Search

Ammonia emission has attracted considerable attention concerning environmental protection in the world. About 33% of the atmospheric ammonia emission comes from farmland fertilization, so the monitoring of ammonia emissions from farmland is of great significance for controlling atmospheric ammonia pollution. In recent years, tuneable diode laser absorption spectroscopy (TDLAS) has been applied to the measurement of farmland ammonia emission; however, the spectral signal of the open optical path is susceptible to distortion due to temperature, noise, turbulence, etc. Meanwhile, the inversion of the distorted spectrum will result in abnormal concentration values that reduce the detection accuracy. This paper proposes a TDLAS spectral quality evaluation method based on dynamic time warping (DTW) algorithm and support vector data description (SVDD) algorithm. DTW algorithm was used to analyse the similarity between the measured spectrum and the reference spectrum, then SVDD algorithm was used to generate a classification model according to the results of DTW algorithm. DTWSVDD algorithm could effectively avoid obtaining wrong inversion concentration and improve the accuracy of concentration monitoring with identification accuracy of spectral quality reached more than 98%. Applied to the measurement of farmland ammonia emission in Fengqiu, Henan Province, China at the Agricultural Ecological Experiment Station of Chinese Academy of Sciences in November 2021, this method could effectively reduce the dispersion degree of measured concentration trend, as a result, which was better than that of the reported correlation coefficient method. DTW-SVDD algorithm should be of further research value for the application of the analysis of the Big Data of atmospheric composition detection based on grand-based optical remote sensing system. [ABSTRACT FROM AUTHOR]