Effect of a snow cover on microwave backscatter from sea ice

The effect of a snow cover on sea ice upon radar backscatter at microwave frequencies (X- andKu-baud) can be important. The effect of scattering from the snow cover on thesigmadegof first-year ice is shown to be severe (5 cm of dry snow can raisesigmadegby 8 dB at 9 GHz), while that onsigmadegof multiyear ice is shown to be smaller. The low thermal conductivity of snow compared to that of sea ice effectively raises the temperature of the upper surface of the ice, resulting in higher dielectric constants for the ice, thereby modifying the backscatter both from the ice surface and from the scattering volume. The temperature effect of a 10-cm snow cover on 3-m-thick multiyear ice is to lower thesigmadegby only about 0.3 dB for air temperature of-20degC. The effect on 1-m-thick first-year ice is even less. Hence, the volume-scattering effect of snow is more important than the temperature effect. The presence of a wet snow cover can block the volume-scattering contribution of the multiyear ice. The effect of wet snow cover on first-year ice should be smaller than that Of dry, snow, becausesigmadegof wet snow is lower than that of dry snow.     

A semi-empirical sea-spectrum model for scattering coefficient estimation

A semi-empirical sea-spectrum model is proposed to be used in a two-scale radar sea scatter model to obtain estimates of radar backscatter over the frequency bandsLtoKu, the incidence angular range20deg-65deg, the azimuth angular range0deg-180degfrom the wind direction and wind speed range 3.5-30 m/s at 19.5 m above the mean sea level. It is shown that the theoretical estimates obtained are consistent with the existing measurements.     

Short gravity and capillary wave spectra from tower-based radar

Radar backscatter measurements made as part of Project MARSEN in 1979 from the Noordwijk tower off the Dutch coast are used to calculate apparent ripple (capillary and short-gravity wave) spectra by inverting the small-perturbation scattering theory. The measurements were made at 10 and 15 GHz for angles of incidence ranging from20degto70deg; this means that the range of Bragg-resonant spatial wavenumbers covered is from 1.43 to 5.90 cm^-1. Results of coincidentC- andX-band experiments by the Institute Francais du Petrole (IFP) andX-band experiments by a group of Dutch researchers (TNO) are compared with our results and good general agreement is found. Our initial results show a steeper falloff of the spectra with increasing wavenumber than reported previously, particularly at low windspeeds. When the spectra are modified to account for the difference between previous aircraft and tower measurements [1], the observed spectra agree well with the appropriate part of Pierson's wave spectrum as modified by Fung and Lee [2].     

Active microwave measurement from space of sea-surface winds

Radar backscatter measurements from the ocean were made at 13.9 GHz from Skylab. The radar signal increased rapidly with wind speed over the entire range of winds encountered, and for angles of incidence of30degand larger. Signals observed were normalized to a nominal incidence angle (from values withinpm2degof the nominal) and to a nominal upwind observation direction, using a theoretical model that has been verified as approximately true with aircraft experiments. The wind speed was regressed against the resulting scattering coefficientssigma^{0}and the values ofbetain windpropto sigma^{0beta}were obtained for incident angles of1deg , 17deg , 32deg , 43deg,and50deg, and for vertical, horizontal, and cross polarizations. For the three larger angles,betavaries from 0.3 to 0.6. Observations during the summer and winter Skylab missions were treated separately because of possible differences caused by an accident to the antenna between the two sets of observations. The results are in general agreement with the theory [26] in all cases, with the winter and cross-polarized agreement somewhat better than that for summer like-polarized data. The "objective analysis" method used for determining "surface-truth" winds in the Skylab experiment was tested by comparing results obtained at weather ships (using all other ship reports to produce the analysis) with the observations made by the weather ships themselves. In most cases, the variance about the regression line between objective analysis and weather-ship data actually exceeded that about the regression line between objective analysis and backscattcr data!     

