Detecting Soil Organic Matter Content through Reflectance Spectra

• Main Authors: Yi-Hao Dai, 戴逸豪
• Other Authors: 申雍
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/36897780761600072276

碩士 === 國立中興大學 === 土壤環境科學系所 === 105 === Soil organic matter (SOM) contents had been estimated in soil samples, which were collected from Soil Survey and Testing Center in National Chung Hsing University, Agriculture Research and Extension Station in Taoyuan, Miaoli, Tainan, Kaohsiung and Taitung District. This study added 0, 0.25, 0.50 and 1.00 mL water in soil samples separately. After water content balanced, spectral reflectance (350-2500 nm) was scanned with an ASD FieldSpec3 spectrometer. Then, the prediction models were developed with forward stepwise regression. Experimental results indicated that the wettest sample set (soil moisture content >20%) had the best prediction ability. Visible and near infrared spectroscopy (Vis-NIRS) reflectance can determine soil moisture content well. Spectral continuum removal (CR) and sample classification as Type A and Type B sample set both can optimize prediction ability. Although the prediction accuracy of extra samples reduced, but the method of choosing important wavelength could improve it. In summary, if we have soil sample, we can get its soil moisture content only by the reflectance data. Then, applying on appropriate model to soil reflectance data, consequently, we can get accurate prediction data of SOM content without soil pretreatment process (e.g. soil air-dried or oven-dried).