Share Email Print
cover

Proceedings Paper

Remote sensing of soil organic matter of farmland with hyperspectral image
Author(s): Xiaohe Gu; Lei Wang; Guijun Yang; Liyan Zhang
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Monitoring soil organic matter (SOM) of cultivated land quantitively and mastering its spatial change are helpful for fertility adjustment and sustainable development of agriculture. The study aimed to analyze the response between SOM and reflectivity of hyperspectral image with different pixel size and develop the optimal model of estimating SOM with imaging spectral technology. The wavelet transform method was used to analyze the correlation between the hyperspectral reflectivity and SOM. Then the optimal pixel size and sensitive wavelet feature scale were screened to develop the inversion model of SOM. Result showed that wavelet transform of soil hyperspectrum was help to improve the correlation between the wavelet features and SOM. In the visible wavelength range, the susceptible wavelet features of SOM mainly concentrated 460 ~ 603 nm. As the wavelength increased, the wavelet scale corresponding correlation coefficient increased maximum and then gradually decreased. In the near infrared wavelength range, the susceptible wavelet features of SOM mainly concentrated 762 ~ 882 nm. As the wavelength increased, the wavelet scale gradually decreased. The study developed multivariate model of continuous wavelet transforms by the method of stepwise linear regression (SLR). The CWT-SLR models reached higher accuracies than those of univariate models. With the resampling scale increasing, the accuracies of CWT-SLR models gradually increased, while the determination coefficients (R2) fluctuated from 0.52 to 0.59. The R2 of 5*5 scale reached highest (0.5954), while the RMSE reached lowest (2.41 g/kg). It indicated that multivariate model based on continuous wavelet transform had better ability for estimating SOM than univariate model.

Paper Details

Date Published: 2 November 2017
PDF: 6 pages
Proc. SPIE 10421, Remote Sensing for Agriculture, Ecosystems, and Hydrology XIX, 104211O (2 November 2017); doi: 10.1117/12.2278193
Show Author Affiliations
Xiaohe Gu, Beijing Research Ctr. for Information Technology in Agriculture (China)
Lei Wang, Beijing Research Ctr. for Information Technology in Agriculture (China)
Guijun Yang, Beijing Research Ctr. for Information Technology in Agriculture (China)
Liyan Zhang, Beijing Research Ctr. for Information Technology in Agriculture (China)


Published in SPIE Proceedings Vol. 10421:
Remote Sensing for Agriculture, Ecosystems, and Hydrology XIX
Christopher M. U. Neale; Antonino Maltese, Editor(s)

© SPIE. Terms of Use
Back to Top