 It is my pleasure to introduce our next lecturer, Donata Giglio as an assistant professor at the University of Colorado. She worked as a postdoctoral scholar at the Scripps Institution of Oceanography, and her interests are mainly focused on large scale ocean dynamics, ASC interactions and geophysical fluid dynamics. And one of her favorite foods is sushi. Donata, I'm looking very forward to your talk. Thank you, Judith, for the introduction, and thank you to you, Anisha and the program for inviting me to give this lecture. Hi everyone. Today I will be talking about ocean observations for S2S. Before I start, I would like to thank all that contributed to the studies that are cited in these presentations, and if you are interested, there will be references in the slides. So we will start with some basics on why we care about ocean observations for S2S. We will then discuss what observations are available versus desirable. At the end, I would like to briefly discuss uncertainty quantification in 3D oceanic fields that are based on observation. So the goal is here to improve understanding modeling and forecasting of S2S variability as you have heard last week and also today. So observations, as we will see in this talk, are important for this, as they provide the ocean is a source of predictability. The community has seen that coupled atmosphere ocean modeling and data simulation may be actually key to forecast S2S variability. It's always important, as you have heard at the beginning of last week, as many critical decisions need to be made several weeks to months in advance. For instance, for naval and commercial shipping, we want to be able to design the shipping routes weeks in advance to take advantage of favorable condition to avoid hazards for drought forecast. It's important for farmers to know in advance about the forecast so to make the best decision in terms of which seed variety to use. Finally, one last example, water resource management. The decisions about reservoir levels have to be made weeks, months and seasons in advance of the actual usage of the water. We all know that we are facing a climate crisis. And as a result of this, there are more severe and frequent extreme events that are developing. NOAA has recently published a report that underscores the importance of ocean observations for forecasting, including for forecasting at S2S time scales. And quoting directly from this report that was published in May, NOAA states that NOAA must embark on a new way of doing business by fully integrating the ocean observing communities with its weather, water and climate programs in order to provide society with state of the science actionable data information and services. NOAA goes on to elaborate on this and to point out how the insufficient integration of the ocean observations in numerical models is a clear limiting factor for improving the skill of earth system prediction. And this is across time scales, including S2S. In this report, they show one summary graphic for the importance of integrating, better integrating ocean observations in model and data simulation efforts to predict, to improve the forecast on different time scales. In particular, they show this graphic where on the horizontal axis, you can see time here, specifically on the vertical axis, the number of events that are billion dollar disaster events in the United States during the period 1980 to 2021. And on the right, you see the cost in billions as a consequence of these events. One thing that the NOAA report points out and I would like you to pay attention to looking at this graphic is that 2020 was the sixth consecutive year in which 10 or more billion dollar disaster events have impacted the United States. This is clearly a very important topic for society and NOAA in this report describes the strategic goal to improve NOAA earth system prediction to provide more accurate and I quote reliable and timely forecast across time scales. And to do this, they highlighted the focus of closing gaps in ocean observations and also better integrating the available ocean observations into the medical models. Here on the right you see a schematic of the strategic objectives that are described in the NOAA report. And you can see how aligning and enhancing and maintaining ocean observation is one of the key points on the right. But let's think about why this is, why this focus on ocean observation. So the subsurface ocean stores and releases heat and this is the key reason to understand the important role that the ocean plays in regulating climate. The ocean is a major reservoir of energy for both weather and the climate system. If we want to think about some numbers and think of an example that in one sentence describes this if we think about the top 100 meter of ocean, top 100 meter of ocean at 40 times the heat capacity of the entire atmosphere. And you may have heard of the heat capacity of the ocean in the context of climate change. The climate is changing as we're all aware due to the fact that the input, the energy input to the climate system is greater than the energy that exits the climate system. But where does this heat go? The oceans actually take up more than 90% of the energy that is added to the climate system by humans. For comparison, the atmosphere takes up 2%. You may have seen this schematic on the right from the IPCC report where on the horizontal axis you have the ear, on the vertical axis you have the energy. And this graph shows how the different components of the climate system, the energy stored in different components of the climate system has changed over time. You can see prominently how most of the energy has been stored in the upper and the portion. So these would be the light and dark blue shading. But let's think of an example that has been discussed also during this workshop for S2S forecast. Let's think about where tropical cyclones get their energy from. Tropical cyclones get