Department of Chemistry and Optical Science and Technology Center, University of Iowa, Iowa City, IA 52242, USA.
Appl Spectrosc. 2013 Aug;67(8):913-23. doi: 10.1366/12-06807.
A spectral synthesis strategy is introduced to help obtain estimates of path-integrated concentrations in passive Fourier transform infrared (FT-IR) remote sensing measurements conducted during field-monitoring experiments. Obtaining quantitative information from passive infrared data is challenging because of the joint effects of temperature and concentration on spectral intensities. The collection of calibration data for use in modeling spectral intensities for a given set of experimental conditions is also costly and labor intensive. In the work presented here, a radiance model is defined for use in synthesizing calibration spectra that serve as inputs to partial least-squares (PLS) models that relate spectral intensities to path-integrated concentrations. The field data for which quantitative estimates are desired are used to estimate the background temperature associated with a given time and set of experimental conditions. Sample temperatures can be obtained through either experimental measurement or by estimating one calibration release. Given these temperatures, calibration data can be synthesized and the PLS model developed. This methodology is tested with stack monitoring data obtained from controlled releases of pure and mixture samples of heated ethanol and methanol. Experiments were conducted across 6 days with stack temperatures of 150 to 200 °C and with path-integrated concentrations in the range of 10 to 300 parts per million meters. Median relative errors in the estimates of path-integrated concentration were typically in the range of 20% or less, with the best results observed at the higher stack temperatures.
引入了一种光谱合成策略，以帮助获得在现场监测实验中进行的被动傅里叶变换红外（FT-IR）远程传感测量中路径积分浓度的估计值。由于温度和浓度对光谱强度的联合影响，从被动红外数据中获取定量信息具有挑战性。为了在给定的实验条件下对光谱强度进行建模而收集校准数据也是昂贵且劳动密集型的。在目前的工作中，定义了一个辐射模型，用于合成校准光谱，作为偏最小二乘（PLS）模型的输入，该模型将光谱强度与路径积分浓度相关联。希望获得定量估计的现场数据用于估计与给定时间和一组实验条件相关的背景温度。样品温度可以通过实验测量或通过估计一个校准释放来获得。给定这些温度，可以合成校准数据并开发 PLS 模型。该方法通过对加热乙醇和甲醇的纯样和混合样的受控释放获得的烟囱监测数据进行了测试。实验在 6 天内进行，烟囱温度在 150 至 200°C 之间，路径积分浓度在 10 至 300 ppm 之间。路径积分浓度估计的中位数相对误差通常在 20%或以下，在较高的烟囱温度下观察到了最佳结果。
