Search

An exponential growth in computing power, which has brought more sophisticated and higher resolution simulations of the climate system, and an exponential increase in observations since the first weather satellite was put in orbit, are revolutionizing climate science. Big data and associated algorithms, coalesced under the field of Machine Learning (ML), offer the opportunity to study the physics of the climate system in ways, and with an amount of detail, infeasible few years ago. The inference provided by ML has allowed to ask causal questions and improve prediction skills beyond classical barriers. Furthermore, when paired with modeling experiments or robust research in model parameterizations, ML is accelerating computations, increasing accuracy and allowing for generating very large ensembles at a fraction of the cost. In light of the urgency imposed by climate change and the rapidly growing role of ML, we review its broader accomplishments in climate physics. Decades long standing problems in observational data reconstruction, representation of sub-grid scale phenomena and climate (and weather) prediction are being tackled with new and justified optimism. Ultimately, this review aims at providing a perspective on the benefits and major challenges of exploiting ML in studying complex systems., Comment: 25 pages, 4 figures

Near the ocean surface, river plumes influence stratification, buoyancy and transport of tracers, nutrients and pollutants. The extent to which river plumes influence the overall circulation, however, is generally poorly constrained. This work focuses on the South China Sea (SCS) and quantifies the dynamical impacts of the Mekong River plume, which is bound to significantly change in strength and seasonality in the next 20 years if the construction of over hundred dams moves ahead as planned. The dynamic impact of the freshwater fluxes on the SCS circulation are quantified by comparing submesoscale permitting and mesoscale resolving simulations with and without riverine input between 2011 and 2016. In the summer and early fall, when the Mekong discharge is at its peak, the greater stratification causes a residual mesoscale circulation through enhanced baroclinic instability. The residual circulation is shaped as an eddy train of positive and negative vorticity. Submesoscale fronts are responsible for transporting the freshwater offshore, shifting eastward the development of the residual mesoscale circulation, and further strengthening the residual eddy train in the submesoscale permitting case. Overall, the northward transport near the surface is intensified in the presence of riverine input. The significance of the mesoscale-induced and submesoscale-induced transport associated with the river plume is especially important in in the second half of the summer monsoon season, when primary productivity has a secondary maximum. Circulation changes, and therefore productivity changes, should be anticipated if human activities modify the intensity and seasonality of the Mekong River plume., Comment: Submitted to JGR-Oceans

The predictability of fields at the ocean surface in the northern Gulf of Mexico (GoM) is investigated through five ensembles of regional ocean simulations between 2014 and 2016. The ensembles explore two horizontal resolutions and different representations of the riverine inflow, and focus on the Loop Current system (LCS) and the Mississippi-Atchafalaya River System (MARS) interactions. The predictability of the surface fields is high in the northern GoM if the atmospheric forcing and the flow at Yucatan Channel are known, and the ensembles simulate similar LCS behavior up to 5 months. In terms of LCS-MARS interactions, the ensembles confirm that they are strongly modulated by the LC mesoscale variability. The relationship is two-ways, with the LCS being influenced by - and not only influencing - the freshwater plume. Whenever the freshwater flux is strong, the northward extension of the LCS is constrained. The ensemble simulations also indicate that this influence is stronger if the riverine inflow is simulated in an active fashion with a meridional velocity component proportional to the flux. Sea surface temperature (SST) and salinity (SSS) predictability have opposite seasonality in their signal, with the SST (SSS) field being more predictable in summer (winter). Partially resolving submesoscale instabilities and improving the accuracy of the riverine fluxes' representation causes the spread to increase, especially in SST. Finally, the predictability of surface relative vorticity decreases in amplitude when increasing resolution due to feedbacks between the mesoscale and submesoscale circulations, but retains most of its intraseasonal and interannual signal., Comment: 24 pages, 13 figures, submitted to JGR: oceans

The threat of global warming and the demand for reliable climate predictions pose a formidable challenge being the climate system multiscale, high-dimensional and nonlinear. Spatiotemporal recurrences of the system hint to the presence of a low-dimensional manifold containing the high-dimensional climate trajectory that could make the problem more tractable. Here we argue that reproducing the geometrical and topological properties of the low-dimensional attractor should be a key target for models used in climate projections. In doing so, we propose a general data-driven framework to characterize the climate attractor and showcase it in the tropical Pacific ocean using a reanalysis as observational proxy and two state-of-the-art models. The analysis spans four variables simultaneously over the periods 1979-2019 and 2060-2100. At each time t, the system can be uniquely described by a state space vector parameterized by N variables and their spatial variability. The dynamics is confined on a manifold with dimension lower than the full state space that we characterize through manifold learning algorithms, both linear and nonlinear. The local geometry and local stability of the high-dimensional, multi-variable climate attractor are quantified through the local dimension and persistence metrics. Model biases that hamper climate predictability are identified and found to be similar in the multivariate attractor of the two models during the historical period while diverging under the warming scenario considered. Finally, the relationships between different sub-spaces (univariate fields), and therefore among climate variables, are evaluated. The proposed framework provides a comprehensive, physically based, test for assessing climate feedbacks and opens new avenues for improving their model representation.

