Transcript for How forest disturbance affects streamflow: The Kuczera Curve revisited

Rachel Brown (DEECA):

I’d like to acknowledge the traditional owners of the lands on which we're meeting. In my case, the land of the Wurundjeri people of the Kulin nation, and I’d like to pay my respects to their elders, past and present, and also extend that acknowledgement to those across other parts of Victoria and Australia where you might be joining us from today.

You might be aware of our research program, the Victorian Water and Climate Initiative, or VicWaCI for short. However, our team are also connected to some other research and hydrology activities that formally sit outside of VicWaCI and one of these examples is through the Bushfire and Forest Services group within DEECA, and they invest in the Integrated Forest Ecosystem Research program, or IFER for short.

IFER involves researchers from Melbourne University School of Agriculture, Food and Ecosystem Services, and it aims to enhance the evidence base of forest values to support decision making under uncertain conditions.

Given the importance of streamflow generated from Victoria's forested catchments, water is one of the forest values considered in the IFER research and so our hydrology and climate science team is involved through that program.

Today's webinar will present on work that was partly funded through this IFER program and we've got Professor Pat Lane from Melbourne University here to share findings of his research into Forest Dynamics and the implications on how hydrologists think about water yield following forest disturbance.

Many of you on the call today will have heard about the Kuczera Curve. For those who need some background, the Kuczera Curve has provided a simple and useable model for predicting the long term impact of fire in ash forests based on the age of the forest.

It was established using observational data following the 1939 fires in Victoria's Mountain Ash forests, but since been applied rather indiscriminately to estimate the hydrologic impact of forest disturbance. Despite the widespread application of the Kuczera Curve in hydrological studies, stem replacement experiments in Mountain Ash catchments have not been able to faithfully replicate the Kuczera response, and Pat and his team have been investigating this inconsistency.

So today, Pat will describe this research into forest dynamics and catchment yield. These recent findings challenge the long-held assumption that the Kuczera Curve is universally applicable and I expect there will be some burning questions from the audience. Pat will be able to answer these questions at the end of the presentation, so please submit these via the teams Q&A functionality that you'll find available.

I'd like to take this opportunity to thank Pat for making the time to talk to us today and also for all the researchers that are working tirelessly behind the scenes and have contributed to this work.

In terms of some of  housekeeping information, we'll keep today's webinar to about 3/4 of an hour. Pat will give us his presentation. and there'll be time for questions after that. You should see a Q&A button enabled within your teams application and you can use this to post your questions throughout the session at any time, and then we'll put them to Pat at the end of the his presentation.

If for some reason you're unable to post using that Q&A functionality, you can feel free to send your questions to our team email, which I've got on the slide behind me - that's HCS.team@dewlp.vic.gov.au. We’ll aim to get through any questions today, but if there's an abundance or we need to follow up, we can do that afterwards via email.

Today we're using Microsoft Teams. Hopefully most people are comfortable with this sort of technology now, but if there's any challenges, please let us know. You can use that email address to get in touch with us. By default, we've switched everyone's microphones and cameras off. And finally, I'd like to thank everyone for joining us today and hope you enjoy this event. I'll just introduce Pat and we'll get started.

Pat Lane is a professor of Forest Hydrology at the School of Agriculture, Food and Ecosystem Science at the University of Melbourne. His research interests include the ecohydrology of natural and disturbed forest, streamflow dynamics and erosion processes. He has a particular interest in the effect of fire and climate variability on forest functioning and hydrology.

So welcome Pat and over to you. Thank you.

Patrick Lane (University of Melbourne):

Thanks, Rachel, and thanks everybody who has come along today. It's always great to have people turning up when you've got something to say.

I'm mainly gonna talk about revisiting the Kuczera Curve, but I'll also talk a bit a little bit more at the end about some more general forest hydrology outcomes from some of our research over the last few years. We're particularly focused on fire when we're talking about this.

