Aerosol-cloud droplet conversion stage identification method based on laser radar

Abstract

The invention provides an aerosol-cloud droplet conversion stage identification method based on a laser radar. The method provided by the invention comprises the following steps: acquiring echo signals of a main channel, a depolarization channel and a molecular channel; calculating a particle scattering ratio and a particle depolarization ratio according to the echo signal and an atmospheric molecular scattering model; counting samples of the particle scattering ratio and the particle depolarization ratio in the identification domain, determining joint probability density distribution of the particle scattering ratio and the particle depolarization ratio based on the two-dimensional sample set, and generating a joint probability density scatter diagram; a combined identification criterion is set according to the microscopic physical process characteristics of each stage of aerosol-cloud droplet conversion, and the aerosol-cloud droplet conversion stage is identified according to the particle scattering ratio and the particle depolarization ratio in combination with the combined identification criterion. According to the aerosol-cloud droplet conversion stage identification method based on the laser radar, dynamic and fine identification of the cloud formation stage is realized.

Description

Technical Field

[0001] This application relates to the fields of atmospheric remote sensing and cloud physics, and in particular to a method for identifying the aerosol-cloud droplet conversion stage based on lidar. Background Technology

[0002] Aerosols, as the core component of atmospheric cloud condensation nuclei, are key factors driving cloud microphysical evolution and regulating cloud lifetime and precipitation efficiency. Aerosol-cloud interactions span the entire physical chain of aerosol hygroscopic growth, activation to form nascent cloud droplets, and eventual development into mature cloud droplets. This transformation process exhibits both significant stages and continuity. Accurate dynamic identification of this process is crucial for revealing the indirect climate effects of aerosols, optimizing numerical prediction models of cloud precipitation, and improving the parameterization accuracy of climate models. It is also an important research direction in the fields of atmospheric remote sensing and cloud physics.

[0003] Currently, detection methods for aerosol-cloud conversion processes are mainly divided into two categories: in-situ detection and remote sensing. In-situ detection equipment (such as cloud condensation nucleus counters and optical particle counters) can capture the microscopic physical details of aerosol activation and cloud droplet formation under controlled or near-field conditions, revealing the complex process of cloud droplet formation under different compositions, particle sizes, and environmental conditions, and providing important experimental evidence for understanding the aerosol activation mechanism (for example, the published patent CN116698675A has achieved the measurement of cloud droplet surface tension during aerosol activation and cloud formation). However, in-situ detection is still limited to local spaces and is difficult to monitor large-scale, continuous spatiotemporal evolution in the real atmosphere.

[0004] In contrast, lidar, with its high spatiotemporal resolution, vertical detection capability, and continuous observation advantages, has become the primary tool for monitoring atmospheric aerosols and clouds. However, existing lidar identification methods mostly focus on binary classification of "aerosols" and "clouds" (for example, published patents CN112698354A uses a threshold method, CN116819490A uses a neural network method, and CN116543300A uses a multi-source fusion method, etc.). While these methods can identify typical aerosol layers and cloud layers, they cannot distinguish the continuous transformation stages between the two, and are particularly difficult to effectively capture the formation process of nascent cloud droplets in a transitional state. In-situ detection is limited by the observation range, enabling only point-like monitoring in a local space, and cannot capture the large-scale, continuous spatiotemporal evolution characteristics of aerosol-cloud transformation in the real atmosphere.

[0005] This "either / or" classification logic ignores the intermediate state of high-humidity aerosols evolving into nascent clouds, which is actually widespread. This makes it difficult to effectively utilize effective data in the ambiguous areas of aerosol-cloud characteristic signals, thus hindering a deeper understanding of the continuous evolution mechanism of aerosol activation. Consequently, existing lidar remote sensing technologies can only perform binary classification of aerosols and clouds, making it difficult to identify their continuous transformation stages. Summary of the Invention

[0006] In view of this, this application provides a method for identifying the aerosol-cloud droplet conversion stage based on lidar, so as to achieve dynamic and precise identification of the cloud formation stage.

