论文部分内容阅读
氮是湿地生态系统重要生源要素,基于高光谱(350~2500 nm)遥感数据对其进行估算以实现湿地土壤全氮(TN)含量无损、快速和准确定量化具有重要意义。选取闽江河口湿地为研究区,于2013年5月,沿潮滩(高潮滩到中潮滩)采集16个土壤剖面80个样本,室内测定其光谱反射率和TN含量,并基于原始反射率(R)和光谱指数(比值指数RI、归一化指数NDI和差值指数DI)建立土壤TN含量高光谱估算模型,并进一步分析反射光谱与铵态氮(NH_4~+-N)、硝态氮(NO_3~--N)、有机质(SOM)和电导率(EC)之间的关系,以期揭示河口湿地土壤TN含量估算的机理。结果表明:土壤光谱反射率在350~600 nm,表现为高潮滩<中潮滩,而在600~2500 nm,表现为高潮滩>中潮滩;闽江河口湿地土壤TN含量与R在500 nm附近相关关系较好,并在490 nm有最大相关系数(-0.508);RI、NDI和DI大大提高了反射光谱与土壤TN含量的相关关系,其相关系数较高区域集中在600~1000 nm的波段组合,以RI(590,640)、RI(610,940)、NDI(940,590)、NDI(940,610)、DI(640,920)和DI(640,940)相关关系表现较好,能较好地实现研究区湿地土壤TN含量反演,其估算与检验模型r~2均大于0.610,RMSE均小于0.208,其中以RI(610,940)估算精度最好,估算与检验模型r~2分别为0.832和0.631,RMSE分别为0.178和0.202;闽江口湿地土壤TN含量与SOM含量密切相关是土壤TN含量估算的重要机理,而NH_4~+-N、NO_3~--N和盐分含量对其估算精度影响不大。
Nitrogen is an important biogenic element in wetland ecosystems. It is of great importance to estimate the total nitrogen (TN) content of wetland soils in a nondestructive, rapid and accurate manner based on the hyperspectral (350-2500 nm) remote sensing data. The Minjiang estuary wetland was chosen as the research area. In May 2013, 80 samples of 16 soil profiles were collected along the tidal flat (high tide beach to mid-tide beach). The spectral reflectance and TN content were determined indoors. Based on the original reflectance R and spectral indices RI, NDI and DI were used to establish the hyperspectral estimation model of TN content in soils. The reflectance spectra were compared with those of ammonium nitrogen (NH 4 + -N), nitrate nitrogen (NO_3 ~ - N), organic matter (SOM) and electrical conductivity (EC), in order to reveal the mechanism of estuary wetland soil TN content estimation. The results showed that the spectral reflectance of the soil was in the range of 350-600 nm, showing high tide beach medium tidal flat; TN content in the wetland soil of Minjiang estuary was correlated with R around 500 nm The relationship between the reflectance spectrum and soil TN content was greatly enhanced by RI, NDI and DI. The higher correlation coefficients were found in the band of 600 ~ 1000 nm The correlations between RI (590,640), RI (610,940), NDI (940,590), NDI (940,610), DI (640,920) and DI (640,940) , The estimated and test models r ~ 2 are all greater than 0.610, and the RMSE are less than 0.208, of which the RI (610,940) estimation accuracy is the best, the estimation and test model r ~ 2 are 0.832 and 0.631, RMSE is 0.178 and 0.202 respectively; The correlation between TN content and SOM content in the estuarine wetland soil is an important mechanism for estimating soil TN content. However, NH 4 + -N, NO 3 - -N and salt content have little effect on the estimation accuracy.