And just a little bit of background for anyone who doesn't know this, that in 1939 we had a massive fire. I can't quite remember the area now, but it burned a lot of our upland forests from the Otways all the way up into NSW. And where that fire intersected with Mountain Ash or Alpine Ash, it killed them in the main and then they regenerated as a single age, which is what Ash species do. And that means that even now, all those decades later, still the majority of our Ash forests have a 1939 origin.

In the mid-70s, early to mid-70s, there was a flurry of hydrologic research that came out of the Melbourne Metropolitan Board of Works, the MBW, which eventually turned into Melbourne Water. And there was a lot of, a whole stack of, about a dozen or so small paired catchment experiments put in in the North Maroondah area to look at various aspects of what happens to streamflow if you modify the forest, but around about that same time, people also started noticing this really strong pattern of that there was way less streamflow as a function of the rainfall than they would have expected.

And so that led to, first of all, John Langford published a paper in 1976, I think it was, in exploring this and then there was a whole bunch of work done over the ensuing decade to try, and I suppose, characterize this and think about why it's happening. And this culminated in George Kuczera coming out with his famous curve.

There's George's curve, and this curve represents, so George basically fitted this model to observed data that came out of catchments, the Grace Burn, the Watts River, Donnelly's Creek and some others. And this effectively says that relative to an old growth or a pre-fire condition, which George defined as old growth, that as the as the stand recovers, there is a very distinct and quite dramatic change in water yield. And if you flip this curve or put it in front of a mirror, the mirror image of that would be we assume to be the evapotranspiration.

So basically, you have an older forest as we see on the left-hand side on this on this slide where it's very sparse large trees, but only a few of them, roughly say 20 trees per hectare. And then on the other side, we have this really dense regrowth greater than 10,000 stems per hectare. And effectively, that's what we've got here is a vigorously regrowing forest using much more water in its younger age.

And so this, I might add here that I really want to emphasize this is not about bashing George Kuczera's work. George himself cannot believe that this is still being used, but the reason why it is, is that it's a really beautiful, quite simple conceptual model that can actually and has actually been used in water resource, you know, allocate or not allocation, sorry, in modelling water resource after a disturbance.

Now it's also however, been misused rather a lot over the years, because it really only pertains or definitely only pertains to Ash catchments, not the vast majority of our Eucalypt species. And it is of course, this assumes the entire catchment of interest has all, is all the regrowth forests. So, it's all been killed in this case by wildfire, but I suppose it has been extended to harvesting impacts. So, it's for an entire catchment, but of course we know particularly around harvesting that when that was happening, that there were much smaller areas of the landscape were impacted. So, it meant that the implications of that land use were quite often overplayed. However, certainly you know it is a great concern to citizens of Melbourne if, after a big fire, they could only have half the water that they had before, you know, in a decade or so, in a couple of decades, going forward, so that's a bit problematic.

And the reason for this is really that, on the right-hand side of this schematic, is you have this kind of schematic of this dense regrowth, that might be, say, 15, 10-15 years old. And then on the left-hand side, you have an old growth which has perhaps you know in the order of 150 plus years old and you can see that these are very different looking stands. And it's really about this stand density and the dynamics of the overstory, which in this case is the Eucalyptus regnans or the Mountain Ash that we think has been driving this, the response that George Kuczera found.

Now there's been a lot of studies since George came up with that in the mid 1980s. There's been a lot of research kind of trying to get a better handle on this, and in particular the work that Rob Vertessy and others did in the 19, mostly in the 1990s out of the CRC for catchment hydrology, looked to come up with a process understanding of this. And on the left hand side there this is work that was published and really this is where uh, people went out in the bush and put in all these sort of small scale water balance plots, measured all of the components of the water balance, transpiration, interception, soil evaporation, etcetera, etcetera and came up with basically a finding that supported the Kuczera Curve where that as the forest ages, the black shading here is probably the big deal here is where the transportation of the overstory declines over time. It peaks in these early years, declines over time as the forest thins out, the understory becomes more important in its water balance, but not to quite the same extent. And this leads to a lessening of evapotranspiration as a forest ages.

