Differential Absorption LIDAR Signal Denoising using Empirical Mode Decomposition Technique

doi:10.21203/rs.3.rs-2888953/v1

Download PDF

Research Article

Differential Absorption LIDAR Signal Denoising using Empirical Mode Decomposition Technique

https://doi.org/10.21203/rs.3.rs-2888953/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 02 Sep, 2023

Read the published version in Optical and Quantum Electronics →

You are reading this latest preprint version

Differential Absorption Lidar (DIAL) technique is a potential method for the remote detection of hazardous chemicals in the atmosphere. These hazardous chemicals can be due to industrial pollution or may be intentionally released by the terrorist groups or military forces of the enemy country to endanger both the military's personnel and the civilian population's lives. DIAL technique may be used for probing such chemicals from far-off distances of several km ranges and generating an early warning for the response teams. The output of the DIAL system normally consists of three parameters viz. name/class of hazardous chemical detected; its location in terms of distance and the concentration. The maximum standoff distance capability for any Lidar system depends on the signal to noise ratio which is governed by the parameters like atmospheric conditions, Lidar subsystem specifications, noises, etc. SNR is often limited by several noises embedded in the signal from various sources. Due to the presence of noises in the signal, the errors are introduced in the concentration estimation of chemicals from Lidar signal. The methods for improvement of SNR of lidar signal has been often limited by application of conventional denoising techniques like multi-pulse temporal averaging and spatial averaging and further requires nonlinear techniques for noise reduction due to nonlinear behavior of lidar signals. In the present work, Empirical Mode Decomposition (EMD) technique has been implemented on the Lidar signal from Differential Absorption Lidar system. The signal has been denoised and improved SNR is compared with that achieved from temporal averaging and spatial averaging. It was observed that the EMD technique is a better technique as compared to other conventional techniques like multi-pulse temporal averaging and spatial averaging for denoising the signal and increasing Lidar SNR. It is seen that SNR can be improved 4–5 times the original SNR using EMD technique.

Differential Absorption LIDAR

SNR

Denoising methods

Empirical Mode Decomposition

The Lidar technique is a remote sensing technique often used in numerous applications. One of the important applications is the remote detection of hazardous chemicals from long standoff distances[1–2]. It is done by the Differential Absorption Lidar method; in which the released chemical site is targeted by two nearby laser pulses that are sent into the atmosphere. The transmitted laser pulses interact with the chemical species and undergo scattering as well as absorption. Thereafter, an optical telescope gathers the light that has been backscattered. The output is then filtered and finally focused on an optical detector. The output signal may be digitized by a DAQ card which may be PCI based card [3]. By analysis of the return signal changes, detection, identification, and quantification of chemical present can be done by this lidar technique.

The received lidar signal follows the exponentially decaying pattern with respect to time. This is in accordance to the standard lidar expression as given by Eq. 1. The exponential falling signal at longer distances is not that much useful for analysis as this useful signal is fused into noise. Based on the frequency spectrum analysis of the lidar signal, the noises in the lidar backscattered signal are spread across very wide frequency band and at the same time, the useful lidar signal frequency is also present in the same band. Thereby, effectiveness of digital filters for removing the noise to clean the signal and preserving the useful signal at same time is very less as the digital filters work on selection of cut-off frequencies.

Various researchers have reported the many methods for denoising of Lidar signals from different types of Lidar systems. Few of them have used the EMD technique or its variants for improving the Lidar signal. Lee et. al. (2012) has reported noise reduction of dual field of view Lidar signal using Ensemble Empirical Mode Decomposition technique and thereby increasing the maximum observation range of Lidar [4]. Tian et.al., 2015 also reported the effectiveness of the EMD technique in denoising Lidar signal and has described the Clear First Interval Thresholded Empirical Mode Decomposition (EMD-CIIT) technique for the same [5]. Tian et al (2014) have described the criteria based on typical range (TR) and low frequency fraction (LFF) for removing intrinsic mode functions (IMFs) arising from the EMD technique [6]. Li et. al. (2015) proposed the threshold selection in the EMD technique based on multiple iterations called EMD interval thresholding (EMD-IT) [7]. Wu et. al. (2006) compared the performance of low pass noise filter, multiple pulses averaging and EMD technique for the reduction in noise for Lidar backscatter signal [8].

