摘要
以兴国县稻田土高光谱反射率为研究对象,对比分析了同一种光谱反射率变换形式下土壤全钾、速效钾与高光谱反射率的相关性,提取了全钾和速效钾的高光谱敏感波段,建立了基于反射光谱特征的南方丘陵稻田土全钾、速效钾高光谱反演模型.经分析可知,在355~620 nm波段,土壤全钾、速效钾含量与光谱反射率相关性同增同减,而在621~2 250 nm波段内,土壤全钾含量与光谱反射率相关性要大于土壤速效钾;通过分析兴国县稻田土全钾、速效钾含量与光谱反射率18种数学变换的相关系数,提取全钾的敏感波段为602、804 nm,速效钾的敏感波段为602、1 058、1 638、2 214 nm;采用偏最小二乘回归,利用高光谱指数构建的反演模型能较好地预测全钾、速效钾含量,模型建模的相关系数和验证系数都较高,基于速效钾含量建立的南方丘陵稻田土高光谱反演模型预测能力较好.
This article took the hyper-spectral reflectance of paddy soil in Xingguo County as the research objectives,analyzed the correlations of soil available potassium and total potassium with the hyper-spectral reflectance,derived the hyper-spectral indices of the soil total potassium and available potassium,then built the paddy soil total potassium and available potassium predicting model based on the correlation between potassium content and spectral indices. The results showed that the composition of soil available potassium and total potassium caused the difference in their hyper-spectral characteristics. In the band of 355-620 nm,the available potassium and total potassium had the same effect on the hyper-spectral reflectance;in the band of 621-2 250 nm,the correlation of soil total potassium with hyper-spectral reflectance was greater than that of the available soil potassium. By analyzing the correlation coefficient of paddy soil potassium content and 18 kinds of mathematical transformations of hyper-spectral reflectance to extract sensitive wavelength,the sensitive wavelengths of total potassium were 602 nm and 804 nm,the sensitive wavelengths of available potassium were 602 nm,1 058 nm,1 638 nm and 2 214 nm. The predicting models for paddy soil potassium content were built with hyper-spectral indices as independent variables and available potassium as dependent variable,and the models were quite good in stability and predictability.
出处
《广东农业科学》
CAS
2015年第7期37-42,共6页
Guangdong Agricultural Sciences
基金
国家"十二五"科技支撑计划项目(2012BAD 04B11)
江西省博士后择优项目(JX2013018)
国家自然科学基金(41361049)
关键词
土壤全钾
土壤速效钾
高光谱特征
反演模型
soil total potassium
soil available potassium
hyper-spectral characteristics
inversion model