Performances of atmospheric density models during satellite reentry prediction campaigns at sunspot minimum

In order to assess the performances of air density models on the accuracy of satellite reentry predictions in condition of low solar activity, five different thermospheric density models were used, under the same conditions and assumptions, during the reentry predictions campaigns for Cosmos 1025 (2007) and EAS (2008). Three of the models (JR-71, MSISE-90 and NRLMSISE-00) were common to both campaigns, while MSIS-86 was used only for Cosmos 1025 and JB2006 only for EAS. Many reentry predictions were carried out, with each density model, over several weeks preceding the actual reentry date. Then, the corresponding percentage errors in the residual lifetime estimation and the mean residual prediction errors were calculated and compared among them. In both campaigns, the error in the single residual lifetime estimation remained well below 25% for each density model analyzed. The mean residual prediction errors remained below 10% all over the first campaign, with the exception of the last prediction with MSIS-86 and MSISE-90. However, JR-71 resulted to be the best, with mean residual prediction errors below 8% during the campaign and below 5% during the last three days. Also during the second campaign the mean residual prediction errors were, with all the models, well below 10% up to a few days preceding the reentry. Afterwards, they showed a tendency to increase, particularly during the last few hours before the final decay. But in this case the mean residual prediction error was observed to be generally lower when using JB2006.