about the gulf um questions have you tried comparing the your results with this one the one that is used in you know in terms you know of wind direction um we have not compared these results to the swan model although this is something we can do once we start um once we start operationally use with this tool once we get a little more into the cold season and this tool it's a little more effective and forecasters can provide their input and see how the tool is doing I've been sitting all day so this is great walking around thanks great job I was wondering with uh some of the work um Inland water bodies with water temperature being a signature for groundwater but I was wondering if there's any influence you could see in the estuaries related to the amount of flow coming in like moderating temperatures and that sort of thing um so uh they um currently the operational use of this tool is just a little bit limited uh since we developed this over the summer but um once we have uh more use with this uh with this when this upcoming winter we'll be able to um we'll be able to see that and have a little more answers for you thank you uh I just wanted uh to ask the question Logan because we know the answer but how much Marine forecasting experience did you have when you came into this internship absolutely none so he did a lot of work over 10 weeks we're very proud of what he was able to do yeah absolutely a round of applause any more questions one okay what's next for you um so currently I am uh exploring a couple grad schools but I would eventually like to apply and join the Weather Service really good answer thank you thank you again [Applause] so we have another friend from the National Weather Service here to talk to us uh Matthew Scalora oh nice um and he's going to talk to us not about New Wave information but New Wave information um included in the National Weather Service Coastal Waters forecast thank you let's do this if we can't find it we could switch up the order of the speakers this is all I have access to is it possible in one of the jobs I remember seeing those three and mine was right next to it they updated there's a date I put mine in on Sunday for the first time and I think they had done theirs on Friday and Saturday I'll take responsibility for this find it I can't find it let's just switch if you're okay with switching speakers well let Chrissy go and then we'll find yours unless you have it on a drive no I don't have it um I could we'll find it I could send it to you from my account wow okay yep let's just let's just do a little switch you know um Chrissy it's here there's Chrissy um from the U.S Geological Service another Federal agency that we love um is Chrissy Hopkins uh her presentation is well you can see the presentation title right there thank you Chrissy for being adaptable yeah yeah sticking it out and um excited to talk to you a little bit about uh green storm water infrastructure and some monitoring that we've been doing especially in Maryland but I think it um is a good example of when we Implement these practices across a whole watershed um what are the impacts that we see so we don't have a lot of places where we have a really dense installation of storm water practices like rain gardens and infiltration trenches and these types of newer stormwater practices um so I think this has some useful information that people can use along uh in coastal areas as well uh so here's a webcam uh video here showing a stream in Charlotte uh combined with USGS stage data so we can see um you know Urban streams are really flashy they rise and fall very quickly um and when we combine things like webcam data with uh you know discrete measurements of something like stage we can more easily visualize kind of the impact in the scale at which um flooding happens in our urban areas uh so what my research has to focused on is looking at comparing a few large centralized stormwater practices so things like large Retention Ponds to what happens when you install a a bunch of smaller practices distributed further up in the Watershed to a range of different uh stream functions so I'm going to call these large centralized storm water practices versus many small distributed storm water management practices so we've been looking at changes over time as an area goes from agricultural land use to Suburban neighborhoods so here's just an example of the land cover changes that we've seen in the study areas that we're working in Maryland you can see the agricultural fields here and some of the forested areas that were converted into roadways and houses for people's neighborhoods and then a school in the northern part of the image here um so this is in Clarksburg Maryland which is a suburb of Washington DC we have been monitoring this area since 2004 so it's a pretty long term study that we've been doing over a decade and comparing control watersheds to um three watersheds that went under went suburban development so you can just see kind of where those watersheds are and show the watershed boundaries there and then the diamonds indicate where USGA has been monitoring uh stream flow and then the county has been monitoring a range of different functions for a water quality and the benthic community within these watersheds so I'm just going to zoom in a little bit so you can see what these watersheds look like these are the control sites we have a forested control site which is the county park these are small watersheds so you can drive from one end to the other in about 15 minutes um you can see like a forested site here that only has about two percent uh impervious cover so things like roads and uh parking lots very limited here because it's a park but it also was formerly agricultural land so it's kind of secondary growth for us it's hard to find anywhere in this part of the country that hasn't been disturbed in some way in the past so it's kind of like the least disturbed area that we can find within this small study area the second control is um