[0007] Specifically, this application is implemented through the following technical solution:

[0008] The first aspect of this application provides a method for identifying the aerosol-cloud droplet conversion stage based on lidar, the method comprising:

[0009] Acquire echo signals from the main channel, depolarization channel, and molecular channel;

[0010] The particle backscattering coefficient and the molecular backscattering coefficient are calculated based on the echo signal. The particle scattering ratio and the particle depolarization ratio are calculated based on the particle backscattering coefficient and the atmospheric molecular scattering model.

[0011] Determine the identification domain, collect samples of particle scattering ratio and particle depolarization ratio in the identification domain to form a two-dimensional sample set, determine the joint probability density distribution of particle scattering ratio and particle depolarization ratio based on the two-dimensional sample set, and generate a joint probability density scatter plot based on the joint probability density distribution.

[0012] A joint identification criterion is set based on the microscopic physical process characteristics of each stage of aerosol-cloud droplet conversion. The aerosol-cloud droplet conversion stage is identified by combining the particle scattering ratio and the particle depolarization ratio with the joint identification criterion.

[0013] This application provides a lidar-based method for identifying aerosol-cloud droplet transformation stages, aiming to address the problem in existing lidar remote sensing technologies that can only perform binary classification of aerosols and clouds, making it difficult to identify their continuous transformation stages. Specifically, existing methods cannot effectively distinguish the transitional states between aerosol hygroscopic growth, nascent cloud formation, and mature cloud formation. Therefore, this invention provides a lidar-based method for identifying aerosol-cloud droplet transformation stages, which, through joint analysis of particle scattering ratio (R... p ) and debias ratio (δ) pThis system utilizes a two-dimensional probability density distribution to achieve dynamic and precise identification of the continuous transformation process from aerosols to nascent clouds to mature clouds. It overcomes the limitations of traditional binary classification of "aerosol-cloud," enabling the identification of the complete transformation chain from aerosol hygroscopic growth to nascent cloud formation and then to mature cloud formation, providing more refined atmospheric process observation information. Furthermore, based on two-dimensional joint probability density analysis, it effectively suppresses the influence of single-point noise and local outliers, and can be integrated into existing lidar data processing systems. It is suitable for operational and scientific research scenarios such as real-time monitoring, climate change research, and cloud physics parameterization improvement. Attached Figure Description

[0014] Figure 1 A flowchart of Embodiment 1 of the LiDAR-based aerosol-cloud droplet conversion stage identification method provided in this application;

[0015] Figure 2 A lidar monitoring time-height map shown as an exemplary embodiment of this application;

[0016] Figure 3 This is a schematic diagram illustrating the aerosol-cloud conversion identification result, which is an exemplary embodiment of this application. Detailed Implementation

[0017] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application.

[0018] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used herein are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0019] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0020] The following specific embodiments are given to illustrate the technical solution of this application in detail.

[0021] Figure 1 This is a flowchart of Embodiment 1 of the lidar-based aerosol-cloud droplet conversion stage identification method provided in this application. Please refer to... Figure 1 The method provided in this embodiment may include:

[0022] S101. Acquire the echo signals from the main channel, depolarization channel, and molecular channel.

[0023] Specifically, the multi-channel detection module of the lidar is activated to perform vertical or scanning detection on the atmospheric area, and the raw photoelectric signals of the main channel, depolarization channel and molecular channel are received simultaneously. The raw signals of each channel are synchronized and time-registered, and then background noise removal, distance square attenuation compensation and channel gain calibration are performed in sequence to finally obtain the effective echo signals of the three channels.

