Pan-Arctic melt pond fraction trend, variability, and contribution to sea ice changes

Abstract

Because the Pan-Arctic melt pond significantly influences the mass balance of sea ice, analyzing changes in the Pan-Arctic melt pond can offer insights into the Arctic sea ice and environment. In this study, the MODIS level 2 surface reflectance product (MOD09A1) was used to calculate variability and trends in the Pan-Arctic melt pond fraction (MPF) from May 9 to August 29 during the period 2000–2019, along with its relationship with September sea ice extent (SIE). The results showed that as a result of advances in Arctic ice melt onset and the extension of the melting period, the MPF as a whole showed an insignificant upward trend but that the MPF in the multiyear ice area increased significantly. Melt pond distribution is closely related to atmospheric circulation over the Arctic. Significant positive and negative geopotential height anomalies at different locations in the Arctic will cause poleward winds from different directions, bringing warm, humid air and leading to an increase in MPF in different regions of the Arctic. There is a strong negative correlation between integrated MPF and September SIE, concentrated in the northern part of the Beaufort Sea. On June 18, the integrated MPF had the strongest correlation with September SIE, at −0.96. The use of linear regression and support vector machine combined with integrated MPF and 850-hpa average wind, Arctic Oscillation (AO) and North Atlantic Oscillation (NAO) indexes to forecast September SIE shows that the support vector machine is more suitable for forecasting September SIE. When the forecast factor is only the integrated MPF, the MPF is integrated from May 9 to June 10 has the forecast accuracy is the highest, with a considerable forecast skill. When the average May–June AO is combined with the integrated MPF, the forecast skill will be further improved.