Rasmus Tonboe

Ocean Winds from RADARSAT-1 ScanSAR

This paper discusses an algorithm designed to retrieve high-resolution wind fields from scanning synthetic aperture radar (ScanSAR) data acquired on board the Canadian satellite RADARSAT-1. The ScanSAR operates at C-band with horizontal polarization. The wind directions are extracted from wind-induced streaks, e.g., from atmospheric boundary layer rolls or wind shadowing, which are approximately in line with the mean wind direction near the ocean surface. The wind speeds are derived from the normalized radar cross section (NRCS) and image geometry of the calibrated ScanSAR images, together with the local wind direction retrieved from the image. Therefore the semi-empirical C-band model CMOD4, which describes the dependency of the NRCS on wind and image geometry, is used. The CMOD4 was originally developed for the scatterometer of the European remote sensing satellites ERS-l and ERS-2 operating at C-band with vertical polarization. Consequently, the CMOD4 required modification for horizontal polarization, which is performed by considering the polarization ratio. To demonstrate the applicability of the algorithm, wind fields were computed from 20 RADARSAT-1 ScanSAR wide-swath images and compared to co-located results from the Danish high-resolution limited-area model (HIRLAM). In addition, the error sources in ScanSAR wind retrieval are discussed and sensitivity studies were carried out to estimate wind speed errors due to uncertainties in the NRCS, wind direction, and incidence angles.

Intercomparison of passive microwave sea ice concentration retrievals over the high-concentration Arctic sea ice

[1] Measurements of sea ice concentration from the Special Sensor Microwave Imager (SSM/I) using seven different algorithms are compared to ship observations, sea ice divergence estimates from the Radarsat Geophysical Processor System, and ice and water surface type classification of 59 wide-swath synthetic aperture radar (SAR) scenes. The analysis is confined to the high-concentration Arctic sea ice, where the ice cover is near 100%. During winter the results indicate that the variability of the SSM/I concentration estimates is larger than the true variability of ice concentration. Results from a trusted subset of the SAR scenes across the central Arctic allow the separation of the ice concentration uncertainty due to emissivity variations and sensor noise from other error sources during the winter of 2003–2004. Depending on the algorithm, error standard deviations from 2.5 to 5.0% are found with sensor noise between 1.3 and 1.8%. This is in accord with variability estimated from analysis of SSM/I time series. Algorithms, which primarily use 85 GHz information, consistently give the best agreement with both SAR ice concentrations and ship observations. Although the 85 GHz information is more sensitive to atmospheric influences, it was found that the atmospheric contribution is secondary to the influence of the surface emissivity variability. Analysis of the entire SSM/I time series shows that there are significant differences in trend between sea ice extent and area, using different algorithms. This indicates that long-term trends in surface and atmospheric properties, unrelated to sea ice concentration, influence the computed trends.

Automatic detection and validity of the sea-ice edge: an application of enhanced-resolution QuikScat/SeaWinds data

Sea-ice edge detection is an essential task at the different national ice services to secure navigation in ice-covered seas. Comparison between the Remund and Long ice mask image from enhanced-resolution QuikScat/SeaWinds (QS) products and the analyzed ice edge from high-resolution RADARSAT synthetic aperture radar has shown that the automatically determined QS ice mask underestimates the Arctic ice extent. QS data was statistically analyzed by colocating the data with ice charts around Greenland and with the National Aeronautics and Space Administration Teams Special Sensor Microwave/Imager (SSM/I) ice concentration algorithm over the whole Arctic region. All variables, i.e., the backscatter in vertical and horizontal polarization, the active polarization ratio (APR) and the daily standard deviation, are sensitive to ice types and are strongly correlated with ice concentration when the relationship is expressed in exponential form. Our study showed that the APR is especially suitable for ice-ocean separation, and a threshold of -0.02 was determined. An ice edge algorithm based on this APR threshold was developed using the other variables with conservative season-dependent thresholds to eliminate additional ocean noise. Also, the history of the ice cover is considered in order to detect single ice fields that are separated from the main Arctic pack ice. Validation with RADARSAT 1 and with the Advanced Very High Resolution Radiometer showed that the new algorithm successfully detects very low ice concentrations of about 10% during the entire year. The validity of the detected ice edge for near-real-time issues is also discussed in relation to the ice motion in the Marginal Ice Zone and the integration time necessary to produce the enhanced-resolution images. The new algorithm improves the automatic global ice edge resolution by a factor of two when compared to SSM/I products and could be used in both model initialization and data assimilation.

