Prediction of Soil Organic Matter Using Visible-Short Near-Infrared Imaging Spectroscopy
JIAO Cai-xia1, ZHENG Guang-hui1*, XIE Xian-li2, CUI Xue-feng3, SHANG Gang1
1. School of Geographic Sciences, Nanjing University of Information Science & Technology, Nanjing 210044, China
2. Institute of Soil Science, Chinese Academy of Sciences, Nanjing 210008, China
3. School of Systems Science, Beijing Normal University, Beijing 100875, China
Abstract:Soil organic matter (SOM) is a crucial indicator of soil fertility and an important form of global soil carbon. It is the premise and basis of ensuring food security and assessing climate change to estimate SOM content and its changes rapidly. The traditional method of SOM determination is time-consuming with the high cost and environmental risks. Soil reflectance spectroscopy can be faster and cheaper than the conventional method, do not generate chemical residues and are non-destructive to the samples. However, spatial interpolation technique was still needed to map SOM after estimation of SOM in a point soil sample by reflectance spectroscopy. Imaging spectroscopy (also known as hyperspectral imaging) collects a spectral curve for each pixel, which enlarges the envelope of point spectrometry into a spatial domain and provides a technical basis for spatial mapping of SOM. This novel technique has not yet been fully utilized for SOM mapping. Therefore, the spectral index established by laboratory visible-short near-infrared imaging spectroscopy data can be used to estimate SOM and explore the mechanism, which lays a theoretical foundation for SOM mapping of remote sensing. In this study, a spectral index, named deviation of an arch (DOA), was established using the information of three wavelengths. The correlation between DOA and SOM was analyzed by the scatter diagram. Then, the samples were randomly split into training and validation data sets for 1 000 times. Nonlinear regression and partial least square regression (PLSR) were used to calibrate DOA or spectroscopy to SOM, respectively. The performances were compared to explore the feasibility of SOM estimation using imaging spectroscopy. The results indicate that the SOM content in the study area was relatively low, and its variation range was large. There was a significant logarithmic relationship between DOA and SOM. Logarithm function can be used to model DOA and SOM and provide reasonable and stable results. The performance of DOA regression is better than PLSR. The possible reason is that the spectral data used by PLSR contains some information unrelated to SOM, which affects the accuracy of PLSR. We can conclude that this spectral index, DOA, can be used for SOM mapping, although it is deduced from three wavelengths. It provides a new idea and methods for SOM mapping based on satellite remote sensing data in the future.
Key words:Imaging spectroscopy; Soil organic matter (SOM); Partial least squares regression (PLSR); Deviation of arch (DOA)
[1] ZHAO Yong-cun, XU Sheng-xiang, WANG Mei-yan, et al(赵永存, 徐胜祥, 王美艳, 等). Bulletin of Chinese Academy of Sciences(中国科学院院刊), 2018, 33(2): 191.
[2] Shi Z, Ji W J, Viscarra Rossel R A, et al. European Journal of Soil Science, 2015, 66(4): 679.
[3] Ji W J, Li S, Chen S C, et al. Soil and Tillage Research, 2016, 155: 492.
[4] ZHAO Xiao-min, YANG Mei-hua(赵小敏, 杨梅花). Acta Pedologica Sinica(土壤学报), 2018, 55(1): 1.
[5] Viscarra Rossel R A, Behrens T, Ben-Dor E, et al. Earth-Science Reviews, 2016, 155: 198.
[6] ZHU A-xing, YANG Lin, FAN Nai-qing, et al(朱阿兴, 杨 琳, 樊乃卿, 等). Progress in Geography(地理科学进展), 2018, 37(1): 66.
[7] ZHANG Gan-lin, ZHU A-xing, SHI Zhou, et al(张甘霖,朱阿星,史 舟,等). Progress in Geography(地理科学进展),2018,37(1):57.
[8] PENG Jie, ZHOU Qing, ZHANG Yang-zhu, XIANG Hong-ying(彭 杰, 周 清, 张杨珠, 向红英). Acta Pedologica Sinica(土壤学报), 2013, 50(3): 517.
[9] Zheng G H, Ryu D, Jiao C X, et al. Pedosphere, 2016, 26(1): 130.
[10] XU Bin-bin, DAI Chang-da(徐彬彬, 戴昌达). Chinese Science Bulletin(科学通报), 1980, 6: 282.
[11] Zeng R, Rossiter D G, Yang F, et al. Geoderma, 2017, 303: 78.
[12] Peng J, Biswas A, Jiang Q S, et al. Geoderma, 2019, 337: 1309.
[13] Mouazen A M, Kuang B, De Baerdemaeker J, et al. Geoderma, 2010, 158: 23.
[14] Lucà F, Conforti M, Castrignanò A, et al. Geoderma, 2017, 288: 175.
[15] Viscarra Rossel R A, Walvoort D J J, McBratney A B, et al. Geoderma, 2006, 131: 59.
[16] Brown D J, Shepherd K D, Walsh M G, et al. Geoderma, 2006, 132: 273.