In the process of monitoring the concentration of chlorophyll-a in water,the destruction of the spectral authenticity by the atmospheric medium is a difficult problem to be solved.In response to this problem,this study focuses on atmospheric correction of ZY-1 02D hyperspectral satellite imagery from three perspectives:external atmospheric products,internal atmospheric compensation parameters and spectral mean of the image,so as to enhance water signal accuracy;At the same time,with the help of multi-dimensional spectral index,combined with the CatBoost machine learning algorithm,the inversion accuracy of water chlorophyll-a concentration is further improved.The results show that in the application of atmospheric correction algorithm in Dushan Lake,6S is better than QUAC,and FLAASH is the worst.CatBoost model can better fit the prediction error and improve the inversion accuracy.6S algorithm-Four-band parameter-CatBoost model inversion combination works best(R2 reaches 0.80).
维普
万方数据