Edge Detection Method for Determining Boundary Layer Height Based on Doppler Lidar

The top of the boundary layer, referred to as the planetary boundary layer height (BLH), is an important physical parameter in atmospheric numerical models, which has a critical role in atmospheric simulation, air pollution prevention, and climate prediction. The traditional methods for determining BLHs using Doppler lidar vertical velocity variance (σw2) can be classified into the variance and peak methods, which depend on atmospheric conditions due to their use of a single threshold, hence limiting their ability to estimate diurnal BLHs. Edge detection (ED) was later introduced in BLH estimation due to its ability to identify the 2D gradient of an image. A key step in ED is automatically identifying the edge of BLHs based on the peaks of the profile, hence avoiding the influence of extreme atmospheric conditions. Two cases in the diurnal cycle on 4 March 2019 and 8 July 2019 reveal that ED outperforms both the variance and peak methods in nighttime and extreme atmospheric conditions. The retrieved BLHs from 2018 to 2020 were compared with radiosonde (RS) measurements for the same time at the neutral, stable, and convective boundary layers. The correlation coefficient (R: 0.4 vs. 0.05, 0.14; 0.26 vs. −0.10, −0.16; 0.35 vs. 0.01, 0.16) and root mean square error (RMSE (km): 0.58 vs. 0.82, 0.90; 0.37 vs. 1.01, 0.50; 0.66 vs. 0.98, 0.82) obtained by the ED method were higher and lower than those obtained by the variance and peak methods, respectively. The mean absolute error (MAE) of the ED method under the NBL, SBL, and CBL conditions are lower than the variance and peak methods (MAE (km): 0.44, 0.14, 0.50 vs. 0.62, 0.34, 0.64; 0.59, 0.75, 0.74), respectively. The mean relative error (MRE) of the ED method is lower than the variance and peak methods under the NBL condition (MRE: −8.88% vs. −18.39%, 13.91%). Under the SBL, the MRE of the ED method is lower than the variance method and higher than the peak method (−38.64%, vs. −152.23%; 14.02%). Under the CBL, the MRE of the ED method is lower than the variance method and higher than the peak method (−15.07% vs. 2.24%; 5.64%). In addition, the comparison between ED and wavelet covariance transform (WCT) method and RS measurements showed that the ED method has a similar performance with the WCT method and is even better. In the long-term analysis, the hourly and monthly BLHs in the diurnal and annual cycles, respectively, as obtained by ED, were highly consistent with the RS measurements and obtained the lowest standard error. In the annual cycle, the retrieved BLHs in summer and autumn were higher than those retrieved in spring and winter.

Single-threshold Image Segmentation Algorithm Based on Improved Bat Algorithm

The Improved Bat Algorithm (IBA) is proposed for the image segmentation based on the maximum interclass variance method. Firstly, the principle of image segmentation based on the maximum interclass variance method is explained, and secondly, the bat algorithm is improved by using chaotic logistic mapping to initialize the population to improve the diversity of solutions, using adaptive parameter optimization to avoid falling into local optimum, using Monkey algorithm for individual selection, and finally, the image segmentation function in image segmentation is used as the individual fitness function of the bat algorithm for solving. The simulation experiments show that compared with the bat algorithm and the monkey group algorithm, this algorithm has better segmentation effect under different threshold values.

Accurate Determination of Conversion Gains of SVOM VT CCDs Based on a Signal-Dependent Charge-Sharing Mechanism

The signal-variance method and the photon transfer curve method are the most valuable tools for calculating the conversion gains of charge-coupled device (CCD) detectors. This paper describes the phenomena that arise in the conversion gain measurements of space multi-band variable object monitor (SVOM) visible telescope (VT) CCDs, where the results of the signal-variance method increase with the image gray level, and the results of the photon transfer curve method appear with nonlinearity, which is caused by the signal-dependent charge sharing mechanism of back-illuminated CCDs. A numerical simulation model based on random variables was adopted to analyze the influence of the mechanism on the gain determination. The model simulates all the signals and noise in the flat field image, including the photon signal and photon-shot noise, readout noise, fixed pattern noise, and the signal-dependent charge-sharing signal, and it demonstrated agreement with the experimental data. Then, we proposed a quadratic polynomial curve-fitting formula for the photon transfer curve, and we quantitatively analyzed the relationship between the fitting coefficients and the gain, the signal-dependent charge sharing coefficient, and the full well capacity using the control variable method. Finally, the formula was used to accurately determine the conversion gains of SVOM VT CCDs.