A sea ice concentration estimation algorithm utilizing radiometer and SAR data

Karvonen, Juha

doi:10.5194/tc-8-1639-2014

Cited by 41 publications

(40 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…[12][13][14][15]17,18,22,23]), autocorrelation methods ( [24]), wavelet based features ( [10,25,26]), Gabor wavelet techniques ( [12]), and Markov random fields (MRF) ( [16]). Efforts have been undertaken to improve the classification quality based on single pol SAR data by data fusion with scatterometer data ( [27]) from AMSR-2 or by model data based on previously known ice drift ( [5]), where model output is fused with SAR classification by segmentation/texture analysis. Despite the usefulness and success of such tools, certain major issues in sea ice classification still defy a solution in any of the mentioned approaches.…”

Section: Introductionmentioning

confidence: 99%

Comparing Near Coincident Space Borne C and X Band Fully Polarimetric SAR Data for Arctic Sea Ice Classification

Ressel

Singha

2016

Remote Sensing

View full text Add to dashboard Cite

This work compares the polarimetric backscatter behavior of sea ice in spaceborne X-band and C-band Synthetic Aperture Radar (SAR) imagery. Two spatially and temporally coincident pairs of fully polarimetric acquisitions from the TerraSAR-X/TanDEM-X and RADARSAT-2 satellites are investigated. Proposed supervised classification algorithm consists of two steps: The first step comprises a feature extraction, the results of which are ingested into a neural network classifier in the second step. Based on the common coherency and covariance matrix, we extract a number of features and analyze the relevance and redundancy by means of mutual information for the purpose of sea ice classification. Coherency matrix based features which require an eigendecomposition are found to be either of low relevance or redundant to other covariance matrix based features, which makes coherency matrix based features dispensable for the purpose of sea ice classification. Among the most useful features for classification are matrix invariant based features (Geometric Intensity, Scattering Diversity, Surface Scattering Fraction). This analysis reveals analogous results for all four acquisitions, in both X-band and C-band frequencies. The subsequent classification produces similarly promising results for all four acquisitions. In particular, the overlapping image portions exhibit a reasonable congruence of detected ice types.

show abstract

Section: Introductionmentioning

confidence: 99%

Comparing Near Coincident Space Borne C and X Band Fully Polarimetric SAR Data for Arctic Sea Ice Classification

Ressel

Singha

2016

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…We successfully demonstrated the usage of the MODIS OWSI and SIT charts for the validation of SIC charts based on the SENTINEL-1 SAR and AMSR2 radiometer data and two different algorithms [45,65]. The validation experiment showed the applicability of the two SIC algorithms trained for the Baltic Sea, also for the Barents and Kara Seas.…”

Section: Discussionmentioning

confidence: 88%

“…We also conduct a comparison in open water vs. sea ice mapping between the thickness thresholded SIT chart and the OWSI chart. Finally, we demonstrate the usage of the charts for validation of SENTINEL-1 SAR and AMSR2 radiometer data based SIC products [45,65].…”

Section: Resultsmentioning

confidence: 99%

“…[65]. Here, AMSR2 gradient and polarization ratios alone are first used to estimate SIC with the MLP at the resolution of the AMSR2 data (around 10 km).…”

Section: Sea Ice Concentration Estimation From Sar and Radiometer Datamentioning

confidence: 99%

“…SIC charts were calculated from the SENTINEL-1 SAR and AMSR2 radiometer data, with two different algorithms: A1 is SIC with the MLP having as inputs SAR segment-wise median AMSR2 and SAR features [45] (shortly SAR + AMSR2 SIC), and A2 is SIC first with the MLP having only the AMSR2 polarization and gradient ratios as inputs, and then the SIC estimates are updated using the SAR segmentation [65] (AMSR2 SIC + SAR segm. ); for details see Section 3.10.…”

Section: Validation Of Sar and Microwave Radiometer Based Sea Ice Conmentioning

confidence: 99%

See 2 more Smart Citations

MODIS Sea Ice Thickness and Open Water–Sea Ice Charts over the Barents and Kara Seas for Development and Validation of Sea Ice Products from Microwave Sensor Data