On the other side here, here's a plot of some work I was involved in some years ago where we revisited a catchment experiment in the Coranderrk catchments out near Badger Weir. This is a catchment that was clear-felled, it was actually not, pure Mountain Ash stand as it's a kind of on a marginal rainfall for Mountain Ash, but Mountain Ash was then resewn onto that site. So we turned the kind of marginal Ash site into an Ash site. And here is the result looking at this quite a few years later and you can see there is what we would call a Kuczera effect. There's the zero line and there is a dip in streamflow as we go forward.

However, an interesting, two interesting things about this one is that it's a pretty shallow dip, nowhere near the minima of the curve that George produced. And also, you will notice up the end here this very wavy line. This is actually the stream drying up in the Millennium drought, and this is notable because what this actually represents, in my mind at least, is a kind of climate change impact where we put a high water using forest onto a site that wasn't quite suited to it. And basically that regrowing stand used all of the water and that is something that we may well see coming to us with the climate, changing climate coming.

Anyway, so, despite the fact that people come up with some process explanation, there's a lot of data and there's been a lot of measurement out there in the forests around this that didn't quite add up. And some of this was catchment scale, some of it was plot scale and some of it was people going out and measuring stand attributes, which included you know kind of standard forest inventory that Melbourne Water did a lot of in the early days in particular. And then people like us going out sort of doing stand attribute surveys. And we found that there's a really big spread of particularly those stand attributes and this really made us feel that using age as the independent variable in trying to predict what will happen over a lengthy time period probably has some significant limitations and so we felt that we needed to try and get forest structural attributes into the model.

And so, this is our problem statement, that we felt that it was worth revisiting this whole question of post disturbance stream flow in the Ash by representing forest dynamics. But this is a pretty tricky thing to do because there are hundreds and thousands of hectares that we're dealing with. You can't go out there and count every tree. How can you do this in a reasonable way?

And the reason why the stand attributes are so important and so useful is that we, a lot of research over the years, which included the work of Rob Vertessy and and other people around that time, and it's been more work that we've done in our group is that, the sapwood area, which is represented here in this diagram on the top left-hand, which is basically the xylem vessels of a tree are, the conducting tissue, that's where all the water moves, and that happens to scale really, really well with evapotranspiration and therefore streamflow. And it does this really well in the Mountain Ash forests, it works pretty well in the damp mixed species Eucs, it starts to fall over a bit when you start to get into dry forest, but in this case it works really, really well. And the other great thing is that the sapwood area scale is really well with the basal area of any given tree. So, if you can measure the basal area, you know the sapwood area pretty well, and from that you can predict the hydrology pretty well.

But how to do this? So, we have done a bit of work in this group and others, but probably mostly in this group actually, of using LIDAR to characterize catchment vegetation and using those relationships I’ve just talked about between basal area and sapwood area to be able to scale up. And there's a couple of pieces of work we've published over the years. However, it's a bit tricky because we don't have... if you're trying to that… that gives you a snapshot of what's happening now, but what we really need and what George's curve did so beautifully was give us a prediction of what would happen over time. So, unless you're doing LIDAR surveys, you know every year for multiple decades, and it's a bit of an issue.

So then we hit upon this idea of using the concept called self-thinning. Now, this, we didn't come up with the idea of self-thinning, this is actually an old forest inventory concept, that I think dates from the 1930s, where people were wandering around doing forest inventory as foresters do, and started realizing that there seemed to be some recognizable patterns for a range of species over time. So, what I'm saying here is it seemed like as the forest ages, there seem to be particular dynamic patterns that were, that could be represented in some nice curves that would allow foresters to understand what the wood resource, if you wanna put it in those production terms, would be over time.

And Raphael Trouve, Craig Nitschke, Pat Baker and Andrew Robinson in this case, and some others, started using this. These are people within our school and collaborators. We're using this to think about regeneration trajectories of our forests, and we hit upon this as being something that we could use, and that's what we've tried to do here.