[0024] During the initial detection phase, the laser beam emitted by the lidar interacts with particles and molecules in the atmosphere. The main channel receives the total backscattered light, including aerosols and cloud droplets, while the depolarization channel receives scattered light perpendicular to the polarization direction of the emitted light. The molecular channel, through spectral filtering, receives only Raman or Rayleigh scattered light from atmospheric molecules. After signal reception, synchronization is first performed to ensure strict temporal alignment of the signals from the three channels, eliminating deviations caused by system response delays. Then, the background noise baseline is determined by statistically analyzing the grayscale values ​​of signal-free areas, and this baseline is subtracted point-by-point from the original signal to remove noise. Next, distance-squared attenuation compensation is performed because the intensity of the lidar echo signal attenuates with the square of the detection distance; the signal intensity of each range gate is multiplied by the square of its distance to restore the true scattering intensity. Finally, based on the gain coefficients of each channel obtained from laboratory calibration, the signal is normalized and calibrated to eliminate differences in photoelectric conversion efficiency between different detection channels, ensuring that the echo signals used in subsequent calculations have dimensional accuracy and comparability between channels.

[0025] Furthermore, the lidar system will receive three signals with different characteristics, which can be expressed by the lidar equation as follows:

[0026] (1)

[0027] (2)

[0028] (3)

[0029] Among them, B || B ⊥ , These are the signals from the main channel, depolarization channel, and molecular channel of the lidar, respectively.

[0030] β is the backscattering coefficient. Let be the backscattering coefficient of the molecule in the main channel of the lidar. This represents the backscattering coefficient of atmospheric particulate matter in the main channel of the lidar.

[0031] denoted as the backscattering coefficient of the molecule in the depolarization channel. This represents the backscattering coefficient of atmospheric particulate matter in the depolarization channel;

[0032] denoted as the backscattering coefficient of the molecule in the molecular channel. This represents the backscattering coefficient of atmospheric particulate matter in the molecular channel;

[0033] α is the extinction coefficient. The extinction coefficient of the molecule is . The extinction coefficient of atmospheric particulate matter;

[0034] η is the multiple scattering factor, which is more pronounced in mature clouds;

[0035] f m f p This is the transmittance factor of the spectral discriminator, obtained through system calibration. Note B. || B ⊥ The signal should be calibrated to the system gain ratio. The lidar signal is a calibrated complete component signal, B=B. || +B ⊥ .

[0036] S102. Calculate the particle scattering ratio and particle depolarization ratio based on the echo signal and the atmospheric molecular scattering model.

[0037] Specifically, based on the standard number density profile of atmospheric molecules, the molecular backscattering coefficient is obtained by inversion; then, based on the effective echo signals of the main channel and the molecular channel, the signal contribution corresponding to the molecular backscattering coefficient is subtracted to obtain the particle backscattering coefficient; subsequently, the ratio of the particle backscattering coefficient to the molecular backscattering coefficient is calculated to obtain the particle scattering ratio; finally, the ratio of the particle backscattering coefficient in the depolarization channel to the particle backscattering coefficient in the main channel is calculated to obtain the particle depolarization ratio.

[0038] Furthermore, through signal inversion and ratio calculation, the original echo signal is transformed into dimensionless parameters that characterize the physical properties of particles. When inverting the molecular backscattering coefficient, the known laws governing atmospheric molecular scattering characteristics are utilized, combined with the molecular number density at different altitudes provided by the standard atmospheric model, to invert the calibrated signal of the molecular channel, directly obtaining the backscattering coefficients of atmospheric molecules at each detection altitude. When inverting the particle backscattering coefficient, the main channel signal contains contributions from both particles and molecules. Therefore, the molecular backscattering signal obtained from the molecular channel inversion needs to be subtracted from the total backscattering signal of the main channel to obtain the scattering signal contributed only by aerosols and cloud droplets. This is then inverted using the lidar equation to obtain the particle backscattering coefficient. When calculating the particle scattering ratio, the particle backscattering coefficient at the same detection altitude is used as the numerator, and the molecular backscattering coefficient as the denominator, and a point-by-point division operation is performed. This ratio directly reflects the relative strength of particle scattering contributions to molecular scattering contributions and is a core parameter for distinguishing between aerosols and cloud droplets. When calculating the particle depolarization ratio, the contribution of molecular depolarization scattering is first subtracted from the total signal of the depolarization channel to obtain the particle depolarization scattering signal and invert it into the particle depolarization backscattering coefficient. Then, the ratio of this coefficient to the particle backscattering coefficient of the main channel is calculated point by point. This ratio reflects the degree of non-sphericity of the particles. Aerosol particles are mostly non-spherical, while cloud droplets are mostly spherical. Therefore, this parameter can help distinguish the phase characteristics of particles.