In the present work, three denoising techniques are applied to the backscattered signal for the Differential Absorption Lidar system situated at Laser Science and Technology Centre, Delhi (28.70°, 77.10° E), India. The original signal is denoised using three denoising techniques and SNRs are also estimated. The resultant SNRs are compared with the SNR of the original signal to check the efficacy of three techniques and the EMD technique is recommended to be used for DIAL denoising purposes. Section 2 describes the system used for getting the lidar signal, while section 3 explains the denoising techniques that have been used for current studies. The results have been explained in the section 4.

A mid-infrared wavelength span based Differential Absorption LIDAR (DIAL) system was built at Laser Science & Technology Centre (LASTEC), Delhi, India. The system comprises four major sections i.e. DIAL transmitter, DIAL receiving, DIAL detection section, Data Acquisition, control cum processing section. Figure 1 displays the system's general block diagram. The DIAL transmitter comprises of a mid-infrared tunable pulsed laser and two reflecting mirrors for laser beam bending for coaxial arrangement of transmitter and receiver. The laser pulses of energy of 3 mJ and at a repetition interval of 100 msec and at a particular tuned wavelength tuned between 3000 nm-3450 nm are sent to free space through these two folding mirrors and targeted towards the suspicious hazardous chemical plume. The backscattered laser energy radiations are attained via a DIAL receiver which comprises an optical telescope configured in the Cassegrain with primary mirror of diameter of ~ 200 mm. An interference filter is used at the end of the telescope to cut the radiations outside the wavelength band of 3000nm-3450 nm. The filtered optical signal is further converted into an electrical signal by using an Mercury Cadmium Tellulide (MCT) photovoltaic detector along with a trans-impedance gain preamplifier for necessary amplification.

The analog domain amplified signal transformed into a digital domain signal by data acquisition control and processing unit [9] which is also responsible for various control operations and processing of raw signal for all the necessary output parameters. i.e concentration estimation, range calculation, and identification. The coaxial transmitter, receiver and detection block are integrated on the top of a pan-tilt assembly which meets the requirement of area scanning of suspicious location by the DIAL laser beam with a field of regard between − 60 degrees and + 60 degrees in the horizontal plane and between − 10 and 30 degrees for vertical plane. The scanning can be performed at a travel rate of maximum 10 degrees per second.

For performing DIAL operation for detection, range calculation and concentration estimation of chemical species, pulses tuned to two different wavelengths (ON-wavelength and OFF wavelength) are fired one after another into the free space and pointing towards the direction of hazardous chemical present. The ON-wavelength laser energy has strong absorption by chemical under interrogation whereas the OFF-wavelength shows a weak absorption. This procedure can be extended to a finite number of chemical species in the 3000–3450 nm wavelength band or all the chemical species in the library of the system.

The LIDAR backscattered signal from the free space is estimated by the standard LIDAR equation [10–11]which is written as Eq. 1. From this equation it is inferred that lidar received power profile is an exponentially falling profile with respect to increasing range r.

$$\text{P}({\lambda },\text{r})={\text{P}}_{\text{t}}\left({\lambda }\right){\Delta }\text{r}\frac{{{\pi }{\text{d}}^{2}}_{\text{r}\text{e}\text{c}}}{4{\text{r}}^{2}}{\beta }\left({\lambda },\text{r}\right){\text{O}}_{\text{F}}\left(\text{r}\right){{\eta }}_{1}\left({\lambda }\right){{\eta }}_{2}\left(\text{r}\right){\text{e}}^{-2{\int }_{0}^{\text{r}}{\alpha }\text{d}\text{r}}{\text{e}}^{-2{\int }_{\text{r}1}^{\text{r}2}{\text{{\rm N}}}_{\text{c}}{\sigma }\text{d}\text{r}}$$