Discrimination of different sea ice types from CryoSat-2 satellite data using an Object-based Random Forest (ORF)

Abstract

Sea ice type is one of the most sensitive variables in Arctic sea ice monitoring, and it is important for the retrieval of ice thickness. In this study, we analyzed various waveform features that characterize the echo waveform shape and Sigma0 (i.e., backscatter coefficient) of CryoSat-2 synthetic aperture radar altimeter data over different sea ice types. Arctic and Antarctic Research Institute operational ice charts were input as reference. An object-based random forest (ORF) classification method is proposed with overall classification accuracy of 90.1%. Accuracy of 92.7% was achieved for first-year ice (FYI), which is the domain ice type in the Arctic. Accuracy of 76.7% was achieved at the border of FYI and multiyear ice (MYI), which is better than current state-of-the-art methods. Accuracy of 83.8% was achieved for MYI. Results showed the overall accuracy of the ORF method was increased by ～8% in comparison with other methods, and the classification accuracy at the border of FYI and MYI was increased by ～10.5%. Nevertheless, ORF classification performance might be influenced by the selected waveform features, snow loading, and the ability to distinguish sea ice from leads.     

Aircraft measurements of the microwave scattering signature of the ocean

Microwave scattering signatures of the ocean have been measured over a range of surface wind speeds from 3 m/s to 23.6 m/s using the AAFE RADSCAT scatterometer in an aircraft. Normalized scattering coefficients are presented for vertical and horizontal polarizations as a function of incidence angle (nadir to55deg) and radar azimuth angle (0degto360deg) relative to surface wind direction. For a given radar polarization, incidence angle, and azimuth angle relative to the wind direction, these scattering data exhibit a power law dependence on surface wind speed. The relation of the scattering coefficient to azimuth angle obtained during aircraft circles (antenna conical scans) is anisotropic and suggests that microwave scatterometers can be used to infer both wind speed and direction. These results have been used for the design of the Seasat-A Satellite Scatterometer (SASS) to be flown in 1978 on this first NASA oceanographic satellite.     

HF radar observations of Arctic pack-ice breakup

This paper describes the first reported high-resolution remote measurements of sea-ice velocities during the summer Arctic pack-ice breakup, made with a high-frequency (HF) radar system (CODAR, for Coastal Ocean Dynamics Applications Radar) located on Cross Island, Alaska. Each 36-min observation also gives the positions of the ice edge, the moving ice, and the open water, with an azimuthal and distance resolution of5degand 1.2 km, respectively, to a range of 15 km. The statistical uncertainties in speed are typically 2-4 cm/s. The ice breakup was observed over a two-day period starting with low ice velocity and no open water and ending with ice and current velocities of approximately 40 cm/s. The position of the ice edge is verified by a simultaneous synthetic aperture radar (SAR) image. To compare the ice, current, and wind velocities, a uniform velocity model was fitted to the measurements of radial velocity. The speed of both ice and current under free drift conditions was found to lie between 2 and 5 percent of the wind speed and the direction within20degof the wind direction.     

Radar backscattering from artificially grown sea ice

Fine-resolution X-band backscatter measurements were made at the US Army Cold Region Research and Engineering Laboratory in Jan. 1987. Backscatter data were collected from unmodified smooth ice and snow-covered ice and from ice from which the snow had been removed (slightly roughened ice). The results indicate that vertically polarized returns were consistently higher than horizontally polarized echoes from both the slightly roughened and snow-covered saline ice. A 6.5-cm-thick dry snow layer altered the σ^o of the original smooth-surfaced saline ice only slightly, but introduced a noticeable volume scattering component. It is shown that although substantial agreement exists between the bare first-year ice measurements and commonly used surface-scatter model predictions, a complete model of first-year ice must include a volume-scatter contribution     

An experimental study of a pulse compression scatterometer system

The ideal scatterometer, operating from either an aircraft or a satellite platform, should be capable of making rapid, accurate estimates of the sea backscatter cross sectionsigmaover as wide a range of grazing anglespsias possible. Efficient operation over a large range of grazing angles is desirable because 1)sigmabehavior for90deg geq psi geq 70degyields rms gravity wave slope information and is an indicator of sea state and 2)sigmabehavior for70deg geq psi geq 0degyields data on surface wind magnitude and direction as well as information about the power spectrum of the sea. A "hybrid" estimation procedure has been developed for pulse compression radars which uses both frequency and spatially decorrelated samples ofsigmato provide an unbiased estimate ofsigmahaving minimum variance over the entire range of grazing angles for which radar reception is not noise-limited.     

