## Linear segmentation algorithm for detecting layer boundary with lidar |

Optics Express, Vol. 21, Issue 22, pp. 26876-26887 (2013)

http://dx.doi.org/10.1364/OE.21.026876

Acrobat PDF (1137 KB)

### Abstract

Abstract: The automatic detection of aerosol- and cloud-layer boundary (base and top) is important in atmospheric lidar data processing, because the boundary information is not only useful for environment and climate studies, but can also be used as input for further data processing. Previous methods have demonstrated limitations in defining the base and top, window-size setting, and have neglected the in-layer attenuation. To overcome these limitations, we present a new layer detection scheme for up-looking lidars based on linear segmentation with a reasonable threshold setting, boundary selecting, and false positive removing strategies. Preliminary results from both real and simulated data show that this algorithm cannot only detect the layer-base as accurate as the simple multi-scale method, but can also detect the layer-top more accurately than that of the simple multi-scale method. Our algorithm can be directly applied to uncalibrated data without requiring any additional measurements or window size selections.

© 2013 Optical Society of America

## 1. Introduction

3. M. Feiyue, G. Wei, and M. Yingying, “Retrieving the aerosol lidar ratio profile by combining ground- and space-based elastic lidars,” Opt. Lett. **37**(4), 617–619 (2012). [CrossRef] [PubMed]

5. J. B. Senberg, A. Ansmann, J. M. Baldasano, D. Balis, C. B. Ckmann, B. Calpini, A. Chaikovsky, P. Flamant, A. Hgrd, and V. Mitev, “EARLINET: a European aerosol research lidar network,” in *Advances in laser remote sensing,* (Selected Papers of the 20th International Laser Radar Conference, 2001), pp. 155–158.

*P*(

*r*), or its derivative signals such as the range-corrected signal,

*X*(

*r*) [i.e.,

*P*(

*r*)

*r*

^{2}], the atmospheric extinction coefficient or attenuated backscatter coefficients and so on [6

6. F. Mao, W. Gong, S. Song, and Z. Zhu, “Determination of the boundary layer top from lidar backscatter profiles using a Haar wavelet method over Wuhan, China,” Opt. Laser Technol. **49**, 343–349 (2013). [CrossRef]

7. F. Rocadenbosch, M. Sicard, M. N. M. Reba, and S. Tomas, “Morphological tools for range-interval segmentation of elastic lidar signals,” in IEEE International Geoscience and Remote Sensing Symposium(IGARSS), 2007), 4372~4375. [CrossRef]

10. Z. Wang and K. Sassen, “Cloud type and macrophysical property retrieval using multiple remote sensors,” J. Appl. Meteorol. **40**(10), 1665–1682 (2001). [CrossRef]

11. F. Mao, W. Gong, and C. Li, “Anti-noise algorithm of lidar data retrieval by combining the ensemble Kalman filter and the Fernald method,” Opt. Express **21**(7), 8286–8297 (2013). [CrossRef] [PubMed]

12. Y. Morille, M. Haeffelin, P. Drobinski, and J. Pelon, “STRAT: An automated algorithm to retrieve the vertical structure of the atmosphere from single-channel lidar data,” J. Atmos. Ocean. Technol. **24**(5), 761–775 (2007). [CrossRef]

13. J. Gaumet, J. Heinrich, M. Cluzeau, P. Pierrard, and J. Prieur, “Cloud-base height measurements with a single-pulse erbium-glass laser ceilometer,” J. Atmos. Ocean. Technol. **15**(1), 37–45 (1998). [CrossRef]

8. S. R. Pal, W. Steinbrecht, and A. I. Carswell, “Automated method for lidar determination of cloud-base height and vertical extent,” Appl. Opt. **31**(10), 1488–1494 (1992). [CrossRef] [PubMed]

9. D. M. Winker and M. A. Vaughan, “Vertical distribution of clouds over Hampton, Virginia observed by lidar under the ECLIPS and FIRE ETO programs,” Atmos. Res. **34**(1-4), 117–133 (1994). [CrossRef]