where ${P}_{t}\left(\lambda \right)$ is the power of the laser pulse transmitted from transmitter at the selected wavelength $\lambda$; P(λ,r) is the collected power of the laser scattered radiations from range r at the same wavelength; ${d}_{rec}$ is the diameter of the optical telescope for collection of LIDAR backscattered light; O(r) is the overlap factor between laser divergence and receiver’s field of view and has a value between 0 and 1. In the present case, it is considered as 1 as coaxial configuration has been used and laser divergence is less than the receiver field of view; β (λ, r) is the volume backscattering coefficient at the transmitted wavelength; η₁ (λ) is the combined optical efficiency of the transmitter and receiver;${ \eta }_{2}\left(r\right)$ is the probability of return of the laser radiations from distance r; α (λ,r) being the atmospheric volume attenuation coefficient and r is the distance from a laser exit port to a range where an estimated signal is represented.

The ratio of the backscattered received power at the OFF-wavelength and at the ON-wavelength can be used to assess the chemical species' concentration [12] and is given by the following equation

$${\text{N}}_{\text{c}}\times {\Delta }\text{r}=\frac{0.5}{{\Delta }{\sigma }}\times \text{ln}\left(\frac{\text{P}\left({{\lambda }}_{\text{o}\text{f}\text{f}},\text{r}\right)}{\text{P}\left({{\lambda }}_{\text{o}\text{n}},\text{r}\right)}\right)$$

where Δσ is the difference between the chemical's absorption cross sections for strong absorption at ON-wavelength and weak absorption at OFF-wavelength. The powers at transmitted wavelengths ${{\lambda }}_{\text{o}\text{n}}$ & ${{\lambda }}_{\text{o}\text{f}\text{f}}$ are assumed to be equal.

Figure 2 shows the experimental raw signal from the Differential Absorption LIDAR system at a particular wavelength. The signal is shown as the electrical signal from the DIAL detection and DAQ system and is plotted against the line-of-sight range in meters. The reference range is the physical position of the laser exit of the DIAL transmitter. It is clearly visible that the signal is submerged into noise after a certain distance and thus becomes unsuitable for long range DIAL operations. Additionally, at longer ranges, the noise fluctuations are very high and need to be suppressed for improving SNR of LIDAR signal.

Denoising of the backscattered signal is crucial for suppressing the random and unwanted changes in the Differential Absorption LIDAR signal profile and thereby improving the worth of the DIAL signal to a achieve longer detection distance. The Lidar signal for far-off distances has lower signal strength and lower SNR, which poses a limitation on the detection of chemicals at longer distances; thereby introducing errors and false values of estimated concentration. Therefore, the noise suppression in the signal is crucial to reduce these errors and to increase the detection range.

In the reported work following three signal processing techniques have been considered for denoising LIDAR signal,

Spatial average technique

Pulse topulse temporal average technique

Empirical Mode Decomposition (EMD) technique

Spatial averaging of Lidar signal involves the integration and averaging of certain number of neighboring LIDAR data points and the resultant value is overwritten in the current reference data point. The pulse-to-pulse temporal average technique involves the averaging of number backscatter profiles corresponding to each transmitted laser pulse. The temporary variations like gaussian white noise gets cancelled and averaged out after the integration of the number of backscatter profiles. Empirical Mode Decomposition (EMD) technique involves the splitting of the original signal into numerous components which are called Intrinsic Mode Functions (IMF’s). IMFs are estimated in the signal in the order of decreasing frequency of components. The signal's different frequency components can be selectively filtered out and the signal can be de-noised by analyzing the noise component of the signal. The methodology of implementation of these techniques is described in following sections.

3.1 Spatial Average Technique

The spatial average technique [13] is one of the techniques used in the Lidar signal processing field. In order to calculate the average value using the approach, a number of consecutive data points from a lidar profile are combined and averaged. The number of data points can be variable and can be decided based on the user requirement and range resolution requirement. Mathematically, the operation can be expressed as Eq. 3

$$y\left(i\right)=\frac{1}{N}\sum _{j=0}^{N-1}x\left(i-j\right)$$

where x is the array of digital values representing the lidar signal data at different range bins. N denotes the size of the spatial averaging window. The processed resultant value at a range bin is obtained by averaging N data points in a shifting window method.

