The sea-ice detection capability of synthetic aperture radar
DOI:
https://doi.org/10.18063/som.v2i2.261Keywords:
synthetic aperture radar, sea ice, dielectric constant, normalized radar cross section (NRCS)Abstract
Climate change, increasing activities in areas like offshore oil and gas exploration, marine transport, eco-tourism,
in additional to the usual activities of northerners resident are leading to reductions in sea ice. Therefore, there is an urgent
need for improvement in the sea ice detection in polar areas. Starting from the mechanism of electromagnetic scattering,
based on an empirical dielectric constant model, we apply EM multi-reflection and transmission formulas for coefficients
between the air-ice interface and sea water-ice interface to develop a model for estimating the capability of detection of sea
ice and ice thickness based on a pulse radar system, synthetic aperture radar (SAR). Although the dielectric constant of sea
ice is less than that of sea water, this model can provide a rational methodology as the normalized radar cross section (NRCS)
of sea ice is larger than that of sea water due to multiple reflections. The numerical simulations of this model showed that
the convergence rate is rapid. With 3 or 4 reflections and transmissions (depending on temperature, salinity, and dielectric
constants of sea ice and water), truncation errors can be satisfied using theoretical considerations and practical applications.
The model is applied to estimate the capability of SAR to discriminate ice from water. The numerical results suggested that
the model ability to measure ice thickness decreases with increasing radar incident angles and increases with increasing
radar pulse width. Reflection and transmission coefficients decrease monotonically with ice thickness and are saturated for
ice thicknesses above a certain critical value which depends on SAR incidence angle, frequency and dielectric constants
of sea ice. The capability to detect ice thickness for given different bands of pulse radar widths can be estimated with this
model.