 Instead of you looking up to the sky, I'm going to make you looking down, so I'm going to send you to the space and look down to the earth. So I'm going to be talking and giving you a brief overview of the current available and the next generation of space-based data and mapping technologies that are helping bushfire management to better prepare and respond to bushfires. So well, unfortunately, I said I will go to take you to space, but I cannot do that. And anyway, I'm not sure you prefer to be at home or at the space of the now, but then we already have ice in the skies. So there are a lot of satellite instruments that are imaging our planet and are helping us to observe and understand the earth processes. So this animation here is from the NASA's the United States Space Agency fleet of Earth Conservation spacecraft. And currently the satellite instruments such as these ones here in the animations and other instruments launched by other international space agencies issue an immense amount of data. So terabytes per day, so that remotely sends experts like me profess and convert into useful information or products that are useful in decision making and planning. So when it comes to bushfire management, this basal right of data collected from space can be used to inform different phases of fire management. So in a nutshell, this slide showed that before the fire remote sensing data can be used to monitor fuel condition, which will affect a fire danger likelihood, for example. During the fires, we can use remote sensing data or satellite data to identify areas of the land with with anomalous high temperatures, or to visualize smoke, which help to detect active fires. And also also to know how this the fire will spread based for example on the fuel condition and the weather conditions. And this information helped to most efficiently direct a separation and evacuation decisions. So finally, after the fire, a remote sensing data can be used to assess the fire stand, the mission from the fires, and the impact in terms of fire security, and also to tell us how the vegetation is recovering after fire. So moving to the applications of satellite data to assess a profile conditions, the main objective here is around estimating fire danger. So this is the risk of bushfire occurring. So a good fire danger rating system is a very important as it supports a range of critical decisions such as pre-positioning firefighting resources, issuing public safety warnings and information, or limiting the potential of emissions through the use of total fire bans. And we know very well what this means as we've been going through these total fire bans very often here in Australia and sometimes it prevents us to do that barbecue we wanted to do. And this slide gives an overview of the new fire danger rating system that is under development that include factors for fire weather, fuel condition, fire behavior, ignition, likelihood, fire suppression, and fire impact. So under this complex framework, remote sensing data is mainly used to estimate a fuel condition. And in case you don't know what the fuel condition is, I will explain now. So fuel is all life for the vegetation that accumulates over time and therefore can potentially burn at any time. And its condition includes moisture content, structure and quantity or load. So all these components of fuel condition affect the plamability of the landscape, and therefore the potential severity of a bushfire. So for example, thinking about fuel structure or the arrangement of the fuel that is separated is the fuel that is separated is less likely to carry a fire than a fuel that is continuous and packed. In addition, more fuel means larger flames and greater fire intensity. Finally, when fuel is high, there is less chance of a fire emission than if the fuel is low, fuel moisture is low if the vegetation is high. So let's start with the fuel moisture content. There are various methods that have been developed to estimate this variable from remote sensing data or data collected from satellites. And I'm going to try to briefly explain why this is possible because I thought perhaps some of you were curious to know the physics behind this. So well, when the solar radiation hits the surface of a leaf, a part of it is absorbed, or there is transmitting to other layers in the plant, and most importantly other fraction of this incoming solar radiation is reflected from the leaves surface back to the sensor that are on board of the satellites in the space. This fraction of the solar radiation that travels back to the sensor is called reflectance and is represented in this figure in the life. So basically this figure shows the reflectance for two plants with different moisture content. So you can see there are big difference mainly in this region here of the spectra. And this is because the water in the leaves absorbs a lot of the radiation in this area of the spectrum. And on the leaves tissue again or the, sorry, in the depending on the leaf tissue water content, the reflectance is therefore reduced to a varying extent. And based on these principles, we can create algorithms that convert the reflectance measured by the satellite into few moisture content values, and this is happening in near real time. So this animation here presents as an example the dynamics, the dynamic variation of few moisture content for Australia for 2019. And these maps were derived from data collected by a satellite from the NASA that is called MODIS. And this animation show us, as expected, the few moisture content values are constantly low in the desert central areas of Australia. But there are strong seasonality elsewhere. For example, in January, well now we are moving towards summer, we can see how the southeast coast get more red pixels. So now where are we? Hold on. This is September, December. So we are moving to a sour as we move into summer there is more red areas in the coastal area. But as we move into summer, there are less values and the temporal pattern is the opposite in the tropical region. So in the north of the country, higher life moisture contents are observed during the northern wet season that is December to March, and lower values are during the dry season that is a program. So these seasonal patterns of few moisture content have been demonstrated to be linked to far landscape flammability and before far occurrence. For example, unusually dry, fuel and hot weather was one of the factors that explained the high far activity we had in the south is Australia during the last five seasons. So this map, this figure here, for example, shows you the fuel moisture content nationally for the different years. And we can see how in 2019 the fuel moisture content was the lowest in record. When focusing in the southeast forest, these figures tells a similar story. So in black, we see the satellite base average fuel moisture content for the southeast forest during 2000, 2018. And in blue, we see the low moisture values at the same forest, the same forested areas reached during the 2019, the 