Simulations of the snow covered sea ice surface temperature and microwave effective temperature

The snow surface on thick multiyear sea ice in winter is on average colder than the air because of the negative radiation balance. Beneath the snow surface there is a strong temperature gradient in winter with increasing temperatures towards the ice—water interface temperature at the freezing point around –1.8 ◦C. The sea ice surface temperature and the thermal microwave brightness temperature were simulated using a combination of thermodynamic and microwave emission models. The simulations indicate that the physical snow—ice interface temperature or alternatively the 6 GHz effective temperature have a good correlation with the effective temperature at the temperature sounding channels near 50 GHz. The complete correlation matrix based on the simulations for physical and effective temperatures is given. The physical snow—ice interface temperature is related to the brightness temperature at 6 GHz vertical polarization as expected. However, the emissivity factor normally used when converting brightness temperature to the ice temperature is dependent on the ice temperature. The simulations indicate that a simple model may be used to derive the snow-ice interface temperature from satellite AMSR 6 GHz measurements.

Snow thickness retrieval from L-band brightness temperatures: a model comparison

Abstract The Soil Moisture and Ocean Salinity (SMOS) satellite’s L-band (1.4 GHz) measurements have been used to retrieve Snow thickness over thick sea Ice in a previous study. Here we consider brightness temperature simulations for 2.5–4.5m thick Arctic multi-year Ice and compare the results of the relatively simple emission model (M2013) used previously for the retrieval with simulations from a more complex model (T2011) that combines a sea-Ice version of the Microwave Emission Model for Layered Snowpacks (MEMLS) with a thermodynamic model. We find that L-band brightness temperature is mainly determined by Ice temperature. In the M2013 model, Ice temperature in turn is mainly determined by surface temperature and Snow thickness, and this dependence has been used previously to explain the potential for a Snow thickness retrieval. Our comparisons suggest that the M2013 retrieval model may benefit from a more sophisticated thermodynamic calculation of the Ice temperature or from using independent temperature data (e.g. from 6 GHz channels). In both models, horizontally polarized brightness temperatures increase with Snow thickness while holding surface temperature, Ice thickness and Snow density near constant. The increase in the T2011 model is steeper than in M2013, suggesting a higher sensitivity to Snow thickness than found earlier.

Simulation of the Ku-band Radar altimeter sea ice effective scattering surface

A radiative transfer model is used to simulate the sea ice radar altimeter effective scattering surface variability as a function of snow depth and density. Under dry snow conditions without layering these are the primary snow parameters affecting the scattering surface variability. The model is initialized with in situ data collected during the May 2004 GreenIce ice camp in the Lincoln Sea (73/spl deg/W; 85/spl deg/N). Our results show that the snow cover is important for the effective scattering surface depth in sea ice and thus for the range measurement, ice freeboard, and ice thickness estimation.

Analysis of WindSat Third and Fourth Stokes Components Over Arctic Sea Ice

WindSat has provided an opportunity to investigate the first spaceborne passive fully polarimetric observations of the Earths surface. In this paper, Arctic sea ice was investigated. The passive polarimetric data are provided in the form of the modified Stokes vector consisting of four parameters. The first two components of the modified Stokes vector are the vertically and horizontally polarized brightness temperatures, which have been continuously measured by various radiometers over the last three decades. The third and fourth Stokes components provide in formation on the degree of polarization of the emission. In this paper, three types of analysis are carried out: spatial (maps considering different azimuth angle intervals), temporal (time series of daily averaged Stokes components over a small selected azimuth angle range), and azimuthal (variations w.r.t. the azimuth angle over selected study areas). Analysis has shown the highest brightness temperature variations for the 37-GHz third Stokes component (>; 2 K) during summer. The next highest signals were observed for the 10.7-GHz third and fourth Stokes components (>; 1 K) during summer as well. The 37-GHz fourth Stokes component exhibited the least variability (>; 1 K). Spikes of up to 2 K were identified in the time series of the 37-GHz third Stokes component during mid-January 2004 (winter) over first-year ice regions. The near-surface air temperature of the European Center for Medium-Range Weather Forecasts model data and the Special Sensor Microwave/Imager National Aeronautics and Space Administration Team ice concentrations revealed that, during these events, the surface temperatures reached near melting levels and the retrieved ice concentrations were reduced to about 80%. Moreover, these observations also showed clear evidence of first harmonic azimuthal dependence. Geophysical parameters, such as temperature and ice leads, are likely to be the causes. The larger signals which occurred during summer were identified as being related to the ice surface temperatures being near melting.