The spatial averaging technique replaces each data value at a range bin with the averaged surrounding value in order to smooth out the sudden fast changes in the lidar signal profile. The procedure lasts for all the data values at different range bins. As the spatial averaging technique only performs the smoothening operation on the signal; it is not able to reject the non-useful transient values and those negative voltage values caused by the noise. Another disadvantage is that as the higher number of pulses are used for the spatial averaging, the necessary useful signal amplitude is also suppressed along with the reduction of noise levels. Although the SNR may increase in a good manner, but the atmospheric characteristics may get camouflaged when moving average is executed. This is a drawback of the spatial averaging technique due to which this technique can be used for only small number of averaging points.

3.2 Pulse to Pulse Temporal Average Technique

Pulse to Pulse temporal average technique [13] is the most broadly used technique for enhancement of SNR of the Lidar signals. This technique involves recording in a memory the lidar return signal data values at every range bin associated with a transmitted laser pulse. Similar to this, sets of data corresponding to the required pulses to be summed are acquired and recorded in other memory locations. The processed resultant value at a range bin is obtained by adding and averaging each individual data bin value at all pulses. Eq. 4 illustrates the mathematical expression or this methodology.

$$\text{y}\left(j\right)=\frac{1}{\text{K}}{\sum }_{k=0}^{K}{\text{x}}_{\text{k}}\left(\text{j}\right)$$

Where x₁(j) is the Lidar signal value for first backscatter profile at j_th range bin. For calculating the resultant value y(j) at j^th bin, each pulse's associated Lidar data value at the jth bin is summed, and then the average is calculated.

The advantage of temporal averaging is that the signal values are not lost in the averaging process and range resolution is also not compromised. Disadvantages of temporal averaging include requirement of large integration time for low prf systems and change in atmospheric phenomena in between the averaging process during large integration times.

3.3 Empirical Mode Decomposition Technique

Empirical Mode Decomposition (EMD) technique was first described by Huang et. al. (1998) [14]. The analysis of signals that are non-stationary and nonlinear was suggested using this method. EMD is the signal data-adaptive method in which the base function is determined in real time from the signal itself. It is an iterative method in which the procedure is repeated until the convergence criteria are satisfied. The essence of the EMD technique lies in the fact that a non-stationary signal can be fragmented into a limited number of functions called Intrinsic Mode Functions (IMFs) and one can approximate the original signal by adding these IMFs as given in Eq. 5.

$$y[n]=\sum\limits_{{k=1}}^{Z} {{I_k}} [n]+r[n]$$

Where y[n] is discrete time signal of length N under measurement, ${I_k}$ is the k^th IMF, Z is the total number of IMFs and r[n] is the residue at the end of the process.

The calculated function is considered as IMF if the following requirements are fulfilled [15]:

The total number of extreme locations (maxima and minima) and the total number of zero crossings of the signal in the time domain should be equal, or there should be a maximum difference of one in the number of positions.

The average of the envelope of all maxima points and envelope of all minima points should be zero.

The sifting process, an iterative procedure used to calculate IMFs and residue, is explained as follows.

1) Look for maxima and minima and their locations in time series discrete signal y[n].

2) Generate the maxima envelope S_max[n] by joining all the maxima points using some interpolation method. Repeat the similar procedure to generate minima envelope S_min[n]

3) Determine the average of maxima envelope and minima envelope as following

S_mean1[n]= $\frac{{{S_{\hbox{max} }}[n]+{S_{\hbox{min} }}[n]}}{2}$ (6)

4) Determine the difference d[n] = y[n]-S [n], checked for the criteria of being IMF. If it

a. satisfies both the conditions, then it is denoted as IMF (I₁[n]). Further residue is calculated as r₁[n] = y[n]-I₁[n].

b. does not satisfy the IMF criteria, then consider d₁[n] as signal and compute d₁₁[n]

d₁₁[n] = d₁[n]-S_mean11 [n] (7)

repeat the above procedure until the difference satisfies the condition of being IMF.

