Peculiarities of stochastic regime of Arctic ice cover time evolution over 1987–2014 from microwave satellite sounding on the basis of NASA team 2 algorithm

• Published in: Izvestiya. Atmospheric and oceanic physics Vol. 51; no. 9; pp. 929 - 934
• Main Authors: Raev, M. D., Sharkov, E. A., Tikhonov, V. V., Repina, I. A., Komarova, N. Yu
• Format: Journal Article
• Language: English
• Published: Moscow Pleiades Publishing 01.12.2015; Springer Nature B.V
• Subjects: Algorithms; Arctic ice; Arctic Ice Cover State Monitoring Based on Satellite Data; Climatology; Confidence intervals; Earth and Environmental Science; Earth Sciences; Exploration; Geophysics/Geodesy; Ice; Ice cover; Ice environments; Marine; Mathematical analysis; Multichannel communication; Polynomials; Regression analysis; Satellite sounding; Satellites; Space research; Statistical analysis; Arctic region; PN, Arctic; nonlinear regression; Arctic ice cover; brightness temperature; microwave remote sensing
• ISSN: 0001-4338; 1555-628X
• DOI: 10.1134/S0001433815090169

The GLOBAL-RT database (DB) is composed of long-term radio heat multichannel observation data received from DMSP F08–F17 satellites; it is permanently supplemented with new data on the Earth’s exploration from the space department of the Space Research Institute, Russian Academy of Sciences. Arctic ice-cover areas for regions higher than 60° N latitude were calculated using the DB polar version and NASA Team 2 algorithm, which is widely used in foreign scientific literature. According to the analysis of variability of Arctic ice cover during 1987–2014, 2 months were selected when the Arctic ice cover was maximal (February) and minimal (September), and the average ice cover area was calculated for these months. Confidence intervals of the average values are in the 95–98% limits. Several approximations are derived for the time dependences of the ice-cover maximum and minimum over the period under study. Regression dependences were calculated for polynomials from the first degree (linear) to sextic. It was ascertained that the minimal root-mean-square error of deviation from the approximated curve sharply decreased for the biquadratic polynomial and then varied insignificantly: from 0.5593 for the polynomial of third degree to 0.4560 for the biquadratic polynomial. Hence, the commonly used strictly linear regression with a negative time gradient for the September Arctic ice cover minimum over 30 years should be considered incorrect.