 Alright, thank you for that introduction and thank you so much for the invitation to to come here and talk today about some of our work on fire. I'd like to start off by just acknowledging the help of a number of others over the years on these projects, including France, this is Nina Oakley, Jason keen, Dennis daily, we tang and you Berg. So, as some overall motivation for this I think that there's a particularly urgent need, not only to understand run off an erosion in recently burned environments but also some of the most extreme gene morphic responses we can see in those areas. And in particular that's debris flows you just got an excellent introduction to the force and power and size of these debris flows with the last talk so I don't think I really need to motivate that very much. But in in recently burned environments debris flows tend to initiate once a some critical rainfall intensity duration threshold has been exceeded fire and in particular moderate and high severity fire can reduce hydraulic roughness by by incinerating vegetation and litter on that that normally covers the forest floor, it reduces canopy interception. It can sometimes render the soil hydrophobic so that the soil is unable to absorb water, as it normally would in an unburned environment, and all of this leads to increased runoff and erosion. Currently, we're seeing an increase in area burned in many regions and in particular area that's being burned at moderate to high severity so we're seeing these more extreme fire impacts and a greater number of areas. And at the same time we're also tending to see in many of these same areas increases in the likelihood of extreme precipitation events. So all these things combined mean that we're going to be crossing thresholds for debris flow initiation, more often in the future with with potentially devastating downstream effects to people infrastructure and water resources. This leads to a couple of very basic questions. When and where are we most likely to see debris flows, how big are they going to be, how far are they going to go and what are they going to impact. We're going to draw the distinction again between a debris flow which has an extremely high sediment concentration, probably more than 40% concentration by volume and floods which behave differently they interact with the topography differently and they present different hazards. So we often really want to identify areas where we think debris flows are going to be a problem as opposed to floods. These are some examples here of current hazard assessments this is after the 2017 Thomas fire in Southern California. You can see on the far left here that watersheds that are more likely to experience debris flows are highlighted and warmer red colors. In the center here, warmer colors indicate watersheds that are more likely to produce large debris flows as opposed to smaller debris flows. And I'm going to focus really more on how we can use process based modeling as opposed to empirical models which which are shown here to do some of these same things and answer these same questions. I'm not going to tackle the how far our debris flows going to go question that's very interesting but Alex Gore is giving a talk tomorrow on that so you have to have to tune in there. So what I'd really like to do today is present a brief conceptual model for how debris flows initiate and grow in burned areas, then describe a process based numerical model that can help us understand the transition from run off to debris flows, and demonstrate how the model can be applied to inform our understanding of debris flow processes and hazards. So the direct debris flows that initiate in unburned areas, mainly due to to being mobilized from shallow landslides debris flows and burned areas very frequently initiate when run off concentrates and steep gullies and channels and rapidly and trains a sufficient amount of sediment. Here's an example here from the 2019 Woodbury fire in southern Arizona, we're high up on the hill slopes, during a single rainfall event, this gully that's over a meter deep got carved. You can see that there's some burn remnants of shrubs here so there's no canopy to intercept rainfall, the roughness on the surface here is far less than what it would be in an unburned setting here. And this fire did burn parts of these slopes at moderate severity, and it left the soil fairly water repellent so that it couldn't absorb water as it normally would. Before I go into describing the process based model for for for a debris flow initiation in these areas. I want to first take a step back and ask, why do we need such a model. We've had a decent amount of success using empirical models to answer some of these these first order questions that I that I motivated my talk with. So one of those questions might be how much rain does it take to produce a debris flow and a given watershed. And one common tool to do that is a rainfall intensity duration threshold which is shown here on the left, where we take historic observations of storms that produce debris flows shown with the blue dots and observations of storms that produce, sorry, that produce floods are shown in the blue dots and storms that produce debris flows are shown in the yellow dots. And we can usually draw a curve like this red curve shown here that that fairly well delineate storms that produce floods and debris flows. And in particular, looking at the peak 15 minute average rainfall intensity during a storm often provides a very good indicator of whether or not a storm is going to produce a debris flow. Another example is that these types of methods require a decent amount of data, especially to be accurate and reliable, and fire is starting to expand into a larger number of plant communities and geographic regions where it hasn't been before. So we simply don't have the data in those areas to take that approach. Another another example is that we often have an abundance of data or at least more data in the first year following a fire of storms that produce debris flows versus floods. We tend to lack that data in subsequent years so that we don't really know how this red line is going to shift as the landscape recovers presumably it's going to shift up and it's going to take more intense rainfall as the landscape recovers to produce debris flows but we don't know by how much. So those are a couple examples of why I think process based models can be can be helpful in answering these