[0039] Furthermore, the steps for calculating the particle scattering ratio and particle depolarization ratio based on the echo signal and the atmospheric molecular scattering model include:

[0040] (1) Calculate the particle backscattering coefficient and the molecular backscattering coefficient based on the echo signal;

[0041] Specifically, the steps for calculating the particle backscattering coefficient and the molecular backscattering coefficient based on the echo signal include:

[0042] (1.1) Using the echo signal of the molecular channel as a standard reference, normalize and correct the echo signals of the main channel and the depolarization channel;

[0043] Specifically, effective echo signal data from the molecular channel, main channel, and depolarization channel at the same detection time and distance gate are extracted to determine the calibration reference value of the molecular channel signal; the system response deviation coefficients of the main channel and depolarization channel relative to the molecular channel are calculated, and the echo signals of the main channel and depolarization channel are corrected point by point based on the deviation coefficients; the consistency of the three channel signals after correction is verified, and abnormal data points that still exist after correction are removed to obtain the normalized corrected echo signals of the main and depolarization channels.

[0044] Furthermore, the molecular channel receives only scattered signals from atmospheric molecules, is minimally affected by aerosols and cloud droplets, and exhibits stable signal characteristics; therefore, it was selected as the standard reference for normalization correction. In the operation, precise spatiotemporal and range gate matching was first performed on the three-channel signals to ensure that the signals involved in the correction were all observation data from the same detection time and atmospheric altitude. Then, using pre-calibrated system parameters in the laboratory, combined with the actual observed molecular channel signal reference values, the inherent response deviation coefficients of the main channel and depolarization channel, caused by differences in photodetector sensitivity, optical lens transmittance, and circuit amplification, were calculated. Subsequently, the echo signals from each range gate of the main and depolarization channels were multiplied by the corresponding deviation coefficient to complete the normalization correction, ensuring that the signal response benchmarks of the three channels remain consistent when detecting the same atmospheric target. After correction, the effectiveness was verified by statistically analyzing the correlation and deviation rate of the three-channel signals. Abnormal data points with deviations exceeding the threshold were removed to avoid interference from outliers in subsequent calculations, ensuring the accuracy and reliability of the corrected signals.

[0045] (1.2) Based on the inversion principle of the lidar system and the atmospheric molecular scattering model, the corrected echo signal is solved to obtain the particle backscattering coefficient and the molecular backscattering coefficient.

[0046] Specifically, based on the type of lidar used, the corresponding lidar inversion equation and system calibration parameters are retrieved; the corrected molecular channel echo signal is substituted into the inversion equation to calculate the atmospheric molecular backscattering coefficient; then the corrected main channel echo signal is substituted into the inversion equation, and the signal contribution corresponding to the molecular backscattering coefficient is subtracted to calculate the aerosol / cloud droplet particle backscattering coefficient; the two types of coefficients obtained are smoothed at different altitudes to obtain continuous and stable particle backscattering coefficient and molecular backscattering coefficient profiles.

