No abstract

<div> <p><span data-contrast="auto">T</span><span data-contrast="auto">he ongoing project </span><span data-contrast="auto">“deteCtion and threAts of maRinE Heat waves – CAREHeat”</span><span data-contrast="auto">, funded by ESA in the framework of the </span><span data-contrast="auto">Ocean Health initiative, aims at improving the current Marine Heatwaves (MHW) detection and characterization methodologies at the sea surface, at analysing MHW vertical propagation through the development of 4D temperature fields by using Machine Learning approaches</span><span data-contrast="auto">, at providing a global atlas of MHW at the sea surface,</span><span data-contrast="auto"> at advancing the understanding of the physical processes involved in MHW development and at assessing the MHW impact on marine Ecosystems and Biogeochemistry. </span><span data-ccp-props="{"201341983":0,"335551550":6,"335551620":6,"335559739":160,"335559740":259}"> </span></p> </div> <div> <p><span data-contrast="auto">This presentation will focus on the first phase of the project. The mostly used MHW detection method (Hobday approach) has been revisited by carrying out sensitivity studies on different threshold parameters such as the choice of the percentile threshold and the minimum duration of the events. Specific work has also been done to investigate the impact of sea surface temperature (SST) trends and prominent climate modes, as El Nino Southern Oscillation (ENSO), in order to disentangle the slow-varying SST component and quasi-periodic oscillations from the abrubt changes that are characteristics of these extreme events . Many metrics are provided along with the global atlas to help the characterization of these events. In parallel with this work, a machine learning approach based on observations has been used to reconstruct a 4D temperature field from the surface up to 300-m depth and MHWs have been estimated. Subsurface MHWs can also impact ecosystems and phase shifts with the surface events can be observed. This product helps to analyse this propagation in depth. </span><span data-ccp-props="{"201341983":0,"335551550":6,"335551620":6,"335559685":0,"335559739":160,"335559740":259}"> </span></p> </div> <div> <p><span data-contrast="auto">The work is focused on three areas of interest: the tropical Pacific, the western Mediterranean, the Madeira Island region. In these regions, the main outcomes of the 2D and 4D analysis will be presented</span><span data-ccp-props="{"201341983":0,"335551550":6,"335551620":6,"335559685":0,"335559739":160,"335559740":259}"> </span></p> </div> <div> <p><span data-contrast="none">Please visit the CAREHeat website (</span><span data-contrast="none">www.careheat.org</span><span data-contrast="none">) and follow us on Twitter (@ careheat_) to stay up to date about the project research and results</span><span data-ccp-props="{"201341983":0,"335551550":6,"335551620":6,"335559685":0,"335559739":160,"335559740":259}"> </span></p> </div>

<p>Lakes and reservoirs monitoring is of sheer interest, as in-situ gauging station coverage is dwindling at a global scale. Water storage change show the impact of not only domestic consumption, low water maintenance in rivers or crop irrigation, but also the impact of climate change. In this frame, different approaches are explored in this study to be able to follow from space remote sensing data the lake storage change, either from water height or surface.</p> <p>For the ESA CCI Lake Storage Change option, we focus on a few lakes distributed around the world to establish a methodology suitable at a global scale to different lake behaviors. In particular for highly varying water bodies, the automatic production and use of hypsometric curve is investigated This approach is yet suitable for volume variations only.</p> <p>Complementary to this first approach, an image inpainting algorithm applied to digital elevation models around water bodies is developped to assess their total bathymetry (either for lake or reservoir) where the pixels to be reconstructed are the ones underneath the lake surface. The first results show encouraging estimations that may lead in the near future to the assessment of total water volume of lake and reservoir at a global scale. With the recent launch of SWOT that will provide an unprecedented coverage worldwide, the estimation of global water storage change has a bright future.</p>

This thesis report explains how to control the movement of an underactuated robot using Reinforcement Learning (RL). It presents the theory of RL, the computer model of the robot, the architecture of the code and the two approaches followed to solve the problem. The simulations are done using MuJoCo in a python environment. They simulate the behaviour of the robot in two dierent state spaces: the rst experiments take place in a discrete state space, where the learning is done using Q-Learning and SARSA(λ) algorithms. The second approach is to treat the problem in a continuous state space and implement REINFORCE. With such algorithms, it is demonstrated that the initial chaotic behaviour of the robot can be controlled to move accurately in a straight direction. It manages to move slightly faster in the continuous state space, but with more variance. At the end of this report, some suggestions are proposed to go further and improve the learning process.

No abstract