期刊
GEODERMA
卷 315, 期 -, 页码 170-177出版社
ELSEVIER
DOI: 10.1016/j.geoderma.2017.11.032
关键词
Imaging spectroscopy; Carbon; Nitrogen; Spatial statistics
类别
资金
- National Science and Engineering Council (NSERC) [RGPIN-2014-04693]
Soil is a critical component of global biogeochemical cycles, and there is an increasing need for cost effective tools to measure soil carbon stocks and determine soil nitrogen contents. Reflectance spectroscopy can deliver large volumes of soil carbon data. However, as soil carbon concentrations can be spatially heterogeneous, imaging spectroscopy presents the best potential to provide high resolution measurements and accurately characterize soil carbon heterogeneity. For this study, discrete, intact and unground soil samples were collected and analyzed using a SisuROCK automated hyperspectral imaging system in a laboratory setting, focused on the shortwave infrared portion of the electromagnetic spectrum. Samples were also analyzed for soil organic carbon and total nitrogen concentrations by dry combustion to prepare a training data set. Predictive models were built using continuous wavelet processing along with partial least squares regression and CUBIST models. Spatial variation of carbon and nitrogen was determined using Moran's i and comparisons of spatial variations among soil types and horizons were made using a spatial generalized least squares model. Overall, soil organic carbon was more aggregated in Chernozemic soils and in B and C horizons compared to A horizons. Nitrogen in turn showed more aggregation for all soil types and horizons compared to soil organic carbon. Results indicated that imaging spectroscopy can be successfully used to measure and characterize the spatial variability of soil carbon and nitrogen at the soil aggregate scale.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据