an urban control so this has kind of the development style that was uh happening in the 1980s so large detention ponds are kind of the primary uh form of stormwater management within this watershed and then the newer development that's gone in has more of these green storm water infrastructure practices so there's a few big box stores that went in recently that have micro buyer retention so small rain gardens installed in the parking lots that's what all those little purple circles are there um and then you can see the stream and um where we monitor downstream is the diamond and then um the retention ponds are shown as little diamonds in that um far map there and then the county has monitored um stream cross-sections in these watersheds so that's what those little um crosses are and then the treatment watersheds these are the ones that went from agriculture to development you can see all the storm water practices there as the little circles so hopefully you can see that in the image there but they also designed the roadways to have swales rather than curving gutters so trying to further disconnect those impervious services from the the stream itself so you can see things like drywalls that were installed behind houses so that takes water from the rooftops and tries to infiltrate it into the ground right next to the houses not like directly next to the house but close by uh the houses and then um some other practices so these all of these storm water features are arranged in a treatment train so one uh water from one practice goes to another for sort of redundant uh storm water treatment now this is the second Watershed that um underwent development it has 44 impervious cover so it's like pretty uh built out within this watershed it's a mix of single-family homes and townhouses so a little bit denser development and then it has almost twice the number of storm water practices so this is even like smaller practices things like tree box filters which are just look like kind of a a tree but it has a um gravel area underneath the tree that provides storm water storage and some treatment along the roadways and then there's also things like infiltration trenches near the roadways and then the last site that we've been monitoring this is the one that's kind of currently finishing becoming suburban neighborhood so you can see some construction activities in this image here this one has 20 impervious cover and a fewer number of stormwater practices because they're still we're being built when uh when we finish this study um and then it has a larger sort of forested buffer area so the bottom part uh is still fairly forested so what we've been doing is looking at a suite of different string functions we're looking at hydrology water quality geomorphic changes and then changes in the benthic community itself within the stream so I'm just going to highlight some of the the main findings that we found for these um four general categories of sort of stressors and stream functions that we've looked at um so USGS has been monitoring streamflow in these watersheds since 2004 like I said before so we monitor flow every five minutes uh so we've taken that entire streamflow record and identified every single um storm event that happened and then matched that up with the corresponding rainfall event and this is showing you the peak stream flow during each of those different events matched with the precipitation amount during that event so you can see for small events these treatment watersheds function fairly similarly to the forested watersheds they have lower peaks than the urban control site which is that older type of development so for these small events we're talking about like 10 millimeters of rain the stormwater practices seem to be doing a pretty good job at mitigating that flow but then as you get to larger and larger events these practices are only designed to manage about an inch of rain so you see that benefit sort of filtering off for these larger and larger storm events but the um treatment one watershed still has significantly lower peak flows than the urban control site even for those um large events even though it's not quite as good as the forested site we still see some you know benefit of these practices for some of these larger rainfall events I also compared the peak flows since these watersheds are so close to each other you know we can assume that rainfall rainfall patterns are fairly uh similar across this area although in the summertime that gets a little messy but um we can compare an event that happened in in one of these watersheds to the other one so each of these dots here represents one rainfall event that happened and then the peak flows from the two different treatment watersheds during that event so anything any dots that are above the line here indicate higher peak flows in the treatment two watersheds and you can see that most of the peak flows were larger in that treatment two Watershed than the first one now 86 percent of the events were higher in the treatment two watershed and we can just think about the differences between these two watersheds the treatment to watershed has like 11 more impervious cover so there's a lot more water to manage in that watershed which might be why um those peak flows are higher even though it has a denser uh higher density of stormwater practices so it's not just looking at you know the amount of impervious cover or the number of um it's not just looking at the number of stormwater practices but like the um you know the relative contribution of that storm water to the number of practices that are implemented so because there's so much more impervious cover and these are smaller practices they're not necessarily able to handle that amount of water uh we have a little bit of nutrient data so we've been looking at um base this is base flow nitrate concentrations in these streams over time and um we can see the nitrate concentrations are elevated in all of these um watersheds like above one