14. S. A. Young, “Analysis of lidar backscatter profiles in optically thin clouds,” Appl. Opt. **34**(30), 7019–7031 (1995). [CrossRef] [PubMed]

15. F. Mao, W. Gong, J. Li, and J. Zhang, “Cloud detection and coefficient retrieve based on improved differential zero-crossing method for Mie lidar,” Acta Opt. Sin. **30**(11), 3097–3102 (2010). [CrossRef]

16. F. Mao, W. Gong, and Z. Zhu, “Simple multiscale algorithm for layer detection with lidar,” Appl. Opt. **50**(36), 6591–6598 (2011). [CrossRef] [PubMed]

*X*(

*r*) ≤

*X*(

*r*

_{b}) [8

8. S. R. Pal, W. Steinbrecht, and A. I. Carswell, “Automated method for lidar determination of cloud-base height and vertical extent,” Appl. Opt. **31**(10), 1488–1494 (1992). [CrossRef] [PubMed]

15. F. Mao, W. Gong, J. Li, and J. Zhang, “Cloud detection and coefficient retrieve based on improved differential zero-crossing method for Mie lidar,” Acta Opt. Sin. **30**(11), 3097–3102 (2010). [CrossRef]

9. D. M. Winker and M. A. Vaughan, “Vertical distribution of clouds over Hampton, Virginia observed by lidar under the ECLIPS and FIRE ETO programs,” Atmos. Res. **34**(1-4), 117–133 (1994). [CrossRef]

17. M. A. Vaughan, K. A. Powell, D. M. Winker, C. A. Hostetler, R. E. Kuehn, W. H. Hunt, B. J. Getzewich, S. A. Young, Z. Liu, and M. J. McGill, “Fully automated detection of cloud and aerosol layers in the CALIPSO lidar measurements,” J. Atmos. Ocean. Technol. **26**(10), 2034–2050 (2009). [CrossRef]

18. J. R. Campbell, K. Sassen, and E. J. Welton, “Elevated cloud and aerosol layer retrievals from micropulse lidar signal profiles,” J. Atmos. Ocean. Technol. **25**(5), 685–700 (2008). [CrossRef]

*X*(

*r*), which representation are subsequently used in cloud detection [19

19. W. Gong, F. Mao, and S. Song, “Signal simplification and cloud detection with an improved Douglas-Peucker algorithm for single-channel lidar,” Meteorol. Atmos. Phys. **113**(1-2), 89 (2011). [CrossRef]

*X*(

*r*). False positive rejection criteria are used based on

*X*(

*r*) with varying thresholds. We first search for layer base and top based on a linear fitting of the segmentations. In order to enhance the robustness, we employ an “extrapolate into the layer” strategy. This will refine our estimate of layer base and top as both simulated and real data with various noise levels are designed as a stress test for the detection scheme. We find good consistency between layer-base heights retrieved by both the linear segment algorithm and the simple multi-scale method. However, the tops detected by the linear segment algorithm are much accurate than that detected by the simple multi-scale method. Our algorithm can be directly applied to uncalibrated data as no additional measurements or models are required and no window size needs to be selected.

## 2. Principles and methods

*P*(

*r*) received by the detector can be written as the following [2]:where

*r*is the range,

*C*is lidar constant,

*G*(

*r*) represents the overlap factor [20

20. W. Gong, F. Mao, and J. Li, “OFLID: Simple method of overlap factor calculation with laser intensity distribution for biaxial lidar,” Opt. Commun. **284**(12), 2966–2971 (2011). [CrossRef]

*β*(

_{m}*r*) and

*β*(

_{p}*r*) are the molecule and particle backscatter coefficient respectively,

*α*(

_{m}*r*) and

*α*(

_{p}*r*) are the molecule and particle extinction coefficient respectively, and the noise

*e*(

*r*) is considered as Gaussian. Equation (1) illustrates that the lidar signal will increase considerably when a lidar laser pulse is incident into an optically thick layer. A diagram of the layer detection scheme based on