their energy from the ocean. We have that water evaporates and latent heat goes into the atmosphere. Once this water condenses, this heat is released and fuels the tropical cyclone. And if there is a nice video that explains this mechanism and you can easily find it either looking at the link down here or literally googling where the hurricanes get energy, get their energy from. This is a NASA video and I encourage you all to look at this animation. We can see the effect of the ocean changes as the tropical cyclone comes through. Here we have one example for tropical cyclone Harvey in 2017. We have a map on the left with the track of Harvey and then the color is the maximum sustained winds in knots. And you can see that there are also these magenta stars on the map. And these are ocean observations from the Argo array that are collocated with the track. So these are Argo profiles basically. If we look, if we focus on the Argo profiles that are available next to the track, close to the track in the Gulf of Mexico. We can look at how the ocean changes between before and after the cyclone. And this is actually included here on the right. So on the right we have a graph with temperature on the horizontal axis pressure on the vertical axis and you can think of pressure in decibel as depth in meters more or less. And what you can see is that if you compare the profile before the tropical cyclone with profiles after the tropical cyclone so you clearly see this cooling. And if you're interested in learning more about Hurricane Harvey, I encourage you to look at the publication by Trenberg et al. in 2018, which describes all the observations that are available during Hurricane Harvey. And if you're interested in more in general to look at how the ocean changes as the tropical cyclone comes through, you can look at this kind of plots for other hurricanes using the Argo with web app and database and the Python notebook that is in this that you can find in this in this GitHub repository down here. The ocean is very important to understand how tropical cyclones evolve. And to improve forecast of for instance tropical cyclone energy. Here, I wanted to show I want to show one study that compares the anomaly correlation skill of forecast of tropical cyclone energy at day 16 to 45. So what you see in this map are different regions and then bars that represent the anomaly correlation skill at this. And the different color of the bars indicates a different kind of experiments. In one case, the orange bars represent what happens, what the correlation skill is, if we initialize the ocean using all the observations available in the system. And then we have the green bars that instead show what happens to the skill if we do not include in the initial condition, the information from Altimetry. And you can see that in almost all cases, the orange bars are taller than the green bars indicated the importance of ocean observations to better initialize the ocean and achieve a better forecast. Another example is for seasonal forecast in the tropical Pacific, specifically a seasonal forecast of El Nino. We are all familiar with the normal conditions in the tropical Pacific and specifically the warm water volume in the western Pacific and different studies have showed that the recharge and discharge of this warm water volume in the tropical Pacific is closely related to ANSO variability. So accurate observations of the evolution of this warm water volume can help to improve ANSO process understanding, first of all, and predictions on seasonal time scales. This is one example that we have heard last week about to ocean observations are key to better understand and represents in models RC interactions. We have learned about the MGO and the propagation of the MGO, particularly through the maritime continent seas and understanding those interactions in these regions are key to understand the propagation of the MGO through the region. And this is a part of the world where we haven't had historically a lot of observations and the international program here so the maritime continent has provided for the period 2017-2020 long needed observations in these regions that are helpful to better understand RC interaction. Finally, I want to mention sea ice related predictions. These need to have good initial conditions in terms of operation currents, temperature surface radiation and it fluxes surface wind way by. We need observation at the edge of the ice. We need observations in the marginal ice zone and increasing the amount of observations we have in time will allow us to do a better job at predicting extreme events like the unprecedented springtime retreat in Antarctic sea ice in 2016. We have a schematic a graph showing that here on the bottom on the horizontal axis we have the month in 2016 on the vertical axis we have sea ice extent. And if we compare the the mean for the period 1979 to 2015 which is the black line with the red line that is for 2016 we see this unprecedented springtime retreat which we would want to be able events like these are events that we would want to be able to do a better job at forecasting. So to summarize, we really need ocean observations to understand the interactions and the rule of interaction in S2S variability and prediction. We need ocean observations to improve S2S forecast in high and mid latitudes, including sea ice predictions as we just discussed. And also the NOAA report that I mentioned earlier, highlights how the ocean is among the most poorly known and understood components of the earth system. And this is largely due to the sparseness of the data to the difficulty to do measurements in certain regions or in certain oceanic layers. So what observations are needed in light of what we discussed so far, we want observations of the ocean atmosphere interface and we want these observations to be simultaneous measurement of the ocean and the atmosphere. We want observations in the