[0047] Furthermore, the normalized echo signal is transformed into physical parameters characterizing atmospheric scattering properties. The calculation process strictly matches the lidar system type to ensure the scientific validity of the calculation logic. Different types of lidars possess different signal separation and detection principles. High-spectral-resolution lidars separate particle and molecular scattering signals through spectral frequency discrimination technology, while Raman scattering lidars achieve specific detection by capturing the Raman scattering signals of molecules. Therefore, it is necessary to first retrieve the inversion equation matching the equipment used, and simultaneously import fixed parameters calibrated in the laboratory, such as the lidar's transmit power, receiving telescope aperture, and system transmittance. When calculating the molecular backscattering coefficient, the corrected molecular channel signal is directly substituted into the corresponding inversion equation. Combined with environmental parameters such as molecular number density, atmospheric temperature, and air pressure at different altitudes provided by the standard atmospheric model, the molecular backscattering coefficient is calculated altitude-by-altitude. The molecular backscattering coefficient only reflects the scattering ability of atmospheric molecules and serves as the benchmark for subsequently distinguishing particle scattering contributions. When calculating the particle backscattering coefficient, the signal after main channel correction includes the combined scattering contribution of particles and molecules. Therefore, the signal quantity of the corresponding molecule scattering is first calculated using the molecular backscattering coefficient. This contribution is then accurately subtracted from the total signal of the main channel. The remaining signal is then substituted into the inversion equation to obtain the backscattering coefficient contributed only by aerosols and cloud droplets. After the calculation, the particle and molecular backscattering coefficients are smoothed by moving average with adjacent height gates to eliminate random fluctuations from single-point detection, resulting in a coefficient profile that is continuously distributed along atmospheric height. This makes the results more consistent with the actual vertical distribution characteristics of the atmosphere and meets the parameter requirements for subsequent calculations of particle scattering ratio and depolarization ratio.

[0048] Normalization correction of the main and depolarization channel signals using the molecular channel echo signal as a standard can effectively eliminate signal deviations caused by differences in hardware system response between different detection channels of lidar, ensuring that the signals of each channel have uniform dimensions and comparability, laying a precise and consistent signal foundation for subsequent calculations. Based on this, the corrected signals are calculated using the lidar system's corresponding inversion principle, which can accurately separate the scattering signal contributions of atmospheric molecules and aerosol / cloud droplet particles, accurately calculate the particle backscattering coefficient and molecular backscattering coefficient that truly reflect the atmospheric scattering characteristics, effectively avoiding the interference of channel signal deviation and signal aliasing on the coefficient calculation results, ensuring the accuracy, reliability, and continuity of the calculated physical parameters, and providing high-quality core parameter support for the subsequent calculation of particle scattering ratio and depolarization ratio, as well as the accurate identification of the aerosol-cloud droplet conversion stage.

[0049] (2) Calculate the particle scattering ratio based on the ratio of the particle backscattering coefficient to the molecular backscattering coefficient;

[0050] Specifically, the effective values ​​of particle backscattering coefficient and molecular backscattering coefficient corresponding to the same detection time and space and the same distance gate are extracted. The particle backscattering coefficient is used as the numerator and the molecular backscattering coefficient is used as the denominator to perform point-by-point ratio calculation. The calculation results are screened and removed for outliers to obtain the particle scattering ratio corresponding to each detection position.

[0051] Furthermore, by calculating the ratio of core physical parameters, the scattering ability of particles and molecules is quantified into dimensionless characteristic parameters, intuitively reflecting the relative strength of particle scattering contribution compared to atmospheric molecule scattering contribution. Before the calculation, it is necessary to ensure that the backscattering coefficients of particles and molecules are matched data at the same detection altitude and time to avoid calculation errors caused by spatiotemporal misalignment. After completing the ratio calculation point by point, a threshold is set according to the conventional physical range of atmospheric observation to eliminate abnormal values ​​caused by instantaneous equipment fluctuations and signal interference, ensuring that the particle scattering ratio can truly characterize the scattering proportion of aerosol / cloud droplet particles. This is a key quantitative indicator for distinguishing the evolution stages of aerosols and cloud droplets, and its value can directly reflect the concentration characteristics of particles.

[0052] Furthermore, the particle scattering ratio can be calculated using the following formula:

[0053] (4)

[0054] in, The particle backscattering coefficient;

[0055] The molecular backscattering coefficient;

[0056] δ v = B ⊥ / B || For the body to retreat and be biased;

[0057] The departition ratio of the molecule;

[0058] is the transmittance factor of the spectral discriminator for molecules;

[0059] is the transmittance factor of the particle spectral discriminator;

[0060] χ= B || / , where is the signal mixing ratio.