*X*(

*r*) (Fig. 1) shows that the scheme comprises four key steps that can be described as:

- (1) Signal segmentation: The segmentation is based on
*X*(*r*) with a threshold array of 6σ·*r*^{2}where σ is the standard deviation of the background noise, and can be estimated using the signal at a distant range where noise is prevalent thus no lidar return. The break-bins outputted by the segmentation can be candidates of base, peak and top of layers. - (2) Primary layer base and top selection: We search for the primary layer base when the fitted slope is positive, and select the primary top when fitted slope is negative and larger than or comparable to that below the primary base.
- (3) False positive rejection: When considering the envelope of the noise
*e*(*r*)*r*^{2}is ± 3σ·*r*^{2}, all detections whose signal difference is less than 3σ·*r*_{p}^{2}+ 3σ·*r*_{b}^{2}are considered as a false positive and subsequently removed,*r*_{p}and*r*_{b}are the range of the peak and base, respectively. - (4) Refine the layer base and top: To enhance the robustness, we employ an “extrapolate into the layer” strategy to refine the primary layer base and top based on the linear fitting of the range-corrected signal of the two nearest “clear air” segmentations.

### 2.1 Signal segmentation

22. E. Keogh, S. Chu, D. Hart, and M. Pazzani, “An online algorithm for segmenting time series,” in (Proceedings 2001 IEEE International Conference on Data Mining, 2001), 289–296. [CrossRef]

19. W. Gong, F. Mao, and S. Song, “Signal simplification and cloud detection with an improved Douglas-Peucker algorithm for single-channel lidar,” Meteorol. Atmos. Phys. **113**(1-2), 89 (2011). [CrossRef]

*n*range bins

*X*(

*r*

_{1}),

*X*(

*r*

_{2}), ... and

*X*(

*r*). One of the major challenges of the segmentation is being able to determine the threshold array, which shall be discussed later. If the threshold array,

_{n}*d*(

_{thr}*r*), is given, we can subsequently select a bin as a break-bin according to the intensity difference between the

*X*(

*r*) and

*X*(

_{seg}*r*). The

*X*(

_{seg}*r*) is derived by linear interpolation with the

*X*(

*r*) of the beginning and ending bins. The absolute intensity difference can be written as:

*r*, the corresponding range bin will be selected as a break-bin if

_{m}*d*(

*r*) exceeds the threshold value

_{m}*d*(

_{thr}*r*). Then, recursive segmenting calculations will be carried out based on the two sub-signals segmented by this break-bin until the threshold is no longer exceeded.

_{m}*d*

_{thr}(

*r*) is one of the biggest challenges in the segmentation method. Because the SNR of

*X*(

*r*) decreases with range rapidly, it is not advisable to segment based on a constant threshold array. We previously defined the error array,

*e*(

*r*) as the difference between the

*X*(

*r*) and the denoised signal

*X*(

_{de}*r*) in [19

19. W. Gong, F. Mao, and S. Song, “Signal simplification and cloud detection with an improved Douglas-Peucker algorithm for single-channel lidar,” Meteorol. Atmos. Phys. **113**(1-2), 89 (2011). [CrossRef]

*e*(

*r*). One disadvantage of this strategy is that the

*d*(

_{thr}*r*) is very sensitive to noise, and the signal will be over-segmented as shown in Fig. 2(a).Furthermore, the segment has to be performed on a denoised signal rather than an original one, but a signal can be distorted by denoising technology.

_{m}*X*(

*r*) can be described as:where

*S*(

*r*) is the pure raw signal. It shows that the noise level of

*X*(

*r*) is equivalent to

*e*(

*r*)

*r*

^{2}. Theoretically, the corresponding envelope of the noise

*e*(

*r*)

*r*

^{2}is ± 3σ·

*r*

^{2}. Furthermore, it can be shown that the curves of + 3σ·

*r*

^{2}and −3σ·

*r*

^{2}envelop the detected noise in Fig. 2(a) and the signal (primarily noise) above 4 km in Fig. 2(b), respectively. Thus, a threshold array

*d*(

_{thr}*r*) can be computed as:

*r*

^{2}with

*X*(

*r*) can be seen in Fig. 2(b). The segmentation not only retains “marked-change bins” as break-bins, but also neglects most of the bins in the “no-change range” such as all the range bins above 4 km. The segmentation can avoid over-segmentation, which is beneficial for refining the layer boundary and removing any over-detections.