Simulation of the CryoSat-2 satellite radar altimeter sea ice thickness retrieval uncertainty

Although it is well known that radar waves penetrate into snow and sea ice, the exact mechanisms for radar altimeter scattering and its link to the depth of the effective scattering surface from sea ice are not well known. Previously proposed mechanisms linked the snow-ice interface, i.e., the dominating scattering horizon, directly with the depth of the effective scattering surface. However, simulations using a multilayer radar scattering model show that the effective scattering surface is affected by snow-cover and ice properties. With the coming CryoSat-2 (planned launch in 2010) satellite radar altimeter, it is proposed that sea ice thickness can be derived during winter by measuring its freeboard. In this study we evaluate the radar altimeter sea ice thickness retrieval uncertainty in terms of floe buoyancy, radar penetration, and ice type distribution using both a scattering model and Archimedes’ principle. The effect of the snow cover on the floe buoyancy and radar penetration and on the ice cover spatial and temporal variability is assessed from field campaign measurements in the Arctic resulting in ice thickness uncertainties of about 0.3 m for the snow depth variability and 0.3 m for the snow density variability. In addition to these well-known uncertainties, we use high-resolution RADARSAT synthetic aperture radar (SAR) data to simulate errors due to the variability of the effective scattering surface as a result of the subfootprint spatial backscatter and elevation distribution, sometimes called preferential sampling. In particular, in areas where ridges represent a significant part of the ice volume (e.g., the Lincoln Sea), the average simulated altimeter thickness estimate of 2.68 m is lower than the real average footprint thickness of 3.85 m, making preferential sampling the single most important error source. This means that the errors are large and yet manageable if the relevant quantities are known a priori. Radar altimeter ice thickness retrieval uncertainties are discussed.

Coastal high resolution wind fields retrieved from RADARSAT-1 ScanSAR

An algorithm is presented for retrieving high resolution wind fields from the scanning synthetic aperture radar (ScanSAR) data acquired aboard the Canadian satellite RADARSAT-1. The wind directions are extracted from wind induced streaks, e.g. from boundary layer rolls, Langmuir cells, or wind shadowing, which are approximately in line with the mean wind direction. The wind speeds are derived from the normalized radar cross section (NRCS) and image geometry of the calibrated ScanSAR images, together with the local wind direction. Therefore the semi-empirical C-band model CMOD4, which describes the dependency of the NRCS on wind and image geometry, is inverted. CMOD4 was originally developed for the scatterometer operating at C-band with vertical polarization and has to be modified for horizontal polarization, which is performed by considering the polarization ratio according to Kirchhoff scattering. To verify the algorithm, wind fields were computed from several RADARSAT-1 ScanSAR images and compared to collocated results from the Danish high resolution limited area model. To estimate relative errors of wind speed due to uncertainties in wind direction and NRCS, sensitivity studies were performed.

Analysis of WindSat Data over Arctic Sea Ice

The radiation of the 3rd and 4th Stokes components emitted by Arctic sea ice and observed by the spaceborne fully polarimetric radiometer WindSat is investigated. Two types of analysis are carried out, spatial (maps of different quadrants of azimuth look angles) and temporal (time series of daily averages over small selected ranges of azimuth angles). The 3rd Stokes component at 37 GHz has shown the highest signal during early summer (Gt2 K). The next highest signals were observed at the 10.7 GHz 3rd and 4th Stokes components during the summer months (Gt1 K). The 37 GHz 4th Stokes component has shown the least variability (Lt1 K). The 10.7 GHz 4th Stokes component has higher signal at the ice edge, confirmed from the sea ice concentration maps derived from Advanced Microwave Scanning Radiometer (AMSR-E) data, and similar to signals observed over global coastlines. From the comparison with near surface air temperature of the European Center for Medium-Range Weather Forecasts (ECMWF) model data and Scanning Multichannel Microwave/Imager (SSM/I) NASA Team ice concentrations it was concluded that the microphysical processes during early melting in the topmost sea ice layers are responsible for the high 37 GHz 3rd Stokes signal.