The invention relates to a
cyanobacterial bloom extraction method based on spectrum and phenological characteristics, and relates to the field of
ecology, and the method comprises the steps: selecting a
vegetation index PCI which is excellent in
cyanobacterial bloom recognition, building a localized
phycocyanobilin inversion model, and carrying out the homogeneity test of pixels around a sampling point; by calculating a PCI gradient image
time sequence,
cyanobacterial bloom and background water boundaries are identified, and a threshold value for distinguishing
cyanobacteria and non-
cyanobacteria pixels is determined by using a maximum gradient average value; constructing a cyanobacterial bloom recognition model based on the LSTM neural network; calculating cyanobacterial bloom
outbreak intensity and coverage area, identifying a hot spot area and a
cold spot area, and depicting distribution characteristics of cyanobacterial bloom in space; a PCI
time sequence data set is synthesized, the evolution trend of a cyanobacterial bloom key phenological index
time sequence is described, the innovation of the method lies in that spectrum and phenological characteristics are fused, cyanobacterial bloom spatial and temporal distribution is accurately analyzed, and guarantee is provided for
water environment treatment and drinking
water safety.