So really quickly, self-thinning, you start off with an initial dense stand. Competition for light means that trees are trying to outcompete each other and the ones that are unsuccessful die and you end up with this, competition gives you this stand that over time gives you, you move from a lot of small trees to fewer trees but bigger trees. And anybody who wanders around a forest, particularly Mountain Ash forest and there's a range of ages will have seen this. And indeed, this can be captured by a curve like this which effectively says this, that you start off with small stem diameters, you got lots of trees and as you move along the stem diameter gradient you get less of them. That makes total sense. It's all about size site resources.

Now it turns out, though, that this can be captured in a model like this. Statistical representation of this, and really what this is saying is that there's a well behaved system at work here that we can capture with these statistical models. And so we've got the mean diameters of trees down on the X and the stocking density on the Y. And then if we put them in a log space on the right-hand side, we have this hypothetical self-thinning line. And then, with a lot of trudging around in the forest and measuring trees, you can actually build your own. And what we have here are two parameters - The slope of this line and the intercept. And we have used those as a way to represent the trajectory of forest dynamics.

And so just to reiterate a little bit again here, here's some examples of, of field studies that have populated this statistical representation. And then we'd link that with this actual, with these hydrological, well the structural attributes, which we're able to derive hydrologic properties, mainly being the sapwood area. And so, if we add those together, we have a way of modelling stand dynamics.

Now, I'm not gonna dwell on this. It's a water balance model, some of you will be familiar with these, some of you may not, but I guess what you're trying to do is model the main processes of the system in a way that doesn't, you know, make your computer completely blow up. This is a relatively simple model where we're modelling, basically modelling growth and water use on a daily basis. But the thing to look at here is the self-thinning part comes in here where we're dealing with the actual stocking density, sapwood area, and how that links to transpiration. That's what we're doing in this model. So, we developed this model, parameterized it for the Mountain Ash areas and fortunately here we have all of these incredible, fully valuable long term experiments up in the North Maroondah where there are whole bunch of catchments, small catchments I might add, but with different treatments, stand replacements, various thinnings, some control catchments, and then we had a very rich dataset on which to both parameterise and test our model.

Now, one of the key concepts that I've been talking about here is this idea of the self-thinning line, and forgive me if I'm kind of banging on about this a bit much, but there is a reason for this. So, we went and measured a bunch of, looked at all the inventory data, did a bit of extra measurement, et cetera, et cetera, and came up, we assume a fixed slope of the self-thinning line, but there, the other major parameter in this is the intercept. And so, we're able to come up with intercepts for a bunch of these catchments. But the four to look at are probably these last four. The black shaded bar here is a kind of average for all of those north Maroondah treated catchments or most of them.

And then we also have over here the 1939 - so this is where we collected all of the inventory data that we could on the 1939, from the 1939 origin Ash forests and they come up, they have a self-thinning intercept which is quite high. So really what this means is a dense forest.

We also looked at, we have two others here though that sit quite a lot small, quite a lot lower than the others. One of them is from that Corranderk catchment that I talked about before, where we imposed an Ash regrowth on marginal site. And lo and behold, the regen trajectories were a lot, a lot lower than in the other Ash, in the other catchments, that are much more sort of proper Ash country, if you like, but the other one is this 1851 self-thinning line parameters.

Now 1851 was the big fire before the 1939 fire and so if we think back to the Kuczera Curve, the big change in streamflow is relative to what happened before the fire. And what seems to be happening from the data that we've got, and admittedly these areas are limited, is that it looks like the forest density was a lot less in pre-1939, than post-1939. So, this is a kind of an important thing to keep in mind here.

Anyway, so we made this model, parameterized it, hit the button to see if we could model the system and looks like we can. This is a pretty good, pretty good model fit for any kind of model, so we were happy with that and so then we moved on to what is the main finding, I suppose, of the talk that I'm about to give here, that I'm giving, sorry. And that is that if we've got a model that works, in the top panels, we look to see whether we could replicate George's analysis. And in fact, we could.