milligram per liter because of the agricultural legacy of this area so groundwater um get them up here uh groundwater uh concentrations of nitrate in this area are elevated due to you know the agricultural legacy so this map here is showing just the areas where you would expect groundwater nitrate to exceed three milligrams per liter so but we can see that you know after development happens in this treatment one watershed which is the blue um circles here nitrate concentrations decline by half for base low in these in the stream and in the groundwater which is the triangles there we don't have a ton of groundwater monitoring data but we have a little bit of groundwater data and we can see that those concentrations have declined over time after construction has kind of ended but they still remain fairly elevated and even though concentrations have gone down we've actually seen um base flow in these streams go up so the overall export of nitrate has remained above roughly the same in this watershed over time um and it still remains higher than at the forested um reference site and the urban control site we've also looked at um specific conductance so this is just showing uh changes in specific conductance in the stream so that's kind of like how salty the stream is and you can see that specific conductance has increased in um all of these watersheds uh the treatment watersheds during development but at different sort of rates because of the timing of when that development happened and the urban control site has the highest specific inductance concentrations now you've also been able to look at changes in the stream itself so the particles within the stream channel and you can see that during construction in both the treatment 1 and treatment two watersheds you see that sand and silt and clay increased uh during construction or after construction in these two watersheds this has which has important implications for the habitat that the bent the community relies on we've also looked at changes in stream cross sections so here's just a a gift hopefully that's working um showing changes in the stream channel itself so the county has gone out and monitored uh cross sections within these watersheds since about 2003 so we've been able to look at how much stream bank erosion has happened and then how much down cutting has occurred within these channels and we find that the channels were in size like prior to development and they're continuing to widen and deepen after development has ended which might be related to that increase in peak flows within the watershed it's great to have this like really detailed information about one site but we wanted to look more broadly across you know whole watersheds we have repeat Airborne lidar data that we can look at changes over time within the whole watershed so this is showing a hill shade from a digital elevation model within the watershed in 2013 um and then we can look at how much change has happened since 2002 before development happens you can see how much uh the landscape was changed to um you know grade the landscape to make room for roadways and um houses and we can actually look at uh you know differencing this to DEMs to see areas that were filled in which is the brown um areas on this map so you can see streams and springs that were buried uh due to development and then also areas that were excavated to flatten the hilltops to make a room for the houses and we can try to estimate the total amount of Earth that was moved across um the whole watershed using these types of data um and then lastly I wanted to kind of give you a sense of what's happening in the streams themselves so that's what the county was most interested in was like you know trying to preserve the biotic Integrity of the streams within this area because they're and were deemed to be pretty um high quality and sensitive um habitats before the development ended so they wanted to try to promote uh you know development that minimized the impacts on the actual um benthic community within the stream so they've been monitoring um the benthic macroinvertebrates within the stream over time so we can look at how something like the index of biotic Integrity have has changed over time so this is just going to show you a series of different plots here any any numbers that are higher indicate a healthier stream and lower numbers indicate poor scores so we can look at the forested side and the urban control site and the forested site has remained in good excellent condition over the whole monitoring period whereas the urban control site has remained in threat to poor condition and then we can look at changes over time as these watersheds underwent development so the box the gray box in these figures shows that timing of of development um so this is before kind of when the development's sort of starting you can see all three of the treatment watersheds are in good condition and then as development progresses you see a drop in the scores during construction within the treatment one watershed but then it sort of rebounds to good condition um at the end of sort of the phase of development and the treatment II watershed kind of starts to become developed and then starts to degrade during that construction period you can see kind of the land cover changes that are happening at that same time and has kind of remained in um poor to fair condition after development and then the treatment three watershed which is just starting to um you know undergo construction is is still kind of uh bouncing around but has uh fallen into the sort of fair condition in the more recent years um so just to summarize here some lessons learned from using this sort of distributed stormwater control that this type of stormwater management can attenuate peak flows and runoff volumes but you know the storage capacity within those practices really matters and we can see differences in the performance of those events based on you know how long ago the previous rainfall event had