Polarization effects in seaice signatures

Observations of microwave emissivities of multiyear sea ice showed anomalies at horizontal polarization in the frequency range from 5 to 35 GHz during the Norwegian Remote Sensing Experiment (NORSEX) [1] in September and October 1979. The effect can be explained by layers of solid ice present in the dry snow cover throughout the NORSEX area. A special experiment made on a typical multiyear floe confirms this explanation. Since the results also indicate that at 94 GHz the layers do not affect the radiation, a dual-polarized radiometer in the 90-GHz window is a promising sea-ice sensor.     

Microwave signatures of the sea ice in the east Greenland current

Airborne microwave radiometer measurements at 5, 17, and 34 GHz have been carried out over the East Greenland Current. Sea ice signatures have been established for some of the basic ice types like first-year ice and multiyear ice. Other signatures have been experienced like that of presumably very old arctic ice and signatures associated more with the snow cover on the ice than with the ice itself. During MIZEX-83 measurements of total ice concentration were carried out.     

Bayesian classification of the ice cover of the Arctic seas

A classification of sea ice in the Arctic by age (multiyear; first-year; and first-year deformed ice, nilas, etc.) is developed based on the Bayesian approach using satellite radar data and taking into account regional peculiarities of these types of ice for different sectors of the Arctic. Estimations of a priori probabilities for each ice type, which are required for the use of the Bayesian classification, are obtained by the analysis of ice charts in the Arctic seas developed at the AARI in 2008–2013 using satellite data. A posterior probabilities are estimated visually by an expert. Types of sea ice distinguished by the expert on satellite images make it possible to create sample values of the radar-scattering cross section (RSCS). Examples of the proposed Bayesian classification of ice in the Laptev Sea according to Envisat satellite data are given.     

Fabrication and test results of a conical acoustic lens

This paper treats the fabrication problems and test results of an acoustic lens built in accordance with the acoustic design presented in an accompanying paper. Specific areas covered include the selection of the internal fluid, the window, construction of the case, sound-absorbing lining, and acoustic test results. With a SOAB sound-absorbing lining, the lens model worked well, giving horizontal beamwidths of6degand vertical widths of30deg. There is still a problem of compatibility between the SOAB and the Freon TF lens fluid selected.     

A physical radar cross-section model for a wind-driven sea with swell

A new spectrum model for the ocean surface is proposed. We determine the two unknown parameters in this spectrum by fitting it to radar observations. We find that this spectrum combined with two-scale scattering theory can predict much of the observed dependence of the radar cross section on radar frequency, polarization, angle of incidence, and wind velocity at incidence angles in the0deg-70degrange. The spectrum model is combined with a model for swell to examine the effect of swell on the radar cross section. We find that the effect of swell is significant for low radar frequencies (Lband) and near normal incidence but can be nearly eliminated by using higher frequencies (K_{u}band) and large angles of incidence (approx 50deg).     

A thermometer for oceanographic research with a time constant of about 1 ms and a resolution of less than 0.001 K: Its properties and results of its application on cruises

The author has developed a new thermometer with a time constant of about 0.5 ms as a part of his two-phase alternating current bridge. The thermometer permits the application of a Pt-wire with a low resistance of about 10-50 mOmega. A resolution of less than 0.001 K in the range of40degC can be reached. Properties, experiments, and results from measurements with this fast thermometer are described.     

The physical structures of snow and sea ice in the Arctic section of 150°-180°W during the summer of 2010

The physical structures of snow and sea ice in the Arctic section of 150°-180°W were observed on the basis of snow-pit, ice-core, and drill-hole measurements from late July to late August 2010. Almost all the investigated floes were first-year ice, except for one located north of Alaska, which was probably multi-year ice transported from north of the Canadian Arctic Archipelago during early summer. The snow covers over all the investigated floes were in the melting phase, with temperatures approaching 0℃ and densities of 295-398 kg/m3 . The snow covers can be divided into two to five layers of different textures, with most cases having a top layer of fresh snow, a round-grain layer in the middle, and slush and/or thin icing layers at the bottom. The first-year sea ice contained about 7%-17% granular ice at the top. There was no granular ice in the lower layers. The interior melting and desalination of sea ice introduced strong stratifications of temperature, salinity, density, and gas and brine volume fractions. The sea ice temperature exhibited linear cooling with depth, while the salinity and the density increased linearly with normalized depth from 0.2 to 0.9 and from 0 to 0.65, respectively. The top layer, especially the freeboard layer, had the lowest salinity and density, and consequently the largest gas content and the smallest brine content. Both the salinity and density in the ice basal layer were highly scattered due to large differences in ice porosity among the samples. The bulk average sea ice temperature, salinity, density, and gas and brine volume fractions were-0.8℃, 1.8, 837 kg/m3 , 9.3% and 10.4%, respectively. The snow cover, sea ice bottom, and sea ice interior show evidences of melting during mid-August in the investigated floe located at about 87°N, 175°W.     