Mäkynen

Karvonen

2017

Remote Sensing

View full text Add to dashboard Cite

Abstract:We have developed algorithms and procedures for calculating daily sea ice thickness (SIT) and open water-sea ice (OWSI) charts, based on the Moderate Resolution Imaging Spectroradiometer (MODIS), ice surface temperature (IST) (night-time only), and reflectance (R) swath data, respectively. The resolution of the SIT chart is 1 km and that of the OWSI chart is 250 m. The charts are targeted to be used in development and validation of sea ice products from microwave sensor data. We improve the original MODIS cloud masks for the IST and R data, with a focus on identifying larger cloud-free areas in the data. The SIT estimation from the MODIS IST swath data follows previous studies. The daily SIT chart is composed from available swath charts by assigning daily median SIT to a pixel. The OWSI classification is simply conducted by a fixed threshold for the MODIS band 1 R. This was based on manually selected R data for various ice types in late winter, early melt, and advanced melt conditions. The composition procedures for the daily SIT and OWSI charts somewhat compensates for errors due to the undetected clouds. The SIT and OWSI charts were compared against manual ice charts by Arctic and Antarctic Research Institute in Russia and by Norwegian Meteorological Institute, respectively, and on average, a good relationship between the charts was found. Pixel-wise comparison of the SIT and OWSI charts showed very good agreement in open water vs. sea ice classification, which gives further confidence on the reliability of our algorithms. We also demonstrate usage of the MODIS OWSI and SIT charts for validation of sea ice concentration charts based on the SENTINEL-1 SAR and AMSR2 radiometer data and two different algorithms.

show abstract

Skill metrics for evaluation and comparison of sea ice models

Dukhovskoy

Ubnoske

Blanchard‐Wrigglesworth

et al. 2015

JGR Oceans

View full text Add to dashboard Cite

Five quantitative methodologies (metrics) that may be used to assess the skill of sea ice models against a control field are analyzed. The methodologies are Absolute Deviation, Root‐Mean‐Square Deviation, Mean Displacement, Hausdorff Distance, and Modified Hausdorff Distance. The methodologies are employed to quantify similarity between spatial distribution of the simulated and control scalar fields providing measures of model performance. To analyze their response to dissimilarities in two‐dimensional fields (contours), the metrics undergo sensitivity tests (scale, rotation, translation, and noise). Furthermore, in order to assess their ability to quantify resemblance of three‐dimensional fields, the metrics are subjected to sensitivity tests where tested fields have continuous random spatial patterns inside the contours. The Modified Hausdorff Distance approach demonstrates the best response to tested differences, with the other methods limited by weak responses to scale and translation. Both Hausdorff Distance and Modified Hausdorff Distance metrics are robust to noise, as opposed to the other methods. The metrics are then employed in realistic cases that validate sea ice concentration fields from numerical models and sea ice mean outlook against control data and observations. The Modified Hausdorff Distance method again exhibits high skill in quantifying similarity between both two‐dimensional (ice contour) and three‐dimensional (ice concentration) sea ice fields. The study demonstrates that the Modified Hausdorff Distance is a mathematically tractable and efficient method for model skill assessment and comparison providing effective and objective evaluation of both two‐dimensional and three‐dimensional sea ice characteristics across data sets.

show abstract

A sea ice concentration estimation algorithm utilizing radiometer and SAR data

Cited by 41 publications

References 32 publications

Comparing Near Coincident Space Borne C and X Band Fully Polarimetric SAR Data for Arctic Sea Ice Classification

Comparing Near Coincident Space Borne C and X Band Fully Polarimetric SAR Data for Arctic Sea Ice Classification

MODIS Sea Ice Thickness and Open Water–Sea Ice Charts over the Barents and Kara Seas for Development and Validation of Sea Ice Products from Microwave Sensor Data

Skill metrics for evaluation and comparison of sea ice models

Contact Info

Product

Resources

About