[0061] (3) Calculate the particle depolarization ratio based on the ratio of the particle backscattering coefficient of the depolarization channel to the particle backscattering coefficient of the main channel.

[0062] Specifically, the effective values ​​of the backscattering coefficients of the depolarized channel particles and the main channel particles corresponding to the same detection time and space and the same distance gate are extracted. The backscattering coefficients of the depolarized channel particles are used as the numerator and the backscattering coefficients of the main channel particles are used as the denominator to perform point-by-point ratio calculations. The calculation results are smoothed to obtain the particle depolarization ratio corresponding to each detection position.

[0063] Furthermore, quantifying the non-spherical characteristics of particles by using the ratio of particle scattering coefficients in different polarization channels is a core basis for distinguishing between aerosol and cloud droplet morphologies. Before calculation, it is necessary to confirm that the particle backscattering coefficients in both channels have deducted the contribution of molecular scattering, retaining only the polarization scattering signals of aerosol / cloud droplet particles. After calculating the ratio point by point, the results are smoothed by a moving average of adjacent detection points to eliminate the influence of single-point random noise. The magnitude of the particle depolarization ratio is negatively correlated with the sphericity of the particles. Aerosol particles are mostly non-spherical, corresponding to higher depolarization ratio values, while cloud droplet particles gradually become spherical after absorbing moisture and growing, corresponding to lower depolarization ratio values. The trend of this parameter can accurately reflect the morphological evolution process of particles from aerosols to cloud droplets, providing important morphological characteristics for subsequent identification of the transformation stage.

[0064] Furthermore, the particle depolarization ratio can be calculated using the following formula:

[0065] (5)

[0066] in, This represents the particle backscattering coefficient in the depolarization channel;

[0067] The particle backscattering coefficient of the main channel;

[0068] δ v = B ⊥ / B || For the body to retreat and be biased;

[0069] This refers to the particle scattering ratio;

[0070] The departition ratio of the molecule;

[0071] The particle scattering ratio, calculated by the ratio of particle backscattering coefficients to molecular backscattering coefficients, can intuitively quantify the relative strength of particle scattering contribution to molecules, accurately reflecting the evolution characteristics of particle concentration and size. The particle depolarization ratio, calculated by the ratio of particle backscattering coefficients in the depolarization channel to the main channel, can effectively quantify the degree of particle non-sphericity, accurately characterizing the morphological evolution characteristics of particles from aerosols to cloud droplets. Both calculations are based on the same spatiotemporally matched effective physical parameters. After outlier removal and smoothing, the resulting dimensionless characteristic parameters possess accuracy, stability, and clear physical meaning. Together, they constitute the core quantitative indicators for identifying the aerosol-cloud droplet conversion stage, providing quantifiable and analyzable key data support for subsequent joint probability density analysis and accurate identification of the conversion stage.

[0072] S103. Determine the identification domain, collect samples of particle scattering ratio and particle depolarization ratio in the identification domain to form a two-dimensional sample set, determine the joint probability density distribution of particle scattering ratio and particle depolarization ratio based on the two-dimensional sample set, and generate a joint probability density scatter plot based on the joint probability density distribution.

[0073] Optionally, the steps to form a two-dimensional sample set include:

[0074] Extract sample data of all particle scattering ratios and particle depolarization ratios in the identification domain; combine them into two-dimensional data pairs according to a one-to-one correspondence; and construct the two-dimensional sample set based on all two-dimensional data pairs.

[0075] Specifically, based on the observation objective and meteorological background, the spatiotemporal recognition range of the lidar detection data is defined, and the identification domain is determined. Within the identification domain, the particle scattering ratio and particle depolarization ratio values ​​at all effective detection heights are extracted, and each set of corresponding values ​​is taken as a sample and integrated to form a two-dimensional sample set. The kernel density estimation method is used to perform statistical analysis on the two-dimensional sample set to calculate the joint probability density distribution of particle scattering ratio and particle depolarization ratio. With particle scattering ratio as the abscissa and particle depolarization ratio as the ordinate, the sample points are assigned corresponding colors or transparency according to the magnitude of the joint probability density distribution to generate a joint probability density scatter plot.

