Spectroscopy and Spectral Analysis, Volume. 37, Issue 2, 557(2017)
Study on the Relationship between Soil Emissivity Spectra and Content of Soil Elements
In this paper, based on the measurements of soil elements content and infrared spectra of 26 soil samples collected in more than 10 places, the relationship between soil emissivity in mid-infrared bands and the content of 11 soil elements including organic matters such as NO3-N, P, K, Ca, Mg, Cu, Fe, Mn, Zn and pH are analyzed. The bands where the soil elements content are significantly correlated with emissivity are given. And soil elements content estimation method is established based on the soil emissivity spectra with the partial least squares regression model and multiple stepwise regression model. The results show that: (1) In 8~10 μm, the correlation coefficient (R2) between Ca and soil emissivity is the highest, followed by Mg, Mn and Fe, with the highest correlation coefficient of 085 and the lowest, 052. In the range of 6~8 μm, the correlations between the contents of K, Fe, NO3-N, Zn and emissivity decrease gradually, with the highest correlation coefficient of 075 and the lowest 048. In 10~14 μm, the correlation between soil elements contents and emissivity is the highest for Mn, followed successively by P and K. (2) The scatter plot of soil emissivity and pH value has a parabola relation basically. The emissivity is the highest when pH value is 7, while the emissivity decreases gradually with the gradual decrease of pH value. (3) The accuracy of the estimated soil elements content from the partial least squares regression method is higher than that from the multiple stepwise regression method. It is noted that R2 between the measurements and the estimates for the elements of Cu, Fe and Ca from the partial least squares regression method are very high (larger than 09). Additionally, using the simulated emissivity spectrum in the ASTER thermal infrared bands, modeling R2 and validation R2 between the measurements and the estimates for the elements of Ca from the multiple stepwise regression method are high (0774 and 0892, respectively). Using the simulated emissivity spectrum in the MODIS infrared bands, modeling R2 and validation R2 for Ca and Fe are higher than 085, and modeling R2 and validation R2 for Mg, K are higher than 05. As a whole, the emissivity spectrum in ASTER band 10 and band 11 and MODIS bands 28, 29, 30 are more sensitive to soil elements content, and thus they are more suitable for the estimation of soil elements content.
Get Citation
Copy Citation Text
DONG Xue, TIAN Jing, ZHANG Ren-hua, HE Dong-xian, CHEN Qing-mei. Study on the Relationship between Soil Emissivity Spectra and Content of Soil Elements[J]. Spectroscopy and Spectral Analysis, 2017, 37(2): 557
Received: Dec. 7, 2015
Accepted: --
Published Online: Jun. 20, 2017
The Author Email: