Combining Multi-Indices by Neural Network Model for Estimating Canopy Chlorophyll Content: a Case Study of Interspecies Competition between Spartina alterniflora and Phragmites australis
Runhe Shi 2,3,4
More details
Hide details
School of Surveying and Geo-informatics, Shandong Jianzhu University, Jinan, 250101, China
Key Laboratory of Geographic Information Science, Ministry of Education, East China Normal University, Shanghai, 200241, China
Joint Laboratory for Environmental Remote Sensing and Data Assimilation, East China Normal University & Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Shanghai, 200241, China
School of Geographic Sciences, East China Normal University, Shanghai, 200241, China
Pingjie Fu   

School of Surveying and Geo-Informatics, Shandong Jianzhu University, China
Submission date: 2021-02-18
Final revision date: 2021-06-21
Acceptance date: 2021-06-25
Online publication date: 2021-12-16
Publication date: 2021-12-23
Pol. J. Environ. Stud. 2022;31(1):199–217
The invasive species Spartina alterniflora show a significant coexistence zonation pattern with local Phragmites australis in different mixture ratio, increasing the difficulty to monitor their distribution directly by remote sensing. Canopy chlorophyll content (CCC) is an important indicator to monitor the growth and physiological status. The objective of this study was to estimate CCC under different mixture ratio. Five spectral indices were selected and combined via back propagation (BP) neural network model for estimating CCC. Combining multi-indices yielded better results (R2 = 0.7729, RMSE = 53.01 ug.cm-2) on average than the best single spectral index (R2 = 0.7190, RMSE = 63.53 ug.cm-2) without distinguishing interspecies competition, with a total increase of 7.5% in the R2 and a decrease of 16.56% in the RMSE. Meanwhile, when considering interspecies competition, the estimating results obtained by the BP neural network model achieved a further improvement of the R2 value, ranging from 3.57% to 20.37%, while the prediction error reduced at varying degrees (maximum reduction of 23.78%). The results indicate that combining multi-indices by BP neural network model can alleviate the influence of interspecies competition and achieve higher estimating accuracy.