subsurface. Just a few examples, upper ocean current and temperature, mixed layer depth and properties, ocean boundary currents and we want a resolution observations in ocean boundary currents which provide a source of predictability for S2S variability. We want measurements of way height in and around the marginalized zone. We want observations below the sea ice. We want observations that capture the processes of sea ice growth and retreat. And we want observations of sea ice thickness and concentration. So what observations are available at this point in time. Here are some examples. We have, for instance, moorings that provide a lot of information with a high temporal resolution, yet they have the limitation of being observations at a single site and maintaining the mooring is expensive. This is particularly challenging if we are in a region where there is a strong current, but moorings are a key source of information for Fc fluxes, for instance. They are a key source of information for the urinal variability because they provide the resolution that is needed to describe the urinal variability. We have gliders that are a key source of information for regions where we need high resolution measurements. I mentioned earlier, boundary currents and gliders are very helpful at doing that. We have, more recently, we have had sail drones. And these instruments, as we will discuss in a second, are very, very useful. For instance, to supplement the information in regions where we have moorings to provide information about the gradients between moorings. Surface drifters, they provide a view, a global view of mixed layer currents near surface temperature and sea level atmospheric pressure. We have ship data that although they are sparse, they are concentrated along specific lines and the sampling along these lines is sparse in time. The sections are repeated only every so often if we think about, for instance, the Go ship program that is a key resource of information for the global ocean. This kind of network is key, not only for the measurements that it provides directly and can be analyzed to study climate changes, for instance. But also these are key to provide a gold standard to assure that arrays like the Argo array of profiling floats remain at best quality, provide best quality data. The Argo floats have provided an unprecedented coverage of the global ocean in space and time. Here, if we look at the map on the top right, these are dots that show the location of Argo profiles collected in a three-day window. And just looking at this map, you can realize the magnitude of the revolution that Argo has initiated. Obviously, satellites are also key to gather information on a global scale. They have provided unprecedented coverage, especially and temporarily. So we have these observations and we shouldn't give them for granted if we think about what we had in the past. For instance, if we think about what we had for temperature casts per one degree box where if you think about casts that profile at least to 900 meters and we put together all the casts available from the world in the world ocean database. For the years before year 2000, we end up with a map like this where the black and we see a lot of black and blue dots in this map indicate that there are only in the case of the black color one cast per one degree box in the case of the blue color two to five casts in a one degree box. So all the observations available over time up to year 2000, especially in the southern ocean, there is a scarcity of data as you can see from this map. Even if you include the initial years of Argo you still see that the southern ocean is not well covered. And now if you compare instead with how much observations we are available, we are able to collect now in particular these are the number of observations for years 2015 and 16 so in only two years. We can now do a much better job at observing the ocean on a global scale, and especially we can do a better job in the southern hemisphere. So finally I want to show the casts from glider data. I mentioned earlier how gliders are very important to observe gradients in boundary current systems. And these are the gliders in the world ocean database, and we can see where these data are available. I mentioned about sale drones and how promising and useful these instruments are these instruments are autonomous surface vehicles that basically sale in the ocean and some power that the instruments on board are some power, and this can be used to target specific events if we think about a linear development. These sale drones could capture could could make measurements during the linear development and and as I mentioned earlier they can supplement other sources of observations, they could cover gradients in between fix the moorings, for example. So we want to make to increase the use of sale drones that are of course the sale drones have been used in specific projects and studies we want to increase the use of these observations, which can the sale drones can provide both observations in the surface of the ocean and in the ocean and in the atmosphere. So this collocated information simultaneous observations of the ocean and the atmosphere are very important. Also, we want to think about new mission and concepts to enhance the range of information that are available for from space. The current satellites do not provide measurements for all the surface variables that are needed to estimate RC fluxes. And I mentioned many times today, how important it is to have simultaneous measurements of the ocean and the atmosphere to have a better descriptions of different types of fluxes and concepts like the butterfly concepts are proposed to move to advance in this direction. Also, satellites, most of the satellites measure at a couple of times per day at any fixed location. And we want to have instead multiple satellites that have similar objectives that measure at one location for a