基于高光谱的梭梭幼苗生境土壤氮素含量估算模型研究

Study on Estimation Model of Soil Nitrogen Content in Haloxylon ammodendron Seedlings Habitats Based on Hyper-spectral Data

【:目的】研究荒漠区梭梭幼苗植株冠层光谱特征,建立基于高光谱的幼苗生境土壤氮素含量最佳估测模型,为荒漠区植物与土壤养分状况的实时监测与精确诊断提供重要依据。【方法】以古尔班通古特沙漠梭梭幼苗及其生境土壤为研究对象,采用便携式光谱仪测定幼苗植株冠层光谱反射率,结合室内分析的幼苗生境土壤氮素含量,利用统计学方法分析幼苗植株冠层光谱与生境土壤氮素含量之间的关系。【结果】生长期内幼苗植株冠层光谱反射率一阶微分与土壤氮素含量在403~485、509~538、677~806、1489~1616 nm波段范围内呈显著正相关,其中在428、516、751、801、1534 nm处呈极显著正相关,以751 nm处的相关系数最大,为最强敏感波长。【结论】基于植株冠层光谱反射率一阶微分敏感波段建立的反演模型Y=133.186X751+8.803,其决定系数最大,相对误差最小,拟合效果最优,预测精度较高,可作为荒漠区梭梭幼苗生境土壤氮素含量的最优估算模型。

【Objective】The study aimed to explore the characteristics of canopy spectral reflectance of Haloxylon ammodendron seedlings,establish the best estimation model of soil nitrogen content in seedlings habitats based on hyper spectral,and provide an important basis for real-time monitoring and accurate diagnosis of plant and soil nutrient status in desert area.【Method】The H.ammodendron seedlings and soil of habitats in Guerbantonggute Desert were used as the research objectives.The canopy spectral reflectance of H.ammodendron seedlings were measured by the portable spectrometer.Combined with the indoor analysis on the soil nitrogen content in seedlings habitats,Stepwise Regression Method was used to analyze the relationship between canopy spectral reflectance and soil nitrogen content in seedlings habitats.【Result】During the growth period,the relationship between the canopy first derivative spectral reflectance of seedlings and soil nitrogen content was significantly positively correlated in the band ranges of 403-485 nm,509-538 nm,677-806 nm and 1489-1616 nm.There was extremely significantly positive correlation at the 428 nm,516 nm,751 nm,801 nm and 1534 nm,and the correlation coefficient was maximum at 751 nm,which was the most sensitive wavelength.【Conclusion】Based on the canopy first derivative spectral reflectance of seedlings at the sensitive bands,the inversion model Y=133.186X751+8.803 is established,which shows that the determination coefficient is maximum and the relative error is minimum,with optimal fitting effect and higher prediction precision.It can be used as the best estimation model for soil nitrogen content in H.ammodendron seedlings habitats.