土壤Cd含量实验室与野外DS光谱联合反演

doi:10.3964/j.issn.1000-0593(2019)10-3223-09

Abstract
References
Related Citation (15)

Download: PDF (6076 KB)
Export: BibTeX | EndNote (RIS)

Abstract Theoretically, hyperspectral remote sensing aided content estimation of soilheavy metal can greatly reduce the cost of conventional chemistries. As a result, hyperspectral remote sensing is gradually becoming a key technology to effectively explore the spatial distribution of soil heavy metal and consequently guide theprevention and remediation of heavy metal polluted soil. However, currently reported hyperspectral retrieval models for soil heavy metal estimation are mostly with laboratory spectra under specifically controlled conditions. Due to the impacts of environmental factors, such as illumination, soil moisture content, and roughness onin-situ field spectra, the wide implementation of in-situ field spectra based remote sensing detection of soil heavy metal is still experiencing the difficulty of reliability. For this, 46 soil samples were firstly collected from a mining area in Hengyang of Hunan Province, China. Then the spectra (ranged 350~2 500 nm) and Cd content of these soil samples were measured using ASD field spectrometer and ICP-atomic emission spectrometry under in-situ field and laboratory conditions, respectively. Then, considering the prior knowledge of laboratory spectra, the principal stepwise regression method was used to develop the Cd content estimation model based on combined laboratory and direct standardization (DS) algorithm transformed in situ field DS spectra with the model robust test by cross-validation. In order to further prove the effectiveness of the model with combined laboratory and DS transformed in-situ field DS spectra, the performance of this model was then compared with four types of hyperspectral remote sensing models including those with spectra from the laboratory, in-situ field, DS transformed in-situ field only, as well as with combined laboratory and in-situ field, one by one. The result shows that while the precision of the hyperspectral remote sensing model with in-situ field spectra (R²=0.56) is lower than the one with laboratory spectra (R²=0.64), the precision of the model with DS transformed in-situ field spectra is improved (R²=0.66). The model with combined laboratory and DS transformed in-situ field spectra is the one with the highest accuracy (R², 0.72). Meanwhile, this highest robust model discloses that the wavebands located at 482, 565, 979, and 2 206 nm have significantly strong correlations with the soil Cd content. And this result is physically consistent with the model with laboratory spectra. In summary, results in this study confirm the role of the prior knowledge of laboratory spectra and DS algorithm in enhancing the reliability of in-situ field spectra based hyperspectral remote sensing model for soil Cd content estimation. It could provide new theoretical and methodological evidence for the development of soil Cd content estimation by using hyperspectral remote sensing.

Key words：Soil; heavy metal; Hyperspectral; DS method; Remote sensing retrieval

Received: 2018-08-23 Accepted: 2018-12-12

ZTFLH:

TP79

Cite this article:

URL:

https://www.gpxygpfx.com/EN/10.3964/j.issn.1000-0593(2019)10-3223-09 OR https://www.gpxygpfx.com/EN/Y2019/V39/I10/3223

[1] Gu Y W, Li S, Gao W, et al. Acta Ecologica Sinica, 2015, 35（13）: 4445.
[2] Shi T Z, Chen Y Y, Liu Y L, et al. Journal of Hazardous Materials, 2014, 265（2）: 166.
[3] Jiang Xiaolu, ZOU Bin, Feng Huihui, et al. Journal of Geochemical Exploration, 2019, 196: 1.
[4] TAO Chao, WANG Ya-jin, ZOU Bin, et al（陶超, 王亚晋, 邹滨, 等）. Spectroscopy and Spectral Analysis（光谱学与光谱分析），2018, 38（6）: 1850.
[5] Liu Y L, Li W, Wu G F, et al. Geo-Spatial Information Science, 2011, 14（1）: 10.
[6] Ackerson J P, Morgan C L S, Ge Y. Geoderma, 2017, 286: 131.
[7] Lagacherie P, Baret F, Feret J B, et al. Remote Sensing of Environment, 2008, 112（3）: 825.
[8] Xu C, Zeng W, Huang J, et al. Remote Sensing, 2016, 8（1）: 42.
[9] Choe E, Meer F V D, Ruitenbeek F V, et al. Remote Sensing of Environment, 2008, 112（7）: 3222.
[10] Song L, Jian J, Tan D J, et al. International Journal of Applied Earth Observation & Geoinformation, 2015, 34（1）: 1.
[11] Ji W，Viscarra Rossel R A, Shi Z. European Journal of Soil Science, 2015, 66（4）: 670.
[12] CHEN Yi-yun, QI Kun, LIU Yao-lin, et al（陈奕云, 漆锟, 刘耀林, 等）. Spectroscopy and Spectral Analysis（光谱学与光谱分析），2015, 35（6）: 1705.
[13] TU Yu-long, ZOU Bin, JIANG Xiao-lu, et al（涂宇龙, 邹滨, 姜晓璐, 等）. Spectroscopy and Spectral Analysis（光谱学与光谱分析），2018, 38（2）: 575.
[14] Zhai L, Li S, Zou B, et al. Atmospheric Environment, 2018, 181: 145.
[15] Piekarczyk J, Ka?mierowski C, Królewicz S, et al. IEEE Journal of Selected Topics in Applied Earth Observations & Remote Sensing, 2016, 9（2）: 827.
[16] Siebielec G, Mccarty G W, Stuczynski T I. Journal of Environmental Quality, 2004, 33（6）: 2056.
[17] SONG Lian, JIAN Ji, TAN De-jun, et al（宋练, 简季, 谭德军, 等）. Spectroscopy and Spectral Analysis（光谱学与光谱分析），2014，34（3）: 812.
[18] Cheng Hang, Shen Ruili, Chen Yiyun, et al. Geoderma, 2019, 336: 559.