A full spatial coverage of albedo data is necessary for climate studies and modeling, but clouds and high solar zenith angle cause missing values to the optical satellite products, especially around the polar areas. Therefore, we developed monthly gradient boosting (GB) method based gap filling models. We aim to apply them to the Arctic sea ice area of the 34 years long albedo time series CLARA-A2 SAL (Surface ALbedo from the CLoud, Albedo and surface RAdiation data set) of the Satellite Application Facility on Climate Monitoring (CM SAF) project. GB models are used to fill missing data in albedo 5-day (pentad) means using albedo monthly mean, brightness temperature, and sea ice concentration data as model inputs. Monthly GB models produce the most unbiased, precise, and robust estimates when compared to alternative estimates (monthly mean albedo values directly or estimates from linear regression). The mean relative differences between GB based estimates and original non gapped pentad values vary from -20% to 20% (RMSE being 0.048), compared to relative differences varying from -20% to over 60% (RMSE varying from 0.054 to 0.074) between other estimates and original non gapped pentad values. Also, when comparing estimates from GB models to estimates from linear regression models over three smaller Arctic sea ice areas with varying annual surface albedo cycle (Hudson Bay, Canadian Archipelago and Lincoln Sea), albedo of the melting sea ice is predicted better by the GB models (with negligible mean differences). Gradient boosting is therefore a useful method to fill gaps in the Arctic sea ice area, and the brightness temperature and sea ice concentration data provide useful additional information to the monthly models.