5) The procedure in steps 1 to 4 is repeated by considering residue as calculated in step 4a as the signal.

6) The finite numbers of IMFs are computed in this way and the procedure is terminated as per stopping criteria of EMD, which indicates that there are no more extrema (maxima/minima) or having one extremain the residue.

Every IMF calculated using the aforementioned method represents oscillations at a specific frequency. Following IMFs indicate lower frequency components in decreasing order of frequency, with the first IMF representing the greatest frequency oscillation present in the signal. Effectively, the residue at the end of iterations does not have any oscillations.

The signal from the experimental Differential Absorption Lidar system assembled at LASTEC has been used for denoising studies. For this purpose, laser pulses with the energies of 3 mJ were sent to the open atmosphere using the DIAL transmitter at an elevation angle of 10 degrees from the system horizontal plane. The backscattered signal received by the DIAL receiver was digitized and recorded with a range resolution of 5m. The reference time of transmission of laser pulse at t = 0 sec is taken as a starting range of 0 m. The signal was recorded for a maximum slant range of 3000m.The backscattered signal was used to calculate the IMFs using the procedure and equations defined in the previous sections. This is a well-known fact that the acquired Lidar signal is a nonlinear and non-stationary signal wherein the frequency of noises is not uniform throughout the time series or total range. Figure 2 displays the recorded Lidar signal and as it can be observed, the signal dissolves into the noise after a given distance. The calculated IMFs of the signal that was shown in Fig. 2 earlier are displayed in Fig. 3 where a total of 9 IMFs (i.e. IMF1 to IMF9) are shown and these IMFs satisfy Eq. 5. The first IMF (i.e. IMF1) contains the highest frequency oscillation present in the signal while the last IMF (IMF9) represents the lowest frequency oscillation of the Lidar backscattered signal. The power spectral density for each of the calculated IMF (IMF1-IMF9) is also estimated and is shown in Fig. 4. It is seen that the lower frequency components or IMF7-IMF8 have a larger share in the signal power as compared to that of higher frequency components (IMF1-IMF6). So, it is imperative that the noise is contained in high frequency components and the Lidar signal can be de-noised by subtracting high order IMFs which represent the high order oscillations from the original signal. It is also understood that the changes in dynamics of the atmosphere will not be at very fast phenomena and cannot be represented by such high frequency components in the signal.

After the necessary IMFs have been eliminated, the de-noised signal is recovered and compared to the original signal, as shown in Fig. 5. It is evident that the Lidar signal is now significantly smoother and less irregular after noise has been removed using EMD. The quantification for the improvement in the signal is done on the basis of Signal to Noise Ratio of the signal.

An estimated SNR with respect to range is also plotted in Fig. 6, where SNR profile of original Lidar signal and denoised Lidar signal are shown in single plot. It is observed that the SNR is highly improved for near ranges and the signal can travel farther to a longer distance before it merges with the background signal. The SNR values at different distance values from the DIAL instrument are tabulated in table-1. From the table, it can be inferred that EMD is a useful and efficient technique for improving the Differential Absorption Lidar signal.

The raw signal was also processed with the conventional spatial averaging technique which is also called the moving average technique, wherein a window of the number of samples equal to the number of window width is moved throughout the signal and the signal is smoothened and denoised spatially within neighbouring points. For current sudies, the number of data points considered for the spatial averaging is 5. The range resolution of the raw signal is 5 m and data points upto 25 m are used for averaging window. Any number of data points can be averaged but using more data points will lead to poor range resolution. The original and denoised lidar signal using this technique are shown in Fig. 7.