questions. So the model that I'm going to describe accounts for infiltration and rainfall infiltration and runoff processes. It routes water over the landscape and accounts for entrainment and deposition of sediment and then subsequent changes in the topography. The model inputs include a rainfall time series like you can see here on the left information about vegetation, which may be related to burn severity. With areas burn at moderate and high severity tending to lack a lot of vegetation in areas that are unburned or burn at low severity still still having some information about soil properties is also essential especially soil properties that affect infiltration and the output that this model produces is spatial and temporal variations in flow depth flow velocity and sediment concentration within a watershed and you see an example of that here where we're seeing peak flow depth in this small watershed that you can also see a picture of here on the far right. So this is really designed to simulate the hydrologic and geomorphic response of an individual watershed at the event scale, so a single rainstorm. I'm not going to go into very much detail about the model equations but essentially it's based on a set of PDEs that describe conservation of mass and momentum. It's a shallow water type model that's coupled to K different advection equations that track sediment and where it moves and how it's transported. There's K different advection equations because we're looking at sediment and different particle size classes and and that gets entrained and deposited differently so we want to want to keep track of them individually. So we want to count for rainfall infiltration topography and flow resistance, which varies as a function of sediment concentration to briefly highlight a couple of important things about the sediment transport piece of the model. It's based on the hair sign rose soil erosion equations which account separately for entrainment in areas of concentrated flow like rails and gullies and entrainment in in inter real areas. Entrainment rates are very sensitive to vegetation and ground cover, because sediments mostly being transported through raindrop aided processes, and then in areas of concentrated flow it's more sensitive to stream power, which depends on on slope flow velocity and flow depth so you could see how that would be affected strongly by a hydrophobic soil that would produce more runoff, whereas the raindrop aided sediment transport is going to be highly affected by by burn severity and vegetation cover. So, to begin to kind of apply this model I'm going to focus on this small watershed that you see here in the right, which is a less than point one square kilometers in in area. It's located in Southern California in the San Gabriel Mountains, and we chose this for a site to apply the model to try to answer some of these questions about debris flow processes, because we have a lot of measurements and observations that help us constrain various model components. So the first question that I want to look at is what rainfall is intensity, what rainfall intensity is required to produce a debris flow in this watershed, and how big will it be. And in order to force the model we're going to use rainfall that was actually observed at a rain gauge very close to the site and we've chosen to use rainfall from a particular type that was generated from a particular type of system that's common in this area. They're called narrow narrow cold frontal rain bands. And you can see you can see that here in this radar image, they're characterized by short duration extremely high intensity bursts of rainfall and they've been linked with triggering postfire debris flows so that's why we we chose this particular type of high edigraph. And I'm showing here not only in black the one minute average of the rainfall intensity, but also the 15 minute moving average because I mentioned that averaging rainfall over that 15 minute duration has proven to be a good predictor of whether or not a debris flow over. So I'm going to reference that that value. This is a very recent number of times moving forward. In this case you could characterize this, this rainstorm, we're going to characterize it with a single metric. You could say that the peak 15 minute average rainfall intensity of the storm is roughly 50 millimeters an hour. All right, so what does the model produce if we look at what's going on throughout this rainstorm. Early on 20 minutes into the rainstorm. There's more of all of the water essentially that's fallen on this watershed and so you don't see any runoff. If we progress a little bit farther into the storm 40 minutes in. We're seeing higher intensity rainfall, and that's sufficient to to to generate some runoff and you see it concentrate in rills and gullies and into this main channel. It's a small watershed so these float ups are fairly minimal but on the order of 25 centimeters. So if I progress a little bit farther into the storm you see that that the runoff response kind of die down again is the rainfall intensity decreases. So we can also instead of just looking at these maps we can look at a summary of flow depth at the output, or at the outlet. So not surprisingly you see what you would expect. There's a lot of rain bursts and rainfall intensity you see increases and float up to the outlet, but what we're really interested in is is not when when we get flow but what, what type of flow is it are we getting flows that have high sediment concentrations that are typical of debris flows, or are we mostly seeing flood flow. So we can also look at the sediment concentration at the outlet, and we see that indeed we're getting a decent number of flows with sediment concentrations that are quite high. And during this time between 40 and 50 minutes when we saw the peak flow, and then a couple of other burst through later in the storm where we're also seeing spikes in flow depth and sediment concentration. So these are small debris flows because it's a small watershed but this storm with that with that I 15 of about 50 millimeters an hour would be sufficient to produce debris flows. So, in addition to just, are we going to get debris flows during a particular storm we wanted to look at this more systematically of when exactly do we start to cross that threshold from flood to debris flow. So what we did was we essentially just scale this high edigraph up and down by, you know, 50 to reduce it by 50% reduce