The Jianghan Plain is a flood-prone area in the Yangtze River Basin, and accurate monitoring of floods is of great significance for disaster prevention and mitigation. In this paper, a multi-source data collaborative flood range extraction method is proposed. By constructing long-term remote sensing observation surface water data, and constructing an exclusion layer based on the prior knowledge of water distribution, the flood information extracted by the Normalized Difference Flood Index (NDFI) method method is corrected to eliminate partial misclassification errors. The flood disaster event in the Jianghan Plain in the summer of 2020 was selected to carry out experimental verification. The experimental results show that the proposed method can effectively improve the situation that normal water bodies in the Jianghan Plain are mistakenly classified as floods, mainly in that the pixels that are mistakenly classified as floods in some farmland, lake and river areas are eliminated through the setting of the exclusion layer, which improves the accuracy of flood mapping.