As it is seen that there is no much improvement in the signal after application of spatial averaging but the fluctuations are reduced up to some extent. In order to further reduce the fluctuations, the number of data points for averaging should be increased but this will lead to loss of signal values as well as compromise the range resolution. In Table 1, the SNR of the denoised signal using spatial averaging is also reported. The SNR values of the raw signal and denoised signal are plotted in Fig. 8. It is seen that there is a loss of signal and as a result signal to noise ratio at various range bins is also reduced while there is no substantial gain on the factor of removing unwanted noise with fluctuations. Hence, spatial averaging is not a very useful technique for removing noise in case fluctuations are not very sharp while it can give some advantages for temporary sharp speckles. The noise fluctuations may also take the signal to negative polarity and it becomes impractical to use the negative signal to calculate the concentration in the case of Differential Absorption Lidar signal.

The raw signal was also processed and denoised using the temporal averaging technique. The signal was averaged for 10 number of pulses.

The comparison plot between the raw signal and the denoised signal using the temporal averaging technique is shown in Fig. 9. The processed signal SNR is plotted against raw signal SNR and is shown in Fig. 10. It is seen that the temporal average is able to remove some of the negative noises which may vary from pulse to pulse and no loss in the signal is noticed. The SNR has also been improved in different ranges as seen and tabulated in the table-1.It is noticed that the temporal averaging is also not able to remove the low frequency noise fluctuations and there are a lot of lobes of low frequency in the long range signal pattern. These lobes will give indications of some false clouds with high values of concentrations if we scan the atmosphere for different chemicals and estimate the concentration of clouds. Hence, pulse to pulse averaging technique is not very effective in the removal of noise which is time-varying in non-stationary signals like Lidar.

Table 1

Comparision Table for SNR Values of Original and Denoised Lidar Signal at Different Ranges Using Three Techniques
Range		30 m	40m	50 m`	60 m	70m	80m	100m	110m
SNR Value	Original	20.4	18.9	12.7	7.9	5.2	3.6	1.7	0.9
	EMD Denoised	50.6	52.2	46.7	35.3	20.2	7.1	2.6	1.9
	Spatial Averaging	18.3	13.5	8.7	5.7	3.7	2.5	1.2	0.6
	Temporal Averaging	35.5	32.2	21.2	12.6	8.3	5.9	2.1	0.7

The SNR values from denoised signals using three different noise reduction techniques are given in table-1. It is clearly seen that the EMD denoising technique is best out of three techniques used for denoising of Lidar signal.

The experimental Lidar signal has been recorded using the assembled Differential Absorption Lidar system. It is very much essential to remove low frequency fluctuations and the negative signal from the Lidar signal profile as it gives an impression of false chemical clouds and topographic targets. The conventional signal processing techniques i.e. spatial averaging and temporal averaging is compared with the EMD technique for denoising Lidar signal. It is seen that the EMD technique is much better in removing time-varying noise of Lidar signals and reducing the errors in information processing from Lidar signals.

Acknowledgments

Authors wish to thank Director, Instruments Research & Development Establishment (IRDE) for his constant support and encouragement for executing the current research work.

This work is supported by the Ministry of Electronics and Information Technology, Government of India under the Visvesvaraya Ph.D. scheme.

Competing Interests

The authors have no relevant financial or non-financial interests to disclose.