rainfall rate by 25% systematically to look at different different values of peak I 15 and see when we started to to initiate debris flows. And so what you're seeing here is a plot of the total debris flow volume that was produced during one of these storms. If we scale the I 15 such that the peak I 15 is 10 millimeters an hour. We do see some runoff and we see a little bit of sediment moving but we don't see anything that has sediment concentrations associated with the debris flow. So there's no no debris flow volume comes out of this outlet. So a knife 15 of 15 to 20 millimeters an hour to generate a debris flow. And coincidentally, the, there is an empirically derived regional threshold for the first year after a fire in this area, it's about 18 and a half millimeters an hour. And so we were pleased that our model was roughly reproducing that threshold. And then you can see that as you continue to increase rainfall intensity you get, as you may expect an increase in debris flow volume. This is great for the first year after a fire, but what happens as the landscape starts to recover and as we start to lack those those observations to derive empirical thresholds. You can see here on the left side this ridge of the of the ridge was burned at moderate severity. On the right side of this ridge, you see, it was unburned the tree the forest is shaded a little bit read there from a fire retardant but after three years you see that there's a big difference and we start to see a vegetation grow back. So we we made measurements throughout this area as a function of time following fire to quantify canopy cover and we see through throughout the first three years this is just a very brief summary of some of the results that we get an increase in canopy cover from essentially 0% to 80% over 33 months of recovery. I'm showing here some some measurements that we made in the field that are related to infiltration capacity. In particular saturated hydraulic conductivity in this top plot and wedding front potential in the bottom plot, as we increase both of these variables were essentially increasing the infiltration capacity of the soil. So you see that we are seeing a recovery of soil infiltration capacity over time, probably associated with reductions in post fire water repellency. So what are the implications of this for debris flow initiation and debris flow volume. Well we can run this through the model just like we did before. And what we see is that in post fire year two, it takes a rainstorm of the peak 15 minute rainfall intensity of about 35 millimeters an hour before we start seeing any debris flows being produced. And when we do they're fairly small, and they stay fairly small even at rainfall intensities of 60 millimeters an hour which is roughly a 10 year recurrence interval storm in this landscape. By year three, we wouldn't expect that there would be any debris flows with even a 10 year recurrence interval storms there's substantial recovery and we're actually able to put some numbers on it that are that seem reasonable to you based on on our observations. So you might ask how much of these changes are due to recovery of the soil versus recovery of the vegetation. Another benefit of these types of models is that we can kind of play that game through, and we can, we can explore a hypothetical scenario where we have an extreme case of no vegetation recovery over the first three years, but we allow the soil to recover as we actually observed in the field. And what you see here as a result of that is this dashed line where in year two if we don't allow any vegetation recovery. So maybe there's an extreme drought for example. Then you would still start seeing debris flows at fairly low recurrence interval rainstorms of 10 to 15 millimeters an hour, sorry 15 to 20 millimeters an hour, and in year three you would still produce debris flows. So these landscapes depending on what recovery trajectory they take may remain more susceptible to debris flows. Through two or three years but it's just in the recovery trajectory that we observed, we did see that susceptibility decrease fairly rapidly. So hopefully I've, I've, I've kind of helped you realize how these types of coupled models for runoff infiltration and sediment transport may be useful for understanding debris flow hazards process dominance in in recently burned landscapes. And I think these these three questions here that I have listed are really fundamental not only from a hazards perspective but also for a landscape evolution perspective. One of other takeaways is that we found that the initiation and size of postfire debris flows can change substantially during recovery. We observed or the model would suggest a 3x increase in rainfall threshold over three years that are Southern California site, and temporal changes and rainfall thresholds were driven more by vegetation recovery than soil hydraulic properties or changes in soil hydraulic properties over time, which I think is a really important finding because so much emphasis is often placed on water energy and recently burned areas and how that affects soil hydraulic properties, but vegetation recovery in this case turned out to be equally if not more important. So, thank you all for your attention. I was wondering, does your model take into account both the effect of the sediment on the momentum equation, and also does the sediment in the, in the debris flow help you wrote the soil. Yeah, so there, there is a feedback between sediment concentration and flow resistance so as a sediment concentration starts to increase. There's a essentially a when when when when the sediment concentration reaches a level of 40%. There's a cool on friction term that kicks in, and we kind of scale that up from a sediment concentration of 20%. It starts to be active and then 40% it's fully active. And so it's a bit, it's a bit arbitrary but but there is kind of that that transition in there and flow resistance as you go from completely water dominated to more completely in the debris flow regime. And once, once a flow becomes a fully developed debris flow, we actually, we don't allow it to entrain any more sediment so once, once it reaches a concentration over 50% kind of shut off the traditional kind of grain by grain bulking sediment transport processes in the model. We allow some other types of mass failure processes to still add additional sediment to the flow if they occur. But yeah, that's the basic idea.