*r*

_{1},

*r*

_{2}, ...,

*r*” to “1:n” for convenience and outline the pseudo-code of the recursive algorithm as follows:

_{n}Function [break_bins] = linear_segment [X(1:n), σ] | ||

// obtain X(1:_{seg}n) by interpolating X(1) and _{seg}X(_{seg}n) linearly: | ||

X(1:_{seg}n) = linear_interpolate[X(1), X(n)]; | ||

// obtain the absolute difference between X and X._{seg} | ||

d(1:n) = |X(1:n) -X(1:_{seg}n) |; | ||

// find d and its corresponding range bin._{m} | ||

[d, _{m}d _bin] = max[_{m}d(1:n)]; | ||

// obtain threshold array | ||

d(1:_{thr}n) = 6·σ·r^{2}(1:n); | ||

If d > _{m}d (_{thr}d _bin) // segment the signal _{m}X(1:n) with given threshold. | ||

// recursively segment the left part if necessary. | ||

[break_bins1] = linear_segment [X(1: d_bin), σ];_{m} | ||

// recursively segment the right part if necessary. | ||

[break_bins2] = linear_segment [X(d_bin + 1:end), σ];_{m} | ||

// combine the break bins of the left and right parts. | ||

break_bins = [break_bins1(1:end-1), break_bins2(1:end)]; | ||

Else | ||

// the break bins are the first and end bins of X(1:n) if d < _{m}d(_{thr}d _bin)_{m} | ||

break_bins = [X(1), X(n)]; | ||

End if | ||

End Function |

*X*(

_{fit,k}*r*) =

*a*+

_{fit,k}·r*b*, where

_{fit,k}*a*and

_{fit,k}*b*are derived from all the

_{fit,k}*X*(

*r*) in the

*k-*th sub-signal (i.e., the signal between

*k*and

*k*+ 1 break-bins) by the least-squares method. The fitted slope

*a*(

_{fit}*r*) will be used for the detection of the layer base and top and is further discussed in the next section.

### 2.2 Selection primary layer base and top

7. F. Rocadenbosch, M. Sicard, M. N. M. Reba, and S. Tomas, “Morphological tools for range-interval segmentation of elastic lidar signals,” in IEEE International Geoscience and Remote Sensing Symposium(IGARSS), 2007), 4372~4375. [CrossRef]

10. Z. Wang and K. Sassen, “Cloud type and macrophysical property retrieval using multiple remote sensors,” J. Appl. Meteorol. **40**(10), 1665–1682 (2001). [CrossRef]

10. Z. Wang and K. Sassen, “Cloud type and macrophysical property retrieval using multiple remote sensors,” J. Appl. Meteorol. **40**(10), 1665–1682 (2001). [CrossRef]

**113**(1-2), 89 (2011). [CrossRef]

*a*(

_{fit}*r*) becomes larger than zero. A second issue of layer detection is selecting the layer top. We initially estimate the layer top as the height of the first bin where

*X*(

*r*) ≤

*X*(

*r*

_{b}) above the base in the same manner as other studies [8

8. S. R. Pal, W. Steinbrecht, and A. I. Carswell, “Automated method for lidar determination of cloud-base height and vertical extent,” Appl. Opt. **31**(10), 1488–1494 (1992). [CrossRef] [PubMed]

### 2.3 False positive rejection

*P*(

*r*) has an “inverse square law” (1/

*r*

^{2}) effect,

*X*(

*r*) is used for rejecting false positive. As mentioned above, the noise level of

*X*(

*r*) is equal to

*e*(

*r*)

