<p>This paper presents a real-time trajectory optimization and onboard replanning framework for a lunar lander utilizing solid rocket motors. While solid propulsion offers a simplified and highly reliable architecture, it precludes thrust modulation during combustion, requiring precise predetermination of ignition time and burn duration. To address these operational constraints, we formulate the powered descent phase as a Sequential Convex Programming (SCP) problem, incorporating critical flight conditions such as fixed-thrust operation and vertical landing constraints. Furthermore, propellant temperature variations induced by the lunar thermal environment cause significant thrust magnitude errors that conventional guidance often cannot overcome. A key contribution of this work is an onboard, closed-loop trajectory replanning scheme based on in-flight thrust estimation. This approach periodically re-optimizes the trajectory to compensate for large thrust uncertainties. Numerical simulations demonstrate that the proposed real-time replanning algorithm successfully mitigates severe thrust errors, ensuring precise terminal convergence and reliable soft-landing performance. Numerical simulations demonstrate that the proposed framework effectively constrains horizontal landing errors to within a few meters even under thrust deviations of up to 1.5%, significantly enhancing the landing reliability compared to the open-loop baseline.</p>