2020 far season. So anybody can get access to these maps, if you were wondering, and here I show a screenshot of the public website we have developed to facilitate the access to this information. As a quick tour, for example, you can search for any date, since 2001, you can search for any location here and you will have the map of Australia, that date you can zoom in any area of Australia. So as an example, here what you can also see is in black, the areas in black are represented total burned extent reported by the emergency authorities at a given time and the red flames are the active parts as reported. This is basically the same information that you may have seen in the Farge near me, but with this website, we also provide information of the fuel moisture content. So this screenshot that I prepared, I searched for the fuel moisture content maps during one of the days of the oral ballet far in the 25th of January. And as you can see the fuel was very dry that day, and that they may have the intensity of this far. As a comparison, I also included the same, the map of the same day, but in 2011, and you can see a huge difference, so now the map is pretty much blue, that means it has very high values. And again, if we go back to this map, we can see how red everything was, so the dryness of the landscape was high. So moving into a mapping a fuel structure and load. So this is this capture with a remote sensing method that we call LiDAR. The principle behind LiDAR is really simple, it's a lot simpler than what I explained before. So basically the LiDAR is an active sensor or an instrument that fires a rapid pulse of a laser light at the surface and measures the time it takes to return to its source. So as the light moves at a constant unknown speed, the LiDAR measurements can calculate the distance between itself and the target with high accuracy. So by repeating this quickly in succession several times, the instrument can build up a complex map of the surface that is measuring and this is called a point cloud. So up to a few years ago, you could only have accurate LiDAR data from on-ground sensors such as these two that I display here on the bottom of this slide. But nowadays there are satellite LiDAR observations that are great. I am now greatly increasing with the written launch of, for example, the Jedi mission that is an NASA mission on board of the International Space Station. So LiDAR can reconstruct the three-dimensional structure of a forest, providing extremely detailed information of the forest's structure and load. And here I show you an example of a point cloud of the Black Mountain, so you can see the Telstra tower up there. And here, this animation, yes, it's a specific location assuming to this point cloud. And to this point cloud, we have basically run a complex automatic algorithm to classify the different points in the cloud into the different layers of the forest. And then from this, we can start to extract properties of the fuel, of the different fuel layers that are relevant for fire behaviour like the authenticity of cover of the near surface fuel or the elevated fuel or the height of the trees or things like that. By the way, I just take a time to remember you that if you have questions, remember to write them in Facebook and I will answer them when I finish. So apart from using LiDAR, the right information on fuel structure and load in fire behaviour modelling, also the maps are very important to provide useful information for separation of trees. For example, this is a map of the elevated fuel load, derived from LiDAR. So basically in red you see areas with high loads and in blue you see areas with low fuels. And these maps were used, for example, in 2019, you may remember that there was a small fire, a small in comparison with what we had this year, of course, around the square rocks. So these LiDAR maps were used to locate a site free of trees, to win the specially fire fighters in because it was a remote fire. And also it was used to try to pick up the easy line to construct the walking track to the fire. Okay, so now moving into the applications of remote sensing during the fires, the objective here is to use a satellite based data to detect active fires. And determine how the fire is going to spread and also find both what we call soft and hard containment lines. So soft lines, containment lines can be, for example, differential in fuel moisture content. So there is a part of the forest that is wetter that can act like a soft containment line. So when the fire hits that wet area, it may spread slowly and it will be easier to contain. And the hard containment lights normally refer to roads and paths and things like that. So most people may have seen, let's see this is an animation. So most people may have seen these maps here on the right during the last five seasons as they were very popular in the media. So these maps are hotspots. So what it is really is thermal anomalies used to identify active fire. In the Amazon, I have borrowed from Robbie, so as a time series of the active fires during the last fire season. And it's just spectacular to see how the activity grow over time. So one of the important aspects when it comes to detect active fires using satellite observation is the frequency in mimics acquisition. And of course, the more frequently the satellite collect imagery of a specific area, the higher the chances that you will be able to detect a fire as soon as it ignites. So this is an animation. So you on the left, active fires detected by the NASA Modi satellite I mentioned before. This sensor is on board of the Terra and Aqua platforms that views the entire surface every one or two days. So for the frequency of observation, it's a bit limited for bullfire, active fire detection. So on the right, on the other hand, you have the detections by Imawari-8 that is a Japanese geostationary weather satellite. And that the satellite is pointing always to the same location on the earth and therefore offers significant improvements in the frequency of observations. This satellite provides an image every 10 minutes. And therefore, it is more useful when it comes to the spread of a fire, as you can see in this animation. These are other images offered by Satellite, but I'm sure most of you might be familiar with them because they were also in the media during this last fire season. And these images here are from high-resolution optical data over Bayman's Bay during the 31st of December. And you can clearly see the smoke of the fire front and even the fire cumulus clouds or fire or what we call fire clouds associated with the fire activity. Of course, after the fire season, you can use this information about fire activity or hot spots during a specific period to know how the season was. And this slide that comes out of a report on the state of the environment we released a few weeks ago summarized the rank of fire activity by a region in Australia. And clearly shows us that while fire activity last year was below or average across most of the inland due