So, in this case, the crosses are the data that George used to fit his model to, and the black dots are our model. And so, what we used here was the actual self-thinning line parameters, so observed ones. So, 1851, a less dense forest, followed by the 1939 parameters, which are quite a dense forest. And what we're able to do is pretty much replicate the Kuczera analysis. So, OK, what that means is that the Kuczera analysis made sense and is totally valid for the circumstance that we that we find ourselves in today, post-1939.

But then, the second panel down is, OK we thought, well what would happen then if we if we modelled a less dense regeneration, i.e. that we replaced the 1851 density with an 1851 regrowth. And in this case what you can see is that all of a sudden our model doesn't fit the observed data and really clusters around that zero line, so that, you could interpret that as well, if we had a less dense forest post the 1939 fire, we might have had very little streamflow change.

And then the bottom one is reversing those self-thinning line parameters where we said, OK, what if we had a really dense forest, i.e. the 1939 regrowth forests that we have now, and what if that was followed by, sorry, that was disturbed by i.e. killed, and that was followed by a much lower regeneration scenario, such as we think we had in 1851. And in that case, we would actually have a whole lot more streamflow than we would have had beforehand.

So, what all of this is saying, is that first of all, George Kuczera’s analysis seems sound. And so it should because of all the process work that kind of found that, you know, overall that supported it. But, we are beginning to think, and not only us, this is really based on a lot of discussion and publications from the, from the sort of forest dynamics people in our school and others, that this might not be what happens every time, that it may well be an artifact of the quite likely the rainfall regime and other climate parameters and seed availability and various, all of the things that go into the dynamics of regeneration that in 1939 it might have been really, really optimal for all of those things to produce a really dense forest. And we are, I suppose, hypothesizing that going forward that may not be the case.

Now, so there, there are these, these are our initial conclusions, that the Kuczera Curve, although we still think is valid for the 1939 circumstance, it's not necessarily universal. So, by that we mean if we have another huge fire that goes through our catchments, then we think there is a, at least, well I wouldn't put a probability on it, but there's a certain chance that we would not get a Kuczera-type outcome, and in fact we might have a quite a benign outcome on streamflow or even get more water. We think the self-thinning line seems to work really well in in doing this, so therefore it's quite a powerful tool and I suppose overall we would suggest we'd been as a water, forest hydrology and water availability community, we might have been overconfident in our past forecasting of fire and harvesting effects on long term streamflow.

Now just briefly, why could all this be? So, Richard Benyon and I did a bit of work on looking at the post Black Saturday regeneration and so did a bunch of other people, I might add. So, this is not the whole totality of these data, but it was what Richard and I found at least, was that there was a lot of variability - You don't need to like, look at the numbers too much - but there was quite a lot of variability and we found that in some places there wasn't actually very high rates of Ash regeneration and in some places there are lots of acacias coming up. Now, acacias are a species that could co-occur in Ash sites. Eventually they get outcompeted by the Ash, but they can come back in, they can regenerate in quite, quite densely, and that led us to a bit of a hypothesis that perhaps after the 1851 fire, there was a really widespread regeneration of acacias, which managed to get enough of a foothold to outcompete Ash seedlings in to some extent, and that that might have led to a more sparse Ash forest than we've seen, than what we see now. I won't go through this, but we did some analysis which seemed to suggest that could be a plausible hypothesis and at this, I have to stress, at this stage that's all it is - a hypothesis.

So, I'll just stop there. So, that's the whole Kuczera Curve thing, if you like, that we're left with, we think we're on to something and we have a hypothesis, but as yet we haven't fully run that to ground. There's quite a lot of work to be done in that space and we're still kind of hoping to do some more work on that, although funding is a bit of, is a bit of an issue on this one at the moment, but we're hoping to be able to secure some research funding to keep going. We will keep ticking away on it but, it's fair to say that, you know, a bit more funding would be handy.

