Abstract:  The thermosphere density model used for spacecraft orbit prediction generally has an error of about 30%，which affects the precise orbit determination and load control of LEO satellites.Based on the relationship between the change of horizontal motion of the low-orbit spacecraft and the atmospheric density，the TLE data of the GRACE（gravity recovery and climate experiment）satellite were used to invert the atmospheric density along the orbit in 2003 and 2007.After concluding the cause of the error in the relational analysis，the lognormal distribution was used to fit the density ratios.By analyzing the influence of solar radiation and geomagnetic index on atmospheric density changes，a correction and forecasting method of thermospheric density model based on space environment index was proposed.This method was used to correct the calculated density of the MSIS86 model in 2003，2004，2007 and 2008，the average relative error of the model was reduced from 33.33%~59.62% to 11.55%~15.13%，and the improvement in low solar activity years is 1.5~2 times of high solar activity years.The forecast correction was carried out on the empirical model results in 2009，which reduced the forecast error by 3649% and improved the model accuracy.

Key words:  , thermospheric density determination；MSIS-86 atmospheric model；atmospheric model correction；two-line element sets（TLEs）；space weather index