*r*

^{2}, and is therefore not recommended to use a constant threshold in order to reject the false positive. Thus, the rejection criterion is based upon varying threshold in

*X*(

*r*). By defining

*X*(

_{s}*r*) and

_{p}*X*(

_{s}*r*) as pure range-corrected signals of the peak and base, respectively, the signal difference between layer peak and base can be written as:

_{b}*X*(

_{s}*r*)−

_{p}*X*(

_{s}*r*) ± (3σ·

_{b}*r*

_{p}^{2}+ 3σ·

*r*

_{b}^{2}). Furthermore,

*X*(

*r*) decrease along with range for clear air. Thus, for a false positive which is due to noise effect, the ∆ of noise variations should be no more than

*as a threshold to separate the genuine layer from false positive which are due to noise effect. Based on the threshold, all detections where ∆ is less than ∆*

_{thr}*are considered as false positive and subsequently removed.*

_{thr}*X*(

_{s}*r*), and we use

*X*(

_{up}*r*) and

*X*(

_{low}*r*) to denote the upper envelope [i.e.

*X*(

_{s}*r*) + 3σ·

*r*

^{2}] and lower envelope [i.e.

*X*(

_{s}*r*)−3σ·

*r*

^{2}] of a noised range-corrected signal, respectively. Theoretically, according the defining of ∆

*, the algorithm will be triggered to discriminate a layer when*

_{thr}*X*(

*r*) is larger than

*X*(

_{up}*r*). The algorithm is able to successfully detect the layer as in the case in Fig. 3(a), because the

*X*(

_{low}*r*) is larger than the

_{p}*X*(

_{up-extrap}*r*), which is the extrapolation of the

_{p}*X*(

_{up}*r*) of the clear air lower than 4 km. However, the algorithm may fail to detect the layer for the case in Fig. 3(b), because the

*X*(

_{low}*r*) is less than

_{p}*X*(

_{up-extrap}*r*). In this situation, the possibility of the layer to be detected is equal to the sum of the possibilities

_{p}*X*(

_{layer}*r*) of every range bin larger than

*X*(

_{up-extrap}*r*). Thus, the larger of the geometric thickness of the layer, the larger of the possibility of the layer to be detected. Because the pattern of a layer is variable in the real world, it is difficult to suggest a constant possibility to be a reference for any given detection. Under the condition of Fig. 3(a), once a layer signal with random noise is given, a layer base will be detected in the range of 4-4.25 km. However, a layer base can likely be detected anywhere in the range of 4-5 km for a signal under the condition of Fig. 3(a).

*G*(

*r*) as a unit as well as assuming

*C*·

*β*(

_{m}*r*)·

_{b}*T*

^{2}(

*r*) is equal to

_{b}*C*·

*β*(

_{m}*r*)·

_{p}*T*

^{2}(

*r*), the difference of

_{p}*X*(

*r*) and

_{p}*X*(

*r*) can be written as:where

_{b}*T*

^{2}(

*r*) is the two-way transmittance at range

*r*. The range of ∆ is

*C*·

*β*(

_{p}*r*

_{p})·

*T*

^{2}(

*r*) ± (3σ·

_{p}*r*

_{p}^{2}+ 3σ·

*r*

_{b}^{2}). Because we will judge a detection to be a layer when ∆ is larger than ∆

*, theoretically, we have no difficulty in detecting a layer where*

_{thr}*β*(

_{p}*r*) meet the following condition:

_{p}### 2.4 Refining layer base and top

## 3. Results and discussion

### 3.1 Testing with simulated signals

### 3.2 Testing with real signals

*X*(

*r*

_{p}) to

*X*(

*r*

_{b}) is larger than four, otherwise it is considered as an aerosol layer based on previous studies [12

12. Y. Morille, M. Haeffelin, P. Drobinski, and J. Pelon, “STRAT: An automated algorithm to retrieve the vertical structure of the atmosphere from single-channel lidar data,” J. Atmos. Ocean. Technol. **24**(5), 761–775 (2007). [CrossRef]

