Over the last 25 years, the increase in Arctic air temperature has been twice as high as anywhere else on the planet, resulting in dramatic changes in these areas. The extent and thickness of Arctic sea ice have declined considerably, especially during the summer. Coupled regional / global climate models still fail to replicate this decline: they underestimate the observed reduction in sea ice, showing that physical processes and feedback mechanisms are not yet well represented. The so-called Arctic amplification has repercussions on the low latitudes, but the prediction of these models is still inadequate.
Since the 1970s, passive microwave imagers have been providing an estimate of sea ice concentration (IPCC 2013) and it is one of the longest satellite climate series. These microwave data between 6 and 37 GHz are not or hardly affected by clouds and do not depend on sunlight. However, the quality of the sea ice concentration estimates is still limited, mainly because of the spatial resolution of the instruments. Microwave high frequencies (> 30 GHz) already allow good spatial resolution, but are affected by atmospheric disturbances and they are less sensitive to the ice signal. Microwave low frequencies (<11 GHz) allow a better accuracy of the estimate, but with a degraded spatial resolution. The algorithms for sea ice characterization are not currently optimized to jointly exploit the radiometric sensitivity of the low frequencies and the spatial resolution of the high frequencies.
A new mission is currently under study for the next generations of satellite missions (Copernicus Imager Microwave Radiometer, CIMR), with frequencies between 1.4 and 37 GHz. This is an opportunity to develop new characterization algorithms for sea ice, which will benefit these new instruments but also existing observations (e.g., AMSR-E and AMSR-2).
At first, a fine analysis of the existing algorithms will be conducted. Special attention will be given to databases that contain existing satellite observations (AMSR) coinciding with in situ estimates of sea ice concentration. These databases provide direct access to the sensitivity of observations to the characteristics of the sea ice. They are already used in algorithm calibration, but not always optimally. We propose here to revisit the use of these databases and to develop algorithms that take advantage of this rich information.
In a second step, the question of the spatial resolution of the observations will be examined. It involves combining satellite measurements of different resolutions, taking advantage of the sensitivity of low frequencies, without losing the benefit of the spatial resolution of high frequencies. This is a general problem in remote sensing and it deserves special attention. We will take great care to quantify the errors resulting from the new algorithms and to compare them to the existing ones.
Finally, the developed methods will be applied to the AMSR data over the last 15 years and compared to the sea ice concentration estimates already available. This work will be done in close relation with the groups currently producing these data (for example, Bremen University, https://seaice.uni-bremen.de/sea-ice-concentration/) and with the users (for example IFREMER).
Strong knowledge in physics, remote-sensing Earth sciencs. Good knowlegde of statistical data analysis. Knowledge of at least one programming support.
To apply, we invite you to contact the PhD/research supervisor and fill, with him/her, the co-financing part of the online application form (Reply to the offer) by April 1st, 2019.