32nd URSI GASS, Montreal, 19–26 August 2017 An Experimental Study of Microwave Remote Sensing of Oil-Contaminated Young Sea Ice Nariman Firoozy (1) , Thomas Neusitzer (2) , Durell Desmond (1) , Tyler Tiede (2) , Marcos Lemes (1) , Jack Landy (3) , Gary Stern (1) , Puyan Mojabi (2) , Søren Rysgaard (1) , and David G. Barber (1) (1) Centre for Earth Observation Science, University of Manitoba, Winnipeg, Canada (2) Department of Electrical and Computer Engineering, University of Manitoba, Winnipeg, Canada (3) School of Geographical Sciences, University of Bristol, Bristol, United Kingdom Abstract This paper presents an experiment on remote sensing of oil infested sea ice, and the detection of this contaminant. To this end, an overview of our previously developed electro- magnetic inversion algorithm is first presented. This algo- rithm has been able to reconstruct the complex permittiv- ity profile of snow-covered sea ice, and also retrieve some of its thermodynamic and geophysical properties. Next, a description of our oil-in-sea ice experiment is presented in which crude oil is injected underneath an artificially-grown young sea ice as the resulting radar cross section response is temporally measured. The volume fraction of oil is then in- directly retrieved using the measured radar data via a mod- ified inversion strategy. Although the reconstructed volume fraction is an over-estimation, it has a potential to trigger a warning system. Finally, the reasons behind this over- estimation are discussed. 1 Introduction Arctic sea ice extent has been shrinking in the past few decades, and a summertime ice-free Arctic is predicted to happen before the mid-century, with an estimated two decades uncertainty [1, 2]. Considering the Arctic’s harsh environment, sheer size, and remoteness, microwave re- mote sensing has been the primary choice in observation and quantification of Arctic changes due to its ability in covering large areas during day and night, and being mini- mally affected by weather conditions. This technology can also be utilized for monitoring and parameter retrieval in areas of increased activities in the Arctic, for instance, the revitalized Northwest Passage shipping route [3]. Remote sensing retrieval models are generally either statis- tical (i.e., developed based on previous measurements and their associated observed parameters), or physical-based (i.e., reliant on the simulation of the system’s response). This paper will concentrate on the latter. In particular, we will present our inversion methodology which is based on minimizing the discrepancy between the simulated and measured normalized radar cross section data in the mi- crowave regime. Through such an inversion technique, the complex permittivity profile of the domain of interest can be directly reconstructed by minimizing an appropri- ate data misfit cost function [4]. In addition, we can in- directly retrieve the profile’s thermodynamics, geophysics, or inclusion properties [5, 6]. However, for such an in- version algorithm to be successful, some challenges exist at both data collection and processing levels, e.g., collec- tion of ‘sufficient’ amount of data (say, multi-view and/or multi-frequency) which is not trivial in sea ice remote sens- ing, and handling of the ill-posedness associated with the electromagnetic inverse scattering problem [7]. One can at- tempt to alleviate these issues by various means including the use of prior information, and appropriate regularization techniques [8]. This paper will investigate the possibility of oil detection in a sea ice environment. It should be noted that our fo- cus is on the detection of spills rather than reservoirs (e.g., see [9]). Previously, ground penetrating radar (e.g., based on reflection-waveform inversion [10]), and SAR imagery (e.g., based on polarization-ratio [11]) have been utilized for oil detection in ice-covered areas amongst other radar technologies [12, 13]. In this paper, we will introduce a physical-based retrieval model that utilizes radar cross sec- tion data. To this end, we initially present an overview on the general scheme of our inversion algorithm as applied to various snow-covered sea ice profiles (in the absence of oil spills) in Section 2. Next, we describe our oil-in-sea ice experiment, and present some collected data in Section 3. Herein, the primarily results are presented. In addition, it will be briefly discussed why our inversion strategy needs to be modified for this experiment. Further details pertaining this modified algorithm, and interpretation of the retrieved data will be explained in our presentation. 2 Electromagnetic Inversion In the first part of our paper, we present an overview of our electromagnetic inversion algorithm developed for snow- covered sea ice remote sensing. Our inversion algorithm consists of an inverse solver and a forward solver. To find the parameters of interest, the inverse solver iteratively min- imizes a cost function that is based on a discrepancy be- tween the measured and the simulated electromagnetic re- sponse of the profile. To this end, we utilize the normalized