16. F. Mao, W. Gong, and Z. Zhu, “Simple multiscale algorithm for layer detection with lidar,” Appl. Opt. **50**(36), 6591–6598 (2011). [CrossRef] [PubMed]

**113**(1-2), 89 (2011). [CrossRef]

23. S. Burton, R. Ferrare, M. Vaughan, A. Omar, R. Rogers, C. Hostetler, and J. Hair, “Aerosol classification from airborne HSRL and comparisons with the CALIPSO vertical feature mask,” Atmos. Meas. Tech. **6**(5), 1397–1412 (2013). [CrossRef]

16. F. Mao, W. Gong, and Z. Zhu, “Simple multiscale algorithm for layer detection with lidar,” Appl. Opt. **50**(36), 6591–6598 (2011). [CrossRef] [PubMed]

## 4. Summary

## Acknowledgments

## References and links

1. | K. N. Liou, |

2. | V. A. Kovalev and W. E. Eichinger, |

3. | M. Feiyue, G. Wei, and M. Yingying, “Retrieving the aerosol lidar ratio profile by combining ground- and space-based elastic lidars,” Opt. Lett. |

4. | M. Vaughan, D. M. Winker, and K. Powell, “CALIOP algorithm theoretical basis document, part 2: Feature detection and layer properties algorithms,” (NASA Langley Research Center, Hampton, Virginia, USA, 2005). |

5. | J. B. Senberg, A. Ansmann, J. M. Baldasano, D. Balis, C. B. Ckmann, B. Calpini, A. Chaikovsky, P. Flamant, A. Hgrd, and V. Mitev, “EARLINET: a European aerosol research lidar network,” in |

6. | F. Mao, W. Gong, S. Song, and Z. Zhu, “Determination of the boundary layer top from lidar backscatter profiles using a Haar wavelet method over Wuhan, China,” Opt. Laser Technol. |

7. | F. Rocadenbosch, M. Sicard, M. N. M. Reba, and S. Tomas, “Morphological tools for range-interval segmentation of elastic lidar signals,” in IEEE International Geoscience and Remote Sensing Symposium(IGARSS), 2007), 4372~4375. [CrossRef] |

8. | S. R. Pal, W. Steinbrecht, and A. I. Carswell, “Automated method for lidar determination of cloud-base height and vertical extent,” Appl. Opt. |

9. | D. M. Winker and M. A. Vaughan, “Vertical distribution of clouds over Hampton, Virginia observed by lidar under the ECLIPS and FIRE ETO programs,” Atmos. Res. |

10. | Z. Wang and K. Sassen, “Cloud type and macrophysical property retrieval using multiple remote sensors,” J. Appl. Meteorol. |

11. | F. Mao, W. Gong, and C. Li, “Anti-noise algorithm of lidar data retrieval by combining the ensemble Kalman filter and the Fernald method,” Opt. Express |

12. | Y. Morille, M. Haeffelin, P. Drobinski, and J. Pelon, “STRAT: An automated algorithm to retrieve the vertical structure of the atmosphere from single-channel lidar data,” J. Atmos. Ocean. Technol. |

13. | J. Gaumet, J. Heinrich, M. Cluzeau, P. Pierrard, and J. Prieur, “Cloud-base height measurements with a single-pulse erbium-glass laser ceilometer,” J. Atmos. Ocean. Technol. |

14. | S. A. Young, “Analysis of lidar backscatter profiles in optically thin clouds,” Appl. Opt. |

15. | F. Mao, W. Gong, J. Li, and J. Zhang, “Cloud detection and coefficient retrieve based on improved differential zero-crossing method for Mie lidar,” Acta Opt. Sin. |

16. | F. Mao, W. Gong, and Z. Zhu, “Simple multiscale algorithm for layer detection with lidar,” Appl. Opt. |

17. | M. A. Vaughan, K. A. Powell, D. M. Winker, C. A. Hostetler, R. E. Kuehn, W. H. Hunt, B. J. Getzewich, S. A. Young, Z. Liu, and M. J. McGill, “Fully automated detection of cloud and aerosol layers in the CALIPSO lidar measurements,” J. Atmos. Ocean. Technol. |