I'm just gonna touch on a couple of other things now because I've really been concentrating on the Ash story, but one of the outcomes of George's work is that that model is so, it's such an elegant and simple model and really understandable, is that it's been used out of context a lot and partly this is, I'm not saying maliciously, I think it's usually just a little bit misunderstood, but also our understanding, particularly in terms of fire dynamics or fire hydrology dynamics , was pretty sketchy outside of the Ash. And we realized this when the big fire started coming in 2003 Alpine fire was the first one that we've had in the last two decades when people like me were asked to kind of model what might happen, we found ourselves really not knowing much about the rest of the forests, which are overwhelmingly fire tolerant, i.e. only a few of them die, instead of all of them.

And so, and I think up until now and even though a lot of questions in the wake of the Black Summer fires about, well, what's gonna happen with the hydrology? And the fact was that the vast majority of those forests weren't Ash forests, and so most of them survived pretty well, which means you don't get this huge regeneration impact, which reflects in in water use. And so we've done a bit more work on this over the years and some work that we published just last year, this was Jabbar Khaledi, he was a PhD student with our group, and he looked at this sort of really broad scale analysis of fire, the effect of fire, well his whole project was about, you know, what are the big drivers - Is it climate, is it fire - in stream flow in our temperate forests?

And I won't go too much into all of these models, but basically, Jabbar did a whole bunch of really great analysis and fitted models to all of the things that could be influencing streamflow. And I think the main thing to look at here is that, rather unsurprisingly, the amount of rainfall is the biggest driver of streamflow. But it turned out, particularly in in humid forests, that a wildfire was a really small, you know, less than 10%. So when he looked at what's driving your variability and streamflow, the presence of fire was not very important, particularly in our humid forests. As we get a bit drier, the importance goes up a bit, but remembering this is the, some of the, well, the majority in fact of these, in these subhumid and semi-arid areas are not in Ash forests. So those, that variability in streamflow that is explained by fire is more than likely to be a positive one.

So overall, the effect of fire, and this is much more in concert with what's been happening, with the research happening around the world, particularly in North America, is you're more likely to get an increase in streamflow than the big decrease that that the Kuczera Curve suggested overall. However, still, that the Ash catchments are a little bit of their own circumstance there. But overall, I think it's fair to say our research has found that the effect of fire has probably been overblown, and by people like me as well as anybody else, and that we probably need to be less worried about that in terms of streamflow than we might have been in the past.

And just very briefly, Rachel Nolan's PhD back in the 2010s, we looked at this, in particular at this mix species question, and really what we came up with is, yes, you can get a bit of a change in ET and therefore streamflow depending on the burn severities, but whatever it is, is only gonna last a few years. And really, when you look at a map of burn severities, there's a mosaic of them. So some, in in any given area of any size, usually there will be areas that have been burnt at high severity and areas have been burned moderate severity. And in the case of Rachel's work, suggested that perhaps a moderate severity might lead to a higher ET for a while mainly because there's more leaves there and then there's kind of the regrowing stuff going on around it, and that perhaps a high severity fire might, you might have less ET therefore more stream flow. This tends to cancel itself out, and I did some analysis for DEECA in the wake of the Black Summer fires, and that's really what I came up with, that if you if you base this on fire severities, there's probably not really an impact.

And then, of course, remembering that both in the wake of Black Saturday and particularly the Black Summer, we had really wet periods. So, issues around decreased water availability because of fire seemed to be ameliorated both by the kind of fire tolerance of our forests, of the vast majority of our eucalypt forests, but also there does seem to be a bit of a pattern of wet periods after the really dry periods.

That's all I've got to say. Of course, many people contributed to the research and the ideas that I've just presented, and there's a few of those references. If people want to follow up with me about reading about this stuff afterwards, then I can, I can give you a never-ending reading list on this stuff. But really, thanks for listening. I hope there was some usefulness in it and I'm more than happy to answer any questions.

Rachel Brown:

Wonderful. Thank you, Pat, for talking to us about all of these latest findings. And I'm glad you've got the papers there, so you've given us a taster we can dive into the details for more information.