[0076] Furthermore, by defining the analysis scope and statistical modeling, discrete detection data are transformed into visual maps with statistical characteristics. When determining the identification domain, data from all time periods and altitudes is not used. Instead, specific observation tasks are considered, such as selecting specific time periods when aerosol lifting or cloud formation occurs, and specific altitude ranges from near-surface to the mid-troposphere. Simultaneously, synchronous meteorological radar or radiosonde data can be combined to eliminate anomalous data segments with instrument malfunctions or strong interference, ensuring that the data within the identification domain accurately reflects the process of aerosol transformation into cloud droplets. When constructing the two-dimensional sample set, for each detection range gate within the identification domain, the particle scattering ratio and particle depolarization ratio at that location are extracted. These two values ​​are used to form a one-to-one correspondence for a two-dimensional sample. All two-dimensional samples within the identification domain that meet the validity requirements are aggregated to form a complete two-dimensional sample set. When calculating the joint probability density distribution, a non-parametric statistical method, kernel density estimation, is employed. A Gaussian kernel function is selected as the weighting function, centered on each sample point in the two-dimensional sample set. The bandwidth of the kernel function is adaptively adjusted according to the spatial distribution of the sample points. By superimposing and integrating the kernel functions of all sample points, the probability density value in the entire two-dimensional space is obtained. This distribution reflects the frequency of the combination of particle scattering ratio and particle depolarization ratio in the discrimination domain. When generating the joint probability density scatter plot, each sample point in the two-dimensional sample set is plotted in a Cartesian coordinate system. The probability density value of each sample point or grid, calculated based on the joint probability density distribution, is then rendered using a gradient color scheme. Regions with high probability density are presented as dark colors or high-transparency clusters, while regions with low probability density are presented as light colors or scattered distributions, thus visually demonstrating the statistical correlation characteristics of the two parameters.

[0077] Furthermore, the steps for determining the identification domain include:

[0078] (1) Determine the evolution of particle properties with height and time during the aerosol-cloud droplet conversion process;

[0079] Specifically, the study integrates particle scattering ratio and depolarization ratio data for each altitude gate throughout the entire detection period to construct a spatiotemporal distribution matrix of particle properties. It analyzes the vertical gradient variation characteristics of particle properties with increasing altitude from an altitude perspective, and the dynamic evolution trend of particle properties with the duration of detection from a time perspective. Combining the laws of atmospheric vertical motion and the physical mechanisms of aerosol activation, it summarizes the core evolution law of increasing particle scattering ratio and decreasing particle depolarization ratio, clarifying the numerical ranges and rate characteristics of properties at different evolution stages.

[0080] Furthermore, when constructing the spatiotemporal distribution matrix, detection time was used as the vertical dimension and detection altitude as the horizontal dimension. The particle scattering ratio and particle depolarization ratio values ​​of each spatiotemporal node were filled into the matrix to form a complete spatiotemporal map of particle properties. In the vertical gradient analysis, the focus was on observing the changes in particle properties from the near-surface to the mid-troposphere, identifying the initial altitude of aerosol hygroscopic growth, the formation altitude of nascent cloud droplets, and the development altitude of mature cloud droplets, clarifying the numerical characteristics of particle properties at each altitude level. In the dynamic trend analysis, the changes in particle properties at the same altitude level over time were tracked, capturing key time nodes where the particle scattering ratio increases slowly to rapidly increase, and the particle depolarization ratio decreases slowly to rapidly decrease. Finally, combining thermodynamic and cloud physics principles, the universal evolution law of continuously increasing particle scattering ratio and continuously decreasing particle depolarization ratio during the aerosol-cloud droplet conversion process was verified and determined. At the same time, the rate of property change in different conversion stages was quantified, such as the slow change in the aerosol stage and the rapid abrupt change in the nascent cloud stage, providing a clear physical boundary reference for the subsequent division of the initial range.