18. | J. R. Campbell, K. Sassen, and E. J. Welton, “Elevated cloud and aerosol layer retrievals from micropulse lidar signal profiles,” J. Atmos. Ocean. Technol. |

19. | W. Gong, F. Mao, and S. Song, “Signal simplification and cloud detection with an improved Douglas-Peucker algorithm for single-channel lidar,” Meteorol. Atmos. Phys. |

20. | W. Gong, F. Mao, and J. Li, “OFLID: Simple method of overlap factor calculation with laser intensity distribution for biaxial lidar,” Opt. Commun. |

21. | D. H. Douglas and T. K. Peucker, “Algorithms for the reduction of the number of points required to represent a digitized line or its caricature,” Int. J. Geo. Inf. and Geo. |

22. | E. Keogh, S. Chu, D. Hart, and M. Pazzani, “An online algorithm for segmenting time series,” in (Proceedings 2001 IEEE International Conference on Data Mining, 2001), 289–296. [CrossRef] |

23. | S. Burton, R. Ferrare, M. Vaughan, A. Omar, R. Rogers, C. Hostetler, and J. Hair, “Aerosol classification from airborne HSRL and comparisons with the CALIPSO vertical feature mask,” Atmos. Meas. Tech. |

**OCIS Codes**

(280.1100) Remote sensing and sensors : Aerosol detection

(280.3640) Remote sensing and sensors : Lidar

**ToC Category:**

Remote Sensing

**History**

Original Manuscript: August 23, 2013

Revised Manuscript: October 18, 2013

Manuscript Accepted: October 20, 2013

Published: October 30, 2013

**Citation**

Feiyue Mao, Wei Gong, and Timothy Logan, "Linear segmentation algorithm for detecting layer boundary with lidar," Opt. Express **21**, 26876-26887 (2013)