happened they can also improve water quality we've seen reductions in sediment export within these watersheds in one of the watersheds but we don't see those benefits for all constituents particularly salt in this these watersheds and that it can reduce the impacts on biota but the sensitive families haven't been able to recover in these streams even though you're seeing kind of a rebound n those benthic scores the sensitive families uh still are in pretty low abundance in these streams and that distributed storm water control isn't always better if you have more impervious covers you need to think about you know trade-offs between allowing more development within the watershed but also creating more storage to go along with that development and then lastly we found that the construction phase is pretty important in sort of triggering some of these impacts to the watershed this is the time period when you see substantial changes in the topography and kind of how the flow paths within the watershed get messed up and moved around within the watershed um and we also see an increase in in fine sediments within the stream channel um during the construction phase so even though there's sediment erosion control practices uh you know going along with construction we're still seeing the impacts of that construction in the stream channels downstream of that disturbance so that's all I have um if you want to know more chat with me I'm excited to talk to you more about this and thanks so much and happy to take questions [Applause] questions uh we do have we do have some temperature, we haven't actually analyzed it yet, so uh we're actually trying to incorporate that into a model to kind of look at how these different stressors combined impact the benthic community. any more questions okay okay have you looked into the characteristics every Watershed in terms of a slope average slope or Channel slope because that can be you know a very good Factor actually yeah so we actually see some impacts with um differences in slope especially with the processing data you can you know make one or two profiles communication models both very uh in the channel and also like it increases due to the grading and just kind of a rebounding of like where how all of these different somewhere else the differences that we're seeing between higher slopes in this area here especially when there's roadways division was based a lot of decreases um so we're seeing like longer falling lands within the hydrocraft and those kind of remaining elevated afterwards so that might be one one reason uh some other similar practices there all of them are supposed to be kind of infiltration tokens but they all have like under drains so a lot of that water is like that is yeah so um it's um you know we've mostly looked at this one Watershed which we have fun for the longest uh Stream flow records we're curious to see like each other for watersheds over time I'm gonna see that so you can see that you'll choose the record you know it's a time period and it's not working no no it's not as far as I can so uh foreign well thank you again Christy I look forward to being able to see your presentation online um later because I do believe I didn't get a chance to see it and now we have Matthew we found his presentation New Wave information but I guess there's more pressure on me now that I have to go last but I'll try yeah um so yeah my name is Matt Scalora I'm a lead meteorologist at the National Weather Service forecast office in Wilmington North Carolina I'd also like to recognize Mark Willis the meteorologist in charge of my office as well as Darren Wright from Weather Service headquarters in Silver Spring Maryland for definitely being greatly uh involved with this project as well and you can see there it's going to cover note the new wave information included in the NWS Coastal Waters forecast which I'll commonly refer to as the CWF in the talk and in terms of all the folks that went into this work and this project itself there's definitely a lot of contributors over a couple dozen people on the National Weather Service National wave team with representation from all around the country every coast and even a couple folks from the Hawaii office as well so before we get into how we're inserting the wave forecasts into our products themselves let's take a basic overview of waves waves have three very fundamental variables that being hype period and Direction height is the distance from the crest to the trough of the wave measured in feet and direction is where the waves are coming from sort of like what we do with winds not where it's going to and period uh definitely one of the more important variables that we measure in seconds is the time it takes for two successive crests or troughs to pass a fixed point in the ocean and that could be for instance a buoy or some other Standalone object and it's very common to have coexisting waves at any point in the ocean and that can be near shore it can be way out in the ocean and all these locations are going to have high unique height period and Direction and each of them will likely be interested particular Marine groups in the next couple slides I'll give you two examples of that and for the time being in our Coastal Waters forecast we're definitely oversimplifying the forecast by just providing a significant wave height significant wave height being the highest one-third of all waves so we definitely like to add that wave detail value because we think that providing it is definitely vital for Mariners to decide if it's safe to venture out on any given day so two examples to start off here currently we're just providing a forecast saying C's eight to ten feet uh even though the picture looks a little taller than that you get the general ideas these are big big waves so a big boat might think that's all right they can handle it however if they saw a period forecast of 22 seconds they might realize that a Perry that long is going to generate s
2022-12-19 20:12