Arctic sea ice concentrations from special sensor microwave imager and advanced very high resolution radiometer satellite data

Abstract

Nearly coincident data from the special sensor microwave imager (SSM/I) and the advanced very high resolution radiometer (AVHRR) are used to compute and compare Arctic sea ice concentrations for different regions and times of the year. To help determine overall accuracies and to highlight sources of differences between passive microwave, optical wavelength, and thermal wavelength data, ice concentrations are estimated using two operational SSM/I ice concentration algorithms and with visible‐ and thermal‐infrared wavelength AVHRR data. All algorithms capture the seasonal patterns of ice growth and melt. The ranges of differences fall within the general levels of uncertainty expected for each method and are similar to previous accuracy estimates. The estimated ice concentrations are all highly correlated, with uniform biases, although differences between individual pairs of observations can be large. On average, the NASA Team algorithm yielded 5% higher ice concentrations than the Bootstrap algorithm, while during nonmelt periods the two SSM/I algorithms agree to within 0.5%. These seasonal differences are consistent with the ways that the 19‐GHz and 37‐GHz microwave channels are used in the algorithms. When compared to the AVHRR‐derived ice concentrations, the Team‐algorithm results are more similar on average in terms of correlation and mean differences. However, the Team algorithm underestimates concentrations relative to the AVHRR output by 6% during cold months and overestimates by 3% during summer. Little seasonal difference exists between the Bootstrap and AVHRR results, with a mean difference of about 5%. Although the mean differences are less between the SSM/I‐derived concentrations and concentrations estimated using AVHRR channel 1, the correlations appear substantially better between the SSM/I data and concentrations derived from AVHRR channel 4, particularly for the Team algorithm output.

The Atmospheric Impact of Uncertainties in Recent Arctic Sea Ice Reconstructions

“…For example, in the Beaufort and Chukchi Seas, the difference between NASA team and Landsat was 0.6% Ϯ 7.4% in autumn, Ϫ2.1% Ϯ 3.1% in spring, and 11.0% Ϯ 22.9% in summer. However, another comparison of SSM/I NASA team data to AVHRR calculated a 6% difference, which reduced to only 3% in summer (Emery et al 1994), while Bootstrap data differed by 5% in both seasons (however, Bootstrap tie points have since been adjusted). Comiso et al (1997) extended the results of Steffen and Schweiger (1991).…”

The use of operational ice charts for evaluating passive microwave ice concentration data

“…The average absolute differences were 8.2% with a standard deviation of 8.8%. Emery et al (1994) compared the NT algorithm with AVHRR data and found mean differences of 5.7% with 11.2% rootmean square error and 8% standard deviation during nonsummer months. Steffen and Schweiger (1991) compared 28 Landsat images and found underestimation of as much as 9% and 11% in summer using global tie points in areas of nilas and new ice.…”

A Comparison of Two 85-GHz SSM/I Ice Concentration Algorithms With AVHRR and ERS-2 SAR Imagery

“…Consequently, the IR temperature contrast between open water and ice is large. Here, AVHRR data acquired in April 1999 are used together with a tie point method based on AVHRR channel 4 IR temperatures [28], [30]. Tie points (see Table IV) are estimated as follows.…”