We propose {\delta}-MAPS, a method that analyzes spatio-temporal data to first identify the distinct spatial components of the underlying system, referred to as "domains", and second to infer the connections between them. A domain is a spatially contiguous region of highly correlated temporal activity. The core of a domain is a point or subregion at which a metric of local homogeneity is maximum across the entire domain. We compute a domain as the maximum-sized set of spatially contiguous cells that include the detected core and satisfy a homogeneity constraint, expressed in terms of the average pairwise cross-correlation across all cells in the domain. Domains may be spatially overlapping. Different domains may have correlated activity, potentially at a lag, because of direct or indirect interactions. The proposed edge inference method examines the statistical significance of each lagged cross-correlation between two domains, infers a range of lag values for each edge, and assigns a weight to each edge based on the covariance of the two domains. We illustrate the application of {\delta}-MAPS on data from two domains: climate science and neuroscience., Comment: 12 pages, 6 figures

The processes and instabilities occurring at the ocean surface in the northern Gulf of Mexico are investigated with a regional model at submesoscale-permitting horizontal grid resolution (i.e., HR with dx = 1.6 km) over a three-year period, from January 2010 to December 2012. A mesoscale-resolving, lower resolution run (LR, with dx = 5 km) is also considered for comparison. The HR run is obtained through two-way nesting within the LR run. In HR quantities such local Rossby number, horizontal divergence, vertical velocity, and strain rate are amplified in winter, when the mixed layer is deepest, as found in other basins. In the model configuration considered this amplification occurs in surface waters over the continental slope and off-shore but not over the shelf. Submesoscale structures consist of a mixture of fronts and eddies generated by frontogenesis and mixed layer instabilities, with elevated conversion rates of available potential energy (APE) into eddy kinetic energy (EKE). In all quantities a secondary maximum emerges during the summer season, when the mixed layer depth is shallowest, barely 15-20 m. The secondary peak extends to the coast and is due to the intense lateral density gradients created by the fresh water inflow from the Mississippi River system. Submesoscale structures in summer consist predominately of fronts, as observed in the aftermath of the 2010 Deepwater Horizon oil spill, and their secondary circulations are impeded due to the limited depth of the mixed layer. Freshwater river input is key to the submesoscale activity in summer but modulates it also in winter, as shown with a sensitivity run in which the riverine inflow is absent. Implications for transport studies in regions characterized by intense freshwater fluxes and for submesoscale parameterizations are discussed.

This study examines the mesoscale and submesoscale circulations along the continental slope in the northern Gulf of Mexico at depths greater than 1000 m. The investigation is performed using a regional model run at two horizontal grid resolutions, 5 km and 1.6 km, over a three year period from January 2010 to December 2012. Ageostrophic submesoscale eddies, and vorticity filaments populate the continental slope, and they are stronger and more abundant in the simulation at higher resolution, as to be expected. They are formed from horizontal shear layers at the edges of highly intermittent, bottom-intensified, along-slope boundary currents and in the cores of those currents when they are confined on steep slopes. The horizontal resolution influences the representation of the lateral and vertical transport of neutrally buoyant tracers. This is exemplified by a passive dye deployed near the Macondo Prospect, where the Deepwater Horizon rig exploded in 2010, and at the largest known natural hydrocarbon seep in the northern Gulf, few hundred kilometers to the west of the rig wellhead. The first site is characterized by weak mean currents and, in the high resolution run, by intense submesoscale eddies that form to the west of the Macondo Prospect. GC600, on the other hand, is surrounded by a complicated bathymetry, and its flow is highly variable. The trapping exerted by the submesoscale eddies and vorticity filaments near the rig constrain the spreading in the vicinity of the Macondo Prospect, and as a result, patchiness in the passive tracer distributions differs in the two integrations, despite similar diffusivities expressing the average spreading rate. The tracer released at GC600, on the other hand, is dispersed more effectively in the first six months in the higher resolution case due to enhanced near-bottom turbulent mixing.