Advanced Algorithm for Continuous Melt Onset Detection on Arctic Sea Ice

Abstract

Expansion of the Arctic melting season with an earlier melt onset date (MOD) is a well-known indicator of Arctic warming. Since 1979, the pan-Arctic MOD distributions usually have been estimated using passive satellite microwave radiometer observations. However, there is a poor agreement in MOD between previous MOD detection algorithms based on passive microwave measurements, raising doubts regarding the accuracy of their MOD products. Thus, this study developed a new MOD algorithm, namely TBmax algorithm, to improve the estimation accuracy of continuous melt onset. The TB(max )algorithm utilizes the microwave radiation characteristics of sea ice, and the daily brightness temperature time series shows their maximum brightness temperature on MOD. By using AMSR2 brightness temperature data, the pan-Arctic MOD distributions estimated from 2013 to 2021 using the TB(max )algorithm successfully reproduced a feature of sea ice melting that mainly during May or June over the Arctic, including the late melting tendency of ice at high latitudes and multiyear ice (MYI). Validation with independent dataset (ice mass balance (IMB) buoy data) suggested that the TBmax MODs showed superior performance compared to other previous algorithms (biases of 0.1 days versus -2.7 and 13.9 days). As MOD can provide information about surface emissivity and the energy budget of the sea ice, the improved MOD may contribute to a more precise analysis of Arctic environment change and enhanced estimation of sea ice parameters.