http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-21-22-26876

Sort: Year | Journal | Reset

### References

- K. N. Liou, An Introduction to Atmospheric Radiation (Academic Press, 2002).
- V. A. Kovalev and W. E. Eichinger, Elastic Lidar: Theory, Practice, and Analysis Methods (Wiley-Interscience, 2004).
- M. Feiyue, G. Wei, and M. Yingying, “Retrieving the aerosol lidar ratio profile by combining ground- and space-based elastic lidars,” Opt. Lett.37(4), 617–619 (2012). [CrossRef] [PubMed]
- M. Vaughan, D. M. Winker, and K. Powell, “CALIOP algorithm theoretical basis document, part 2: Feature detection and layer properties algorithms,” (NASA Langley Research Center, Hampton, Virginia, USA, 2005).
- J. B. Senberg, A. Ansmann, J. M. Baldasano, D. Balis, C. B. Ckmann, B. Calpini, A. Chaikovsky, P. Flamant, A. Hgrd, and V. Mitev, “EARLINET: a European aerosol research lidar network,” in Advances in laser remote sensing, (Selected Papers of the 20th International Laser Radar Conference, 2001), pp. 155–158.
- F. Mao, W. Gong, S. Song, and Z. Zhu, “Determination of the boundary layer top from lidar backscatter profiles using a Haar wavelet method over Wuhan, China,” Opt. Laser Technol.49, 343–349 (2013). [CrossRef]
- F. Rocadenbosch, M. Sicard, M. N. M. Reba, and S. Tomas, “Morphological tools for range-interval segmentation of elastic lidar signals,” in IEEE International Geoscience and Remote Sensing Symposium(IGARSS), 2007), 4372~4375. [CrossRef]
- S. R. Pal, W. Steinbrecht, and A. I. Carswell, “Automated method for lidar determination of cloud-base height and vertical extent,” Appl. Opt.31(10), 1488–1494 (1992). [CrossRef] [PubMed]
- D. M. Winker and M. A. Vaughan, “Vertical distribution of clouds over Hampton, Virginia observed by lidar under the ECLIPS and FIRE ETO programs,” Atmos. Res.34(1-4), 117–133 (1994). [CrossRef]
- Z. Wang and K. Sassen, “Cloud type and macrophysical property retrieval using multiple remote sensors,” J. Appl. Meteorol.40(10), 1665–1682 (2001). [CrossRef]
- F. Mao, W. Gong, and C. Li, “Anti-noise algorithm of lidar data retrieval by combining the ensemble Kalman filter and the Fernald method,” Opt. Express21(7), 8286–8297 (2013). [CrossRef] [PubMed]
- Y. Morille, M. Haeffelin, P. Drobinski, and J. Pelon, “STRAT: An automated algorithm to retrieve the vertical structure of the atmosphere from single-channel lidar data,” J. Atmos. Ocean. Technol.24(5), 761–775 (2007). [CrossRef]
- J. Gaumet, J. Heinrich, M. Cluzeau, P. Pierrard, and J. Prieur, “Cloud-base height measurements with a single-pulse erbium-glass laser ceilometer,” J. Atmos. Ocean. Technol.15(1), 37–45 (1998). [CrossRef]
- S. A. Young, “Analysis of lidar backscatter profiles in optically thin clouds,” Appl. Opt.34(30), 7019–7031 (1995). [CrossRef] [PubMed]
- F. Mao, W. Gong, J. Li, and J. Zhang, “Cloud detection and coefficient retrieve based on improved differential zero-crossing method for Mie lidar,” Acta Opt. Sin.30(11), 3097–3102 (2010). [CrossRef]
- F. Mao, W. Gong, and Z. Zhu, “Simple multiscale algorithm for layer detection with lidar,” Appl. Opt.50(36), 6591–6598 (2011). [CrossRef] [PubMed]
- M. A. Vaughan, K. A. Powell, D. M. Winker, C. A. Hostetler, R. E. Kuehn, W. H. Hunt, B. J. Getzewich, S. A. Young, Z. Liu, and M. J. McGill, “Fully automated detection of cloud and aerosol layers in the CALIPSO lidar measurements,” J. Atmos. Ocean. Technol.26(10), 2034–2050 (2009). [CrossRef]
- J. R. Campbell, K. Sassen, and E. J. Welton, “Elevated cloud and aerosol layer retrievals from micropulse lidar signal profiles,” J. Atmos. Ocean. Technol.25(5), 685–700 (2008). [CrossRef]
- W. Gong, F. Mao, and S. Song, “Signal simplification and cloud detection with an improved Douglas-Peucker algorithm for single-channel lidar,” Meteorol. Atmos. Phys.113(1-2), 89 (2011). [CrossRef]
- W. Gong, F. Mao, and J. Li, “OFLID: Simple method of overlap factor calculation with laser intensity distribution for biaxial lidar,” Opt. Commun.284(12), 2966–2971 (2011). [CrossRef]
- D. H. Douglas and T. K. Peucker, “Algorithms for the reduction of the number of points required to represent a digitized line or its caricature,” Int. J. Geo. Inf. and Geo.10, 112–122 (1973).
- E. Keogh, S. Chu, D. Hart, and M. Pazzani, “An online algorithm for segmenting time series,” in (Proceedings 2001 IEEE International Conference on Data Mining, 2001), 289–296. [CrossRef]
- S. Burton, R. Ferrare, M. Vaughan, A. Omar, R. Rogers, C. Hostetler, and J. Hair, “Aerosol classification from airborne HSRL and comparisons with the CALIPSO vertical feature mask,” Atmos. Meas. Tech.6(5), 1397–1412 (2013). [CrossRef]

## Cited By |
Alert me when this paper is cited |

OSA is able to provide readers links to articles that cite this paper by participating in CrossRef's Cited-By Linking service. CrossRef includes content from more than 3000 publishers and societies. In addition to listing OSA journal articles that cite this paper, citing articles from other participating publishers will also be listed.

« Previous Article | Next Article »

OSA is a member of CrossRef.