Abstract: In large shallow lakes, where hydrodynamic conditions are relatively weak, biochemical processes play a dominant role in water quality evolution. This makes key parameters such as degradation coefficient and internal source strength significantly influential in water quality simulation results. However, due to the high complexity of biochemical processes in natural water bodies, these parameters are often difficult to obtain directly, posing a key challenge in improving the accuracy of water quality models. This paper focuses on the degradation coefficient and internal source strength, derives the adjoint system of the depth-averaged convection-diffusion equation based on the adjoint equation method, and establishes a gradient analytical expression for simulation errors with respect to parameters. On this basis, combined with the Broyden-Fletcher-Goldfarb-Shanno (BFGS) optimization algorithm, a collaborative inversion framework for water quality parameters is constructed and implemented on the OpenFOAM numerical platform. Taking the simulation of CODMn in the South Four Lakes of the East Route of the South-to-North Water Diversion Project as an example, the optimized degradation coefficient k=0.0185d?1 and internal source strength S=0.179 g·m?2·d?1 are both within a reasonable physical range; the relative error of the model simulation RMSE is less than 16.6%, significantly reduced compared to the initial parameter scheme; the optimization process converges stably with high computational efficiency. The results show that this method can effectively improve the parameter identification accuracy of shallow lake water quality models, providing methodological support for parameter inversion under complex water environmental conditions.