Fluxes and flows

Abstract

The launch of NASA’s SEASAT satellite in 1978 ushered in the era of radar oceanography. So treasured were the insights provided by this mission that its untimely demise, only three months after launch, did not preclude a legacy of success. In fact, rumors of sabotage sprang up, claiming that the instrument’s remote sensing prowess had jeopardized military assets. Scuttling SEASAT had surely been a matter of national security.

Like plenty of rumors, these, too, were grounded in some truth: SEASAT had shed light on many a mystery, though primarily in the field of physical sciences rather than Cold War conspiracies. Nonetheless, the value of spaceborne oceanography was now clear, and space agencies around the world did not stand idly by. Europe quickly followed suit with its own missions, launching in succession a series of high-resolution imagers, including ERS-1, ERS-2, ENVISAT, and, since 2014, the Sentinel-1 constellation.

Aided by these instruments and many others, the previous decades have been marked by continued scientific advances in the ever-growing field of radar oceanography. Building on that rich heritage, this dissertation aims to advance two distinct (but entwined) subdomains relating to either side of the air-sea interface.

In the first half of this dissertation, Chapters 2 and 3, we investigate the ubiquitous yet often overlooked atmospheric features present in most high-resolution radar images of the ocean. It may seem counterintuitive to study the atmosphere using ocean imagery, but much of the ocean’s surface variability is wind-driven, such that the ocean surface acts as a mirror for the wind field aloft. Till recently, the only systematic use for these atmospheric features had been in studies aiming to relate (an)isotropy to wind directions. But the information content is rich, and much more can be done.

In Chapter 2, we show that significant quantities of atmospheric information remain unexploited. And by using a domain-inspired encoding scheme, we transform high-dimensional radar images into a much lower-dimensional latent space, from which we extract substantially more atmospheric information than preexisting analytical methods.

Then in Chapter 3 we leverage the ever-increasing catalog of Sentinel-1 observations to go one step further. We illustrate that radar-derived atmospheric variables are complementary to state-of-the-art and reference atmospheric estimates. The challenge here is one of persuasion: how to argue that new estimates can complement state-of-the-art estimates when there are insufficient “ground truth” measurements to reconcile discrepancies between them? We aim to address this difficulty through a series of qualitative and quantitative assessments.

In the second half of the dissertation, Chapters 4 and 5, we switch to the other side of the air-sea interface, to investigate different methods of retrieving surface currents—a long-standing challenge in spaceborne oceanography. First, we investigate surface current retrieval for a proposed satellite modification, DopSCA, in Chapter 4. This system could be uniquely capable of retrieving surface current information, but its unorthodox design makes it susceptible to several additional noise sources. One of these noise sources, which we refer to as leakage—the undesired interaction between geometric Doppler, azimuthal backscatter gradients, and the antenna pattern—is investigated to assess its mission-compromising potential. Fortunately, this seems to be a comparatively minor issue.

Then, using a very different instrument in Chapter 5, we consider an alternative surface-current retrieval that incorporates polarimetry. Here we rely on the cross-correlation between wave motion, its radar cross section, and polarization to disentangle the convolved mixture of signals received by the satellite. Simulations and limited real observations illustrate that this approach may be a fruitful avenue towards the long-standing challenge of distinguishing between wave- and current motions. Future systems with additional polarimetric capabilities, such as Harmony or Sentinel-1 Next Generation, stand to gain from this approach in particular.

To summarize, this dissertation investigates radar-based retrieval of heat and momentum transport (from atmospheric convection and surface currents, respectively) by exploiting unique observational properties of a variety of radar satellites. We’ve made progress in both domains: providing the first quantitative and SAR-retrieved air-sea heat-flux estimates, and introducing a new method for surface-current retrieval. But fear not, enough work remains to entertain many radar oceanographers to come ;)