 second speaker is Danika Roth who will be speaking about wildfire and surface roughness. Thanks. Can you guys all see my screen? Should we share a title slide? Yes. Great. Okay. Cool. So let's talk about wildfire and surface roughness. So in the broadest sense my group's work has been centering on the question of how do landscapes respond to wildfire? That's a really big question though. So for today we're going to narrow it down to looking at how steep landscapes respond to moderate to severe wildfire. So kind of more the noticeable and members of both landscape and post-fire responses. All right. So if we're going for a really basic description here we could say that in an unburned landscape roots support soil stability while above-ground vegetation both stores sediment up slow and provides some kind of protective cover for the bare ground. During a fire or right afterward that vegetation gets incinerated releasing sediment to travel from the hillsides to the channels as dry rubble and exposing the surface to wind erosion. And then once rain starts falling on the surface the bare ground gets reworked by rain splash to create micro topography and overland flow starts to develop real and gully networks. Then after this comes a prolonged period of elevated erosion and hazard risk until eventually the hillslopes start to recover as vegetation regrows and starts to store sediment again and the channels incise through all the settlement that the hillslopes dumped into them. And finally many years later we'll return to our initial unburned state. So this describes kind of a basic cycle of wildfire impact and recovery. The details and timing of some of these processes are still kind of unclear but they're starting to matter more and more as wildfire frequency and severity increase with climate change. So the good thing is that many of these processes also are somehow related to surface roughness which we can measure through high resolution modern topographic data. So for the rest of this talk I'm going to take you on kind of a whirlwind tour through this fire cycle by looking at three ways that my group and I have been exploring what we can learn about postfire erosional processes from topographic roughness. We'll start with this initial postfire landscape, go through rill and gully formation then talk about longer term hill slope recovery. So in the earliest postfire stages wildfire simply removes surface roughness. So these are images from a study that Tyler Doan, Josh Boring, David Furbish and Aaron Zettlerman and I have recently submitted to PNAS on some rock job experiments in the Oregon coast range where we compared the roughness at a bunch of burned and vegetated slopes and found that the median micro roughness is up to an order of magnitude lower on the burned slopes which you can see in these cumulative distributions of micro roughness. Just to clarify since there are a lot of ways to measure roughness, roughness height here is a measure of deviation around a standard or an average ground surface and micro roughness means we're using a high-pass Gaussian filter to examine roughness over length scales or spatial wavelengths up to about a meter. So we'll notice we're also showing surface gradient S here. So we'll look more closely just at these two sites which have the same surface gradients and only a one and a half centimeter difference in median roughness. I mentioned this was a rock drop study so we were throwing rocks down hills and these are the histograms of particle travel distances along with representative hill slope profiles of the green lines for each of these sites. I just want to point out that these are on the same horizontal axis is up to four meters same same distance there. So what we're seeing here is that while around 20% of particles traveled over four meters on the burned slopes, some of them quite a bit more, not a single one made it past four meters on the rough vegetated slope. And again these sites have the same surface slope and only one and a half centimeters of difference in their surface roughness. So not only does fire remove roughness but that removal has direct implications for the motion of sediment after fire for example like dry raffle because roughness steals momentum. So particles travel much farther on smoother slopes. One last note about this study you might have noticed that there's a bit of a slope dependence to those roughness distributions. If we look more closely we find that short wavelength micro roughness is only slope dependent at vegetated sites. But the longer wavelength macro roughness so through a low pass filter looks like it scales with slope at all sites. So that implies that there are some specific length scales of roughness that either burn away or erode really quickly in a wildfire. And that's something that we're currently investigating using spectral analysis of burned and vegetated topography. All right so moving on after a wildfire roughness eventually starts to recover. One of the first ways that roughness returns to a post fire landscape is through the development of real and gully networks. So this is some work that my master student Claire Bavres is doing to examine how roads impact the development of those networks. Specifically she's looking at this following Thomas Fire in Southern California. She's still in the preliminary stages here so we don't have real results to show yet. But one of the things she's having to do is to figure out an automated way to detect and measure real network density which she's looking at in a few different ways including flow routing, light-r, return intensity, maybe some spectral analysis and by looking at surface roughness in cloud compare. I just want to point out that here you can see the real showing up pretty nicely. She's also looking at how overland flow can be rerouted by roads which going back to our last study we can maybe think of as kind of spatially explicit macro topography or macro roughness. So if you'd like to find out where her work goes you can look for her results at AGU this year. Okay and then the last project that I wanted to touch on looks at roughness recovery over longer time scales. Excuse me. So my colleague Gabe Walton and I taught a class on point cloud analysis last semester and our students Andrew Graber, Claire Bavres, Kale Kelgan and Ryan Koh are actually putting together a paper now with some pretty neat class results from analyzing Steve DeLong's light-r scans following the horseshoe to fire in Arizona from 2011. So Steve used change detection to get at some very nice high resolution erosion and deposition patterns but our students were looking at roughness. So this is the first of Steve's scans immediately after the fire and a histogram of roughness heights over here from cloud compare again. Compared to the same region nine months later where we can see that surface roughness has increased. So it shifted from smaller roughness values or roughness heights here to larger values over in the tail. But what is really exciting here is how well spectral filtering pulls out these changes so we can clearly see the roughness reforming on these hill slopes. And again going back to our first study our students are also using spectral analysis to investigate whether we can pick out the characteristic frequencies of roughness that are evolving here and identify specific features or processes that they represent. So for example we think the roughness here is probably a combination of vegetation regrowth and erosion excavating around immobile roughness like boulders. So you can see the rest of their results at AGU too and hopefully in a class publication sometime soon. And that's all I've got for you today. Thanks. Okay. And so please put your raise your hand or write a question in the chat. And while you're while we're waiting I had a question about is there any ability to pick out vegetation versus roughness from the signal? And you mentioned that a little bit at the end. Yeah. So our students are they're taking a couple approaches to this they're looking at first trying to remove the vegetation and look at the how the bare ground is evolving. The vegetation is itself a form of roughness. So for example in that first study the vegetation was the roughness there that burns away. And it's it's my guess is for the the short wavelength roughness that's burning off this probably vegetation. And some of this is probably regrowth. But like I said vegetation also can stockpile sediment behind it. So it could be a combination. But yeah they're they're looking into into identifying which parts are are actual vegetation which parts aren't. Interesting. So let's take one more question before moving on to the next speaker. And if you don't need your slides Danica go ahead and take your screen off. The question is from Kristen Sweeney. And it's can you expand on how you chose the window size for measuring surface roughness in cloud compare. Is it all a single neighborhood radius or are you looking across neighborhood sizes. And Arvind could you get your screen up while we wait. Yeah in the in the most recent one that I'm showing I believe that was a one meter window size but they are exploring different window sizes to to see what's kind of capturing different types of signals. I think really kind of looking at a spectrum is is a more universal effect. Then then using cloud comparisons. Cloud compares kind of looking at within one window size. So I think the goal is to kind of examine things overall scales.