This paper describes the target pointing calibration implemented on the Diwata-2 microsatellite for high accuracy Earth observations. An experimental procedure of using Lunar observations is proposed to address the systematic errors that were only apparent during its flight operations. These errors include misalignments between the star tracker sensors and its optical payloads. The circular Lunar outline is presented as the reference target for this in-flight calibration procedure. By analyzing image error patterns, the discrepancies with the onboard attitude measurements and the actual attitude are revealed by the observation images. Further fine tuning of the observations was conducted by managing the satellite's execution of its panoramic capture in a deterministic approach. This accounted for the persisting issues with the satellite, such as system latency and orbital model inaccuracies to a certain extent. With this calibration procedure, an overall average of 0.2° in RMS with a standard deviation of 0.12° pointing accuracy for Earth observation was recorded with the latest calibration iteration. In at least 24 % of the observation trials, Diwata-2 achieved the 0.1° accuracy requirement needed for an effective observation by its High Precision Telescope. This calibrated system and operation strategy were then successfully applied to Diwata-2's routine operations.