论文部分内容阅读
土壤有机碳的有效评估对全球碳循环和农业可持续发展具有重要作用。可见光-近红外光谱技术已广泛用于土壤有机碳含量的反演研究。然而,基于可见光-近红外光谱的土壤有机碳反演模型通常具有一定的区域局限性。本文基于湖北钟祥市和洪湖市两个区域的土壤光谱和有机碳量测数据(样本数分别为100和96),探究土壤有机碳反演模型在不同区域间的传递性。结果表明,钟祥市或洪湖市区域模型都不能用于另一个区域,但基于钟祥样本全集与洪湖区域30个土壤样本数据建立的模型对洪湖区域土壤有机碳含量有很好的预测效果(R~2=0.88,RMSE=2.51g·kg~(-1))。尽管模型在不同区域间的传递性非常有限,但将少量目标区域样本添加到现有区域土壤光谱库中所建立的偏最小二乘回归模型能够估算目标区域土壤有机碳的含量,降低目标区域的采样和量测成本。
Effective assessment of soil organic carbon plays an important role in the global carbon cycle and agricultural sustainable development. Visible-NIR spectroscopy has been widely used in the inversion of soil organic carbon content. However, soil organic carbon retrieval models based on visible-near-infrared spectroscopy usually have some regional limitations. Based on the data of soil spectroscopy and organic carbon (samples of 100 and 96, respectively) from Zhongxiang City, Hubei Province and Honghu City, the transmissibility of soil organic carbon in different regions was explored. The results show that neither Zhongxiang nor Honghu regional model can be used in another region. However, the model based on the data of 30 soil samples collected from Zhongxiang sample and Honghu region has a good predictive value for soil organic carbon content in Honghu region (R ~ 2 = 0.88, RMSE = 2.51 g · kg -1). Although the model has very limited transmissibility between different regions, the partial least-squares regression model established by adding a small sample of target regions to the existing regional soil spectral library can estimate the soil organic carbon content in the target region and lower the target region Sampling and measuring costs.