 Welcome everybody thank you for inviting me to give this presentation. So what you see in the top left is the symbol of SEOS, the Committee on Earth Observation Satellites, which is essentially the way the public funded space agencies in the world collaborate through SEOS. And so these metadata standards have been developed under SEOS and endorsed by SEOS. So I'm going to show you a time series of satellite data in Morton Bay where you see Stratbroke Island and Morton Island. And you see data over about 20 or 30 years where you see the send banks moving, you see a bit of clouds, you see some kind of differences in the water, but on the land in general you didn't see any difference. And this is because all these satellite data have been processed to analysis ready data which means all the atmospheric effects have been removed, the geometric correction has been applied and a lot of other corrections have been applied in order to make the data comparable so that you can use one algorithm for the data for all the data, all the images instead of having to process each image. There was just an example, I'll give you a few more examples as I go on. So the Quartek Reflectance Product Family Statement, BFS's Product Family Statement, which is an official term from SEOS, was done by quite a lot of people from Australia, geoscience Australia and SRO, but also the Joint Research Centre of European Communion, Brockmont Consulting working on a lot of European Space Agency projects. American colleagues, Kent State, universities in Berlin, NASA folks, Belgium, Vito, Netherlands, Water Insights and Estonia, Tietcuzza and Italy, so you can see this was a very international led initiative. So analysis ready data from satellite data. So satellite data have always had the promise of providing long continuous time series which you can analyse for vegetation or for agriculture or for aquaculture or for coral reef bleaching or for many many other applications. However the data from satellites is measured at the top of atmosphere. In order to make an under, sorry at the top of atmosphere, so the satellites are 800 kilometers above us, they go around 30,000 kilometers per hour, so they circle the earth in 100 minutes and they make a tremendous amount of measurements, but under different light conditions, different solar angles, different skylight conditions, there can be haze, there can be dust, there can be smoke in the atmosphere or can be a very clear atmosphere, there can be sun glint on the water surface, you name it there are a lot of effects which you all would like to remove so that the data is all becomes comparable from day to day, over the years so that you can do long term trend analysis and change detection and that's why you need to go from top of atmosphere satellite data to bottom of atmosphere fully corrected data which is called analysis ready data and then you can better capture the where and when things happen and what happens and that can help you answer the question of why, why did it happen and what to do about it and if you did something about it did it help so this supply chain for earth observation is quite complex and there's a lot of specialized preprocessing and the data sets are large, the Lancet archive for Australia which is one image every 16 days for each location in Australia over about 35 years now is petabytes of data if you have to preprocess all of that data each time you want to use it that is a completely impossible task so what you want to do is preprocess the data and then make that available either by the cloud or by some other method so that people can select the data and process it further to for their information requirement and if you have analysis ready data then the big data aspect of this information can be exploited so analysis ready data is a step before you analyze the data for a particular use be it commercial use or a management use or an agricultural use or recreational use etc and this tries to show that if you have analysis ready data and you can process it all centrally to analysis ready data and then multiple users can use that data as they wish this is implemented in a digital earth Australia in Australia so analysis ready data has now been defined properly it's generated from raw data and processed so that it can be used without the need for further process in the context of water quality it's systematically radiometrically, atmospherically, geometrically and spatially corrected full archive earth observation data sets of either normalized water living radians or reflectance and effectively when we see color we see reflectance the satellites are just much better at it than we are interpretation ready data represents derived products so you use the analysis ready data then you apply algorithms to go to a water quality for instance which I'll show in a minute and some of these water quality constituents concentrations we can do nowadays is the total amount of algae or the cyanobacteria color dissolved organic method of tenants in the water transparency turbidity etc this is an example of interpretation ready data made using analysis ready data and this is Lake Hume and an algal bloom on the 27th of February in 2020 this is 10 meter resolution data you can say an enormous amount of patents in the water different concentrations of algae and where it's bright green it's effectively a cyanobacterial bloom going on in the lake this kind of spatial information is check can change how we manage our water bodies because the Lake Hume management authorities sample at one or two locations and as you can see here they're sampling if it's in the northeast they will actually shut the lake for recreation and fisheries etc because it's high cyanobacterial content if they sample in the south they'll say the water is fine for recreational purposes but this is the true story which is much more complex just as an example so one of the reasons for having to do this and having to establish these global metadata standards is that at a certain moment multiple providers of earth observation data were for instance there was an example in some lakes in Africa which were in a high demand to get information of those lakes different providers were providing different forms of earth observation data from the same satellite but reaching different conclusions now if you don't know how they process the data you can't compare the results so it's one of the main reasons for saying we must have global acceptable metadata standards because we must avoid confusion with the end users so this is done to access the very large datasets but also to make sure that the end users get consistent information which is really really important so I'll now go a little bit into the framework of this Committee on Earth Observation Satellites Analysis Ready data and there are product family specifications which are the full documents I'll give you an example of one and in our case we're looking at one of the optical products because remote sensing water quality is done by optical