A multi-mode variable-weight spatiotemporal fusion evidential network method based on an evidence correction mechanism was proposed in order to address the issue that traditional situation assessment methods struggled to continuously provide effective results when information was incomplete due to their fixed models and reasoning mechanisms. Evidence types were classified according to the degree of temporal correlation of information, and a modular evidential network reasoning model was constructed. A multi-mode evidence correction method was proposed combined with temporal information prediction technology based on the type and incompleteness degree of evidence information. An adaptive generation method for network node weights was proposed considering the differences in the reliability of corrected evidence and based on the variable weight principle. The temporal correlation of non-mutant evidence was strengthened, and the accuracy of continuous situation assessment was improved by designing a three-step spatiotemporal fusion mechanism with a time-decreasing discount rate. The effectiveness of the proposed method was verified through UAV air combat simulation experiments under two scales with four levels of information incompleteness.