The effect of microwave backscatter uncertainty of satellite radar altimeter accuracy

The effect of variations in ocean surface roughness characteristics with upwind/downwind direction, reported by other investigators, is used to compute radar cross section (sigmadeg) and to assess the errors which may arise in present and planned altimeter sensors. Based on an analysis of the rough surface impulse response, the uncertainty between attitude angle andsigmadegasymmetry is found to cause height errors as large as 12 cm, depending on off-nadir angles and sea state. Additionally, the previously reported data in conjunction with computed facet backscatter are found to producesigmadegcharacteristics at large off-nadir angles which are in better agreement with experimental results than those predicted by physical optics Gaussian theory.     

Radar backscatter from mechanically generated transient breakingwaves. I. Angle of incidence dependence and high resolution surfacemorphology

This paper describes the results of an experimental investigation of the microwave backscatter from several laboratory generated transient breaking waves. The breaking waves were generated mechanically in a 35 m×0.7 m×1.14 m deep wave tank, utilizing chirped wave packets spanning the frequency range 0.8-2.0 Hz. Backscatter measurements, were taken by a X/K-band (10.525 GHz, 24.125 GHz) continuous wave Doppler radar at 30°, 45°, and 60° angles of incidence, and at azimuth angles of 0° and 180° relative to the direction of wave propagation. Surface profiles were measured with a high-speed video camera and laser sheet technique. Specular facets were detected by imaging the surface from the perspective of the radar. The maximum radar backscatter occurred in the upwave direction prior to wave breaking, was nearly polarization independent and corresponded to the detection of specular facets on the steepened wave face. This peak radar backscatter was predicted through a finite conductivity corrected physical optics technique over the measured surface wave profiles. Post break backscatter was predicted using a roughness corrected physical optics technique and the small perturbation method, which was found to predict the returns for vertical polarization, but to under predict the horizontal returns     

Radar backscatter from mechanically generated transient breakingwaves. II. Azimuthal and grazing angle dependence

For Pt. I see ibid. vol. 26, pp. 181-200 (2001). This paper describes the results of experimental investigations into the microwave backscatter from mechanically generated transient breaking waves. The investigations were carried out in a 110 m×7.6 m×4 m deep model basin, utilizing chirped wave packets spanning 0.75-1.75 Hz. Backscatter measurements were taken by a K-band continuous wave radar (24.125 GHz) at 40° angle of incidence, and at azimuth angles of 0°, 45°, 90°, 135° and 180° relative to the direction of wave propagation. Grazing measurements were conducted using an X-band (10.525 GHz) FMCW radar at 85° angle of incidence, and azimuth angles of 0° and 180°. Results show that the maximum radar backscatter was obtained in the upwave direction prior to wave breaking and was caused by the specular or near specular presentation of the wave to the radar. After breaking, the backscatter transitioned from a specular or near-specular dominated scattering, primarily seen in the upwave direction, to a small scale roughness dominated scattering, observed at all azimuths. Physical optics solutions were found to correctly predict the backscatter for the specular or near-specular dominated scattering and the small perturbation method was found to accurately model the VV polarization post-break radar backscatter     

Short pulse radar used to measure sea surface wind speed and SWH

A joint airborne measurement program is being pursued by NRL and NASA Wallops Flight Center to determine the extent to which wind speed and sea surface significant wave height (SWH) can be measured quantitatively and remotely with a short pulse (2 ns), wide-beam (60deg), nadir-looking 3-cm radar. The concept involves relative power measurements only and does not need a scanning antenna, doppler filters, or absolute power calibration. The slopes of the leading and trailing edges of the averaged received power for the pulse limited altimeter are used to infer SWH and surface wind speed. The interpretation is based on theoretical models of the effects of SWH on the leading edge shape and rms sea-surface slope on the trailing-edge shape. The models include the radar system parameters of antenna beam width and pulsewidth. Preliminary experimental results look promising and indicate that it may be possible to design a relatively compact airborne radar to infer, in real-time, the sea surface SWH and surface wind speed.