The invention discloses a hydrological data abnormal mode detection method based on similarity measurement. The method is based on a linear piecewise representation KPRA-PLR algorithm of a key point,hydrological data is cut according to the definition of the key point, straight line fitting is carried out on each sub-sequence through the PLR algorithm, and the slope ai and the time interval deltat of a straight line are used for representing the sub-sequence; wherein each segmented sub-sequence is called as a meta-mode, adjacent meta-modes are combined to obtain a sequence mode, a weighted distance and an SDTW algorithm are respectively used for similarity measurement of the meta-mode and the sequence mode, and then an abnormal score of each sequence mode, namely a reciprocal of an average distance between the mode and other modes, is calculated; wherein the abnormal score is the k-nearest neighbor distance of the sequence mode Sx, and calculating a local abnormal factor LOF according to a k-nearest neighbor local detection principle. The abnormal mode detected by using the similarity measurement method is more accurate, and a new technology is provided for hydrological data abnormal mode detection from the perspective of data analysis.