earth observation so this is the definition it's processed to a minimum set of requirements and organized into a form that allows immediate analysis with a minimum of additional user effort also very important and interoperability through time and with other datasets then satellite data becomes really useful so how does that how does this work well we've set metadata standards and now providers of data can do a self-assessment there's a whole there's a whole table for that on the web they can do a self-assessment how good they are and how far they reach these metadata standards then their self-assessment goes to an international peer review panel and they then verify whether this is correct and then they can get a effectively a stamp of approval that they are compliant I'm not going to go through all these metadata fields but I'm trying to give you an idea of what's involved which is this is the general metadata which is similar for most earth observation types so also for terrestrial and agricultural and geological applications so these are all the metadata you require including some actions and you've got threshold and target threshold is that's the minimum target is optimum if you can reach that that would be fantastic so the middle column is threshold the right column is target in the next slide I don't have that threshold and target although those columns are still there so this is general metadata which we've been trying to keep as similar as possible for the aquatic as for the terrestrial and other applications then we go to per pixel metadata so for each pixel and so we have massive amounts of pixels petabytes of pixels for each per pixel metadata we do want this information so you also have to look at clouds of course because you can't see them but you also have to look at cloud shadows you also need a land water mask but you also in especially northern hemisphere you need sea like and river ice masks etc etc so you have to identify do you see sun glint at the water surface do you see white caps from waves etc is there floating vegetation on the surface which can if you're looking for water quality those algorithms can't deal with floating vegetation so you need to flag this so there are many of these per pixel metadata then really important for remote sensing you have to do all these atmospheric corrections so you need to be able to know your water vapor your ozone corrections other trace gaseous corrections carbon dioxide for instance or methane or etc so you have to do all these corrections as well and this is a more detailed this is drilling into just one example where you say for instance you know the solar and viewing geometry that the metadata provides every solar and sensor viewing estimate and scenic angles which is quite logical if you think about it and that so there are multiple pages of these kind of metadata that have to be that have have these criteria in them and then the providers will self-assess against them this is lake burley griffin another time series provided by geoscience australia we're looking at total suspended metal levels and you can see the image changing on the left on the right you see a Huffmiller diagram which gives you along the black transect from the right of lake burley griffin to the left it gives you the total suspended sediment patterns and from this you can actually find weather types when there were lots of floods from 1987 to 1991-92 massive amounts of suspended sediment in the lake the millennium drought is visible from 2001 to 2013-11 you can see the lake was very clear because there was no major influx of suspended matter then you saw in 2010 and 2011 there were large inflow events with lots of stability again and you can hide total suspended matter and then you've got a relatively calm player again and we're now extending this time series to present and you would see in the last two years you would see massive increases in suspended matter again so you can see this kind of data can really be used to provide a lot of information so the development of this analysis ready data for aquatic reflectance was initiated in march 2020 it was endorsed one year and three months later and it applies to very specifically remote sensing data over coastal and inland waters and the version one is now available from the seos ard website so you can see there are also analysis ready data products now for surface reflectance surface temperature normalized radar backscatter polarimetric radar aquatic reflectance ocean radar backscatter and nighttime light surface radiance and many more are in development european space agency is actually beginning to use this to do a self-assessment as is the united states geological survey for their land-set products now we did use the land product family statement for terrestrial remote sensing as much as possible but we had to modify all these fields because they did not appear or were used for or need to be used in terrestrial remote sensing but they do are required in aquatic ecosystem earth observation so here you see some requirements we had to modify and new requirements that had to be identified and there are some changes that we had to do in the radiometric and abstract corrections as well so we had to do quite a lot of adjustments so the effect of all of this is that this is another way to mature the use of earth observation coastal inland waters because there's now a international metadata standards people groups that provide this data can now do a self-assessment which will help operationalize the information it will aid in the comparison from different providers in what they output the providers are all going to have to meet this new standard which didn't exist before and that will lead to trust by end users because they know it's now compliant and the most difficult aspects we have are still the surface effects at the water interface and effects in the atmosphere which we still have to work on and the interesting I'll get to that later there's an interesting consequence of this for research these are all some of the issues that we now really have to focus our research on so having these metadata standards getting the global community expert community together we then actually are identifying the weakest spots in our processing of earth observation data which is fantastic because we never had that in such a consolidated way which of course can lead to research programs with space agencies or research agencies or universities so it's fantastic it was an unexpected result that having these metadata standards identifies the area of research and development investment that is going to be needed and even if we do have the metadata fields filled in for threshold which is the minimum there's still quite a lot of research to go to target which is what would be ideal so this is a fascinating consequence we did not even anticipate when we started this process this is the date page on the seals website where you can go to and you can read much more about this and that's the end of my presentation