This document discusses two Kalman filtering approaches to automate the correction of long-term geolocation errors in MODIS and VIIRS instruments. The current manual approach uses least-squares analysis with linear and sinusoidal curves, but the Kalman filtering approaches can provide automated daily updates. The first approach models error trends with linear and sinusoidal curves in a Kalman filter. The second estimates sensor roll, pitch and yaw errors through Euler angle differential equations. Preliminary tests of the first approach on over 10 years of MODIS data show promising results in reducing geolocation errors, while the second approach requires more examination. Further testing is needed to validate and refine the Kalman filtering methods for automated long-term geolocation error correction.