Author Contributions

Study conception and design were done by M K Jindal, and M Mainuddin. Material preparation, data collection and analysis were performed by M K Jindal, and S Veerabuthiran. Analysis was performed by M K Jindal, M Mainuddin and N Jindal. The first draft of the manuscript was written by M K Jindal and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Data Availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Jindal, M. K., Mainuddin, M., Veerabuthiran, S., Razdan, A. K.:. Laser-based systems for standoff detection of CWA: a short review. IEEE Sensors Journal, 21(4), 4085-4096 (2020)
Jindal, M. K., Veerabuthiran, S., Razdan, A. K.: Integrated path DIAL for standoff detection of acetone vapors under topographic target condition. Optics & Laser Technology, 143, 107299 (2021)
Mainuddin M., Beg, M. T., Tyagi, R. K., Rajesh, R., Singhal, G. Dawar, A. L.: Optical spectroscopic based in-line iodine flow measurement system–an application to COIL. Sensors and Actuators B: Chemical, 109(2), 375-380 (2005). doi: 10.1016/j.snb.2005.01.004.
Li, J., Gong, W., Ma, Y.: Atmospheric lidar noise reduction based on ensemble empirical mode decomposition. ISPAr, 39, 127-129 (2012).
Tian, P., Zhang, L., Cao, X., Yi, N.: The Application of EMD-CIIT Lidar Signal Denoising Method in Aerosol Detection. Procedia engineering, 102, 1233-1237 (2015).
Tian, P., Cao, X., Liang, J., Zhang, L., Yi, N., Wang, L.,Cheng, X.: Improved empirical mode decomposition based denoising method for lidar signals. Optics Communications, 325, 54-59 (2014).
Li, M., Jiang, L. H., Xiong, X. L.: A novel EMD selecting thresholding method based on multiple iteration for denoising LIDAR signal. Optical Review, 22(3), 477-482 (2015).
Wu, S., Liu, Z., Liu, B.: Enhancement of lidar backscatters signal-to-noise ratio using empirical mode decomposition method. Optics communications, 267(1), 137-144 (2006).
Jindal, M. K., Mainuddin, M., Veerabuthiran, S.: Differential absorption lidar data acquisition and control system for remote detection of trace chemicals including methane and thiodiglycol. Review of Scientific Instruments, 94(2), 024705 (2023).
Veerabuthiran, S., Dudeja, J. P.: Parametric Study of Differential Absorption Lidar System for Monitoring Toxic Agents in the Atmosphere. Defence Science Journal-New Delhi, 57(5), 755 (2007).
Veerabuthiran, S., Razdan, A. K., Jindal, M. K., Prasad, G.: Open field testing of mid IR DIAL for remote detection of thiodiglycol vapor plumes in the topographic target configuration. Sensors and Actuators B: Chemical, 298, 126833 (2019).
Veerabuthiran, S., Razdan, A. K., Jindal, M. K., Sharma, R. K., Sagar, V.: Development of 3.0–3.45 µm OPO laser based range resolved and hard-target differential absorption lidar for sensing of atmospheric methane. Optics & Laser Technology, 73, 1-5 (2015).
Menyuk, N., Killinger, D. K., Menyuk, C. R.: Limitations of signal averaging due to temporal correlation in laser remote-sensing measurements. Applied Optics, 21(18), 3377-3383 (1982).
Huang, N. E., Shen, Z., Long, S. R., Wu, M. C., Shih, H. H., Zheng, Q., Yen, N.C., Tung, C.C.,Liu, H. H.: The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proceedings of the Royal Society of London. Series A: mathematical, physical and engineering sciences, 454(1971), 903-995 (1998).
Boudraa, A. O., Cexus, J. C.: EMD-based signal filtering. IEEE transactions on instrumentation and measurement, 56(6), 2196-2202 (2007).

No competing interests reported.

Download PDF

Journal Publication

published 02 Sep, 2023

Read the published version in Optical and Quantum Electronics →

Editorial decision: Major revision
24 Jun, 2023
Reviews received at journal
08 Jun, 2023
Reviewers agreed at journal
30 May, 2023
Reviewers agreed at journal
22 May, 2023
Reviewers invited by journal
19 May, 2023
Editor assigned by journal
06 May, 2023
Submission checks completed at journal
06 May, 2023
First submitted to journal
03 May, 2023

You are reading this latest preprint version

Differential Absorption LIDAR Signal Denoising using Empirical Mode Decomposition Technique

Status:

Journal Publication

Version 1

Abstract

Figures

1. Introduction

2. Differential Absorption LIDAR (DIAL) System

3. Improving the SNR of Differential Absorption Lidar signal

3.1 Spatial Average Technique

3.2 Pulse to Pulse Temporal Average Technique

3.3 Empirical Mode Decomposition Technique

4. Results and Discussion

Conclusion

Declarations

References

Additional Declarations

Status:

Journal Publication

Version 1