The aquatic ecosystems and regional habitat quality of brackish Ebinur Lake in arid northwest China are significantly affected by suspended particulate matter (SPM). The Landsat series, Sentinel 2 and MODIS remote-sensing data offer a great potential to overcome the issue of the lack of reliable observation data for continuous SPM monitoring, but more accurate SPM inversion models are required for the arid regions. Hence, this research aims to: (1) determine the optimal strategy for applying the SPM inversion model to various time series images, (2) explore the inversion mode of SPM under the missing-image scenario, and (3) analyze the spatiotemporal changes of SPM over the Ebinur Lake lake from May to October during the 2011–2020 period. The findings show that ESTARFM_IVS (Inversion of SPM using images generated by Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model) produces the best inversion accuracy, with the values of R² ≥ 0.57, RMSE ≤ 44.51 mg/L. Secondly, the exponential model for the Landsat 5/7 and fused images were adopted, while the Quasi-Analytical Algorithms for reference wavelengths of 655 nm and 665 nm (QAA_655(665)) were used for the Landsat 8 and Sentinel 2 images. Thirdly, the SPM fluctuations varied significantly, and the SPM concentrations showed an increasing trend with a widening amplitude over time. The inter-annual SPM increasing trend was significant, while the inter-month SPM variations have not changed much. When compared to the lake's core area, the SPM concentrations in the nearshore area were noticeably higher, and this has been consistent over the years. The results should serve as helpful guides for monitoring brackish lake water environment using satellite technologies.