It is a challenging problem to monitor atmospheric CO₂ leakage due to the complex variation of the atmosphere background. In this research, a new CO₂ leakage identification method in the atmosphere based on the complex time–frequency spectrum of atmospheric CO₂ variation was proposed. First, the complex continuous wavelet transform (CWT) was utilized to analyze the experimental data without and with CO₂ leakage. It was found that CWT could provide distinguished features for atmospheric CO₂ leakage by calculating the time–frequency spectrum and modulus of CWT for the cases with a leakage rate from 5 to 25 m³/h. Further, the atmospheric CO₂ concentration and CO₂ variation rate were compared to recognize abnormal leakage. The results indicated that the CWT spectrum of the CO₂ variation rate performed better than that of concentration. Moreover, the CWT spectrum of the atmospheric CO₂ variation rate with the real-valued wavelet function was also utilized to recognize CO₂ leakage. The tests showed that the CWT spectrum with the complex Morlet wavelet demonstrated a more obvious and wider hot spot than that with the real-valued Morlet wavelet. In addition, a pretreatment method with principal component analysis (PCA) was applied to extract the features of original monitoring signals. It was proved that more obvious abnormal signals in the time–frequency spectrum and modulus variation PCA–CWT method could be captured than that from the original CWT analysis, even for a small leakage. Therefore, it is a feasible method to monitor and recognize atmospheric CO₂ leakage with the complex CWT of the CO₂ variation rate in the atmosphere combined with PCA processing.

中文翻译：

---

基于大气CO2变化复杂时频谱的CO2泄漏识别方法

由于大气背景的复杂变化，监测大气CO ₂泄漏是一个具有挑战性的问题。在这项研究中，一个新的CO ₂基于大气CO的复杂的时间-频率频谱中的大气泄漏识别方法₂的变化中提出的。首先，利用复连续小波变换（CWT）来分析没有和有CO ₂泄漏的实验数据。通过计算CWT的时频频谱和模数，发现CWT可以为大气CO ₂泄漏提供显着特征，在泄漏率为5至25 m ³ /h的情况下。此外，大气 CO ₂浓度和CO ₂变化率进行比较以识别异常泄漏。结果表明，CO ₂变化率的CWT谱表现优于浓度的CWT谱。此外，还利用具有实值小波函数的大气 CO ₂变化率的 CWT 谱来识别 CO ₂泄漏。测试表明，复Morlet小波的CWT谱比实值Morlet小波表现出更明显和更宽的热点。此外，应用主成分分析（PCA）预处理方法提取原始监测信号的特征。经证明，即使是很小的泄漏，时频频谱和模量变化 PCA-CWT 方法中的异常信号也比原始 CWT 分析中捕获的异常信号更明显。因此，将大气中CO ₂变化率的复杂CWT结合PCA 处理是监测和识别大气CO ₂泄漏的一种可行方法。