to low fuel loads because of the dryness. It was the highest since at least 2000 in Tasmania, the East Coast and parts of Western Australia. So finally during the first phase of fire management, the objective here is to map the effect of the fire, basically how much area is being burned, in which severity. And then also to compute the missions resulting from the fire and also to look at how the vegetation is recovering after the fire. So in the same way that we can detect changes in fuel moisture content because plants with different water content have different response. Here we can detect fire severity because the reflect or burn area because the reflectance, reflectance spectra for and burn vegetation canopy and fires affecting different vegetation strata are very different. And those difference differences can be used to map a total burn a stand on fires. So this animation show an example of the total burn spent of the fires with Sydney last season. The yellows indeed show the active fires. Similarly as I previously show the time of the acquisition and the black shows the area damage at its time. So using that also from the NASA Modi sensor, we recently analyzed in this report I mentioned on the Australian environment in 2009. We estimate the burn area per land cover to clearly show that the bush fires we just had were unprecedented in the forested environments of Australia. And, and this analysis only included the area burn up to 2019 that was the, the period. But as I mentioned before, fire is not really a binary process. So the analysis of fire impacts require better discrimination of the variation of burn severity. Satellite data can also provide this information. For example, this is some map of the severity of the 50 fire that happened in Queensland. So this fire went a bit of convective on the 18th of November and burn right up to two significant dams, which are water resources for for two areas in Queensland. So we provided these maps, because they help the Queensland fire emergency services to talk to the local council about targeting remediation and efforts to protect the water supply. So basically, it tells you where you need to be more quicker and making remediation activities because the fire has been affected this area more severe than others. So one important aspect of any satellite based maps of any kind of variable is validation. We need to remember that the satellites records on information that then needs to be converted into the collect data that then need to be converted into useful information and for that we use algorithms. So once we derive maps of whatever we are targeting, in this case, for severity, then we need to do validation and these are some images. So some of the observations we took of the severity of the fire, the Aurora Valley fire here in the city in February, basically we flew with an helicopter and we took visual estimates of the fire severity at the same time that we also took photos with a normal camera. And these are the tracks we did during one of the days we flew. So this is the sample of the severity map for the Aurora Valley that we derived using in this specific case. We use data from the European Space Agency's Sentinel-2 sensor that provides imagery every five days at 10 meters as a resolution on the ground. And again, overlapping this map, I have those dots that are the observations we took from the helicopter. So again, once you do the modeling and you have a map, you always need to do some field validation. And here, for example, this spot that is in the green area that this area affected with low severity, we have this picture taken from the helicopter that shows that indeed that area was not heavily affected. We only have a few panopies that were scored. This other dot that is in the yellow area that are medium severity. Here we see that most of the, well, all the panopies are scored. And if we take up dot in the red area, we see how the vegetation is completely gone, so there's no fuel left there. So the fire was more intense. So I guess just to finalize my take home message, I guess, is that the increasing challenging fire management situation are calling for proactive approaches to reuse the likelihood of catastrophic bushfires. I'm not sensing information has already and will be in the future support for management in Australia to providing additional intelligence to better plan, prepare and response to bushfires. But my, I guess my key message is that on top of using better information technologies, governments and individuals also need to take serious actions to tackle the underlying problem with this climate change. So I think that was everything I wanted to cover tonight. I hope you enjoy my talk. And now I'm ready to take questions. Thank you for listening. All right, so we have a few questions already. So the first question is whether in the future, how can we reduce the potential for bushfires? Well, this is an excellent question and that really touches in my last point. So with better information, we can be better prepared and we can better plan and better respond. But we will not stop us to happen in the future. So for that really, we need to tackle a climate change and make a serious efforts to reduce the increase in the future. So the next question is whether there are predictions for the next summer 2021. Well, I don't think we have predictions yet. But it's very difficult to know what is going to happen because in one, as I said at the beginning, fire risk depends on many factors. One is the fuel load, the other is the dryness of the fuel. On one hand, there has been so many fires that the fuel has been reduced dramatically so there is not much fuel to be left to be burned. So on the other hand, the areas that have not been burned may be drier if we still have a dry winter dry spring and dry summer. So it's still a bit early to know what's going to happen. The next question is if more prescribed burns have been conducted in the areas burned last fire season, would the burn areas be low and not have resulted in fires we had? Well, there's a very hot debate around that and the answer to that is not simple. It really depends on a specific case. In the National Ballet specifically, we show that the areas that were recently burned with prescribed burns in recent years were affected with less severity than areas that were not burning the recent years. This has not been observed in all the fires, so it's very hard to have a direct relationship between a reduction of fuel and the effect on the fire severity and occurrence. Will you always need to verify the satellite data or will you eventually be able to rely on satellites alone? Well, that's an excellent question and I think we always need to verify the satellite data. This is in the research development phase. So once you have validated your algorithm and you know how it works, how accurate it is, then you can just run it and forget about the validation. But again, that initial evaluation is very important to also give an idea of the uncertainty of the algorithm.