The correlation and regression analysis on aerosol optical depth, ice cover and cloud cover in Greenland Sea

Researches on Arctic aerosol, ice cover and cloud cover have received great attention and it related to the regional even global climate changing. We here study the distributions and the coupling relationships of AOD, cloud cover (CLD) and ice cover (ICE) in the Greenland Sea (20°W-10°E, 70°N-80°N) during 2003-2012. Enhanced statistics methods, such as lag regression method and co-integration analysis method are used for correlation and regression analysis. According to the 10 years satellite data, AOD was high in spring, and low in summer. Generally, AOD was higher down south and lower up north. CLD and AOD mainly had negative correlations and ICE and AOD had positive correlations. According to the lag regression analysis by statistical software EViews, both the peaks of CLD and peaks of ICE were all 1 month earlier than the peak of AOD. The co-integration test suggested that both ICE(-1) and CLD(-1) and AOD were all zero-order integration, and there was no unit root in the residual, so there all had long-run equilibrium relationships. ICE and AOD were stationary series, and the residual had no unit root, they were good coupling. The melting of sea ice and decreasing of cloud cover would all result in the increasing of the AOD content. However, the relationship between AOD and CLD was weaker than the relationship between AOD and ICE, indicating that the aerosol in Arctic mostly came from the sea rather than from the air.