Aiming at the problem that one remote sensing image atmospheric correction algorithm is difficult to be applied to the spectral correction of different types of lakes at the same time, we select Nansihu as the study area, collect Sentinel-2 images of the region from 2019 to 2022, collect the spectral data of spacious lakes and lakes covered by aquatic vegetation during the same period. Based on the adaptive weighting algorithm, two new frameworks for atmospheric correction of multispectral images, the Adaptive Weighted Average Atmospheric Correction Algorithm (AWA-AC) and the Improved Adaptive Weighted Average Atmospheric Correction Algorithm (IAWA-AC), are constructed by taking advantage of the advantages of the three traditional atmospheric correction methods, namely, Acolite, Sen2Cor and C2RCC. The results of the atmospheric correction experiments on the Sentinel-2 image of Nansi Lake using each algorithm and the evaluation of the comparative accuracies show that the new framework of atmospheric correction is more effective than the single traditional algorithm, and the new framework of atmospheric correction is better than the single traditional algorithm in terms of coefficient of determination (R²), root mean square error (RMSE), and average unbiased relative error (AURE) of the measured and atmospherically corrected image spectra in the area during the time period of the study. The new framework proposed in this study has the maximum enhancement values of 79.75%, 71.55% and 70.43% for the three metrics, respectively, compared with the single traditional atmospheric correction algorithm. In the absence of measured spectral data to derive R², atmospheric correction of remotely sensed images using the IAWA-AC algorithm constructed in this study is able to obtain better spectral fidelity.