different day so that we can describe the urinal variability and the scales of fast moving synoptic storms. Finally, I mentioned we need more observations, many more observations of the marginal ice zone, more observations at the edge of the sea ice to better understand sea ice growth and retreat. Okay, let's see. I have a few more minutes and I would like in this remaining time to discuss a bit about one more use of optional observations beyond what we have heard a lot so far in terms of assimilating these observations in data assimilating models. So obviously that's a very important part, very important use of ocean observations. Another way ocean observations are useful is to produce gridded products that can be helpful to assess oceanary analysis and improve the initial state of the ocean for S2S forecast. As Magdalena and other speakers have described, this is of key importance to improve S2S forecast. So when we have, when we think about Argo profiles for instance, I mentioned how Argo has provided unprecedented coverage in space and time. We have more than two million profiles. How do we go from these profiles to not only gridded products, which we have quite a few out there, but gridded products that come with meaningful uncertainties. How do we do that and what are the challenges? So if we think of the example of field of ocean it content estimates, that is of key importance, obviously, because we know that as I described in the beginning, we know that the ocean has taken up more than 90% of the excess energy in the climate system. So we want to be able to do a good job at estimating how the ocean it content changes over time and estimating the uncertainty of the ocean content. The challenges are that there are different sources of uncertainty and not all of this and not for and we haven't yet produced the products that take into account all these different sources of uncertainty. So one, a certain sort of uncertainty is the quality control of the data. And as Magdalena pointed out, this could be a problem also for assimilation of the data into a model if bad data are assimilated that will degrade the what we can tell using the model. There are errors in bias correction for some data. And one example is for the XBT data in the past. We have, as Magdalena mentioned, special temporal data sampling density that changes over time. This is just one schematic that shows on the horizontal axis, the year on the vertical axis, the number of casts per year. And as you can see, as the Argo array and the glider kick in, we have a large increase in the number of available observations. And this is challenging, not only for data simulating system as Magdalena described is also challenging when producing maps. Another key component to this that we need to consider is when we map these observations, one of the first steps is to generally to remove a baseline climatology the choices that we make in for that baseline climatology have impact on the end product and have impacts on whether some of the assumptions of the method that then we used to map the residuals are met or not. But let's have a look for this. For the trend, for instance, of the data for oceanic content. This is a very relevant. Very important question. What's the trend in oceanic content and this trend. If you use cross validation, you can show that not including the trend in this baseline climatology leads to an underestimate of the actual warming trend. So the choice we make have are very important for the end result. Also, the choice of mapping method, you could think of going with a very simple box average versus objective mapping this in as impacts on both the estimates and even more on the usability of the data that you can produce along with that estimate. You can make choices about how to estimate the correlation scales and this again as implications on on what happens next. Finally, one source of uncertainty is the special temporal correlation of the measurements. And in a recent work that I'm doing with collaborators. We have seen that using local conditional simulation can help to provide uncertainty with the estimate that are taking into account the special temporal correlation of the measurements. So to summarize, we need to design and implement low cost I value surface and subsurface ocean observations. And not only that. So we need to, of course, maintain the observations we have, we need to enhance these observations adding more measurements, especially to capture. the observations in regions that are sources of predictability, but also we need to develop forecasting systems that are capable of extracting the observations potential in forecast applications. We want to take advantage use what we know from the models to also guide the design of observing system, we want to combine the information from the observation of oceanography with the modeling community to design the best observing system for the future. The, as I try to highlight today. SOS forecast will benefit from high special temporal resolution in regions that are sources of predictability and these include coastal regions, tropical regions and polar regions. I made the example of gliders and how useful they are in boundary current systems to capture strong gradients for instance in sea surface temperature in your surface temperature. We want to think beyond the current technology and advance current technology to observe the ocean atmosphere interface simultaneously using autonomous vehicles but also potentially doing this from from space. And finally, we discussed briefly about observations based gridded products and the importance of producing observation grade based gridded products that have meaningful uncertainty so that we can assess ocean analysis and improve the initial state of the ocean which is important for SOS forecast. So we'll stop here and take questions. Thank you. Thanks very much. Very interesting aspects about ocean observations.