[0081] (2) Divide the initial range according to the evolution law, and filter the data interval of the initial range according to the signal-to-noise ratio threshold to obtain the identification domain.

[0082] Specifically, based on the summarized spatiotemporal evolution laws of particle properties, an initial spatiotemporal range encompassing the entire process of aerosol hygroscopic growth, initial cloud droplet formation, and mature cloud droplet development is defined; the signal-to-noise ratio (SNR) of the echo signal at each spatiotemporal node within the initial range is calculated, and a minimum SNR threshold is set; invalid data intervals with SNR below the threshold within the initial range are eliminated, and valid data intervals with SNR meeting the standard and conforming to the evolution laws are retained, which are then determined as the final identification domain.

[0083] Furthermore, when defining the initial scope, the evolutionary pattern determined earlier is strictly followed. In terms of time, a complete period from the initial activation of aerosols to the maturation of cloud droplets is selected. In terms of altitude, the entire vertical range from the aerosol aggregation layer to the mature cloud layer is covered, ensuring that the initial scope fully encompasses all key stages of the transformation process and avoids omitting core data. During the signal-to-noise ratio (SNR) calculation and screening stage, the ratio of signal power to noise power is calculated for the echo signals of the main channel, depolarization channel, and molecular channel within the initial scope. The final SNR for this spatiotemporal node is determined by comprehensively considering the SNR of the three channels. The set SNR threshold needs to be determined in conjunction with the detection performance of the lidar system and the atmospheric observation requirements to ensure that low-quality data caused by instrument noise and strong atmospheric turbulence interference are eliminated. For spatiotemporal intervals with a signal-to-noise ratio below the threshold, regardless of whether the particle properties conform to the evolution law, they are all judged as invalid data and removed. The final valid data interval is the identification domain. This identification domain not only fully covers the entire physical process of aerosol-cloud droplet transformation, but also ensures high data quality, laying a solid data foundation for subsequent joint probability density analysis and transformation stage identification.

[0084] By determining the evolution of particle properties with altitude and time, the initial range covering the entire transformation stage can be precisely delineated based on the physical nature of aerosol-cloud droplet transformation, ensuring the physical integrity of the identification domain from the outset. A secondary screening of the initial range based on the signal-to-noise ratio threshold effectively eliminates low-quality data intervals affected by instrument noise and atmospheric interference, further improving the reliability and validity of data within the identification domain. This combined physical demarcation and quality screening mechanism provides an identification domain that fully encompasses the core spatiotemporal range of aerosol hygroscopic growth, initial cloud droplet formation, and mature cloud droplet development, while ensuring high signal-to-noise ratio and physical validity of the data. This completely avoids interference from invalid data in subsequent statistical analysis and stage identification, providing a high-quality, highly targeted core analysis range for the accurate construction of the joint probability density distribution and the refined identification of transformation stages.

[0085] Furthermore, in order to accurately identify the continuous transformation process from aerosol to cloud, it is necessary to select observation spatiotemporal regions with typical dynamic evolution characteristics as the identification domain. Figure 2 The lidar monitoring time-height map shown as an exemplary embodiment of this application is referred to in [reference]. Figure 2 This embodiment selects four regions with different macroscopic characteristics for comparative analysis based on the lidar time-elevation map. These regions range in altitude from approximately 4.5 to 6 km and cover the time period from 23:30 to 02:00 local time. Four different types of identification domains were selected:

[0086] Judgment domain 1: The signal is relatively strong overall and the structure is relatively uniform.

[0087] Judgment domain 2: The signal shows a clear trend from weak to strong.

[0088] Judgment domain 3: The signal has obvious local enhancement spots.

[0089] Judgment domain 4: The signal has no enhancement spots, but is slightly stronger than the signal at higher levels.

[0090] Based on the analysis of the macroscopic characteristics of the signal, the number of cloud droplets in identification domains 1 to 4 shows a decreasing trend. Valid data points are extracted from each identification domain to form data point sets (particle depolarization ratio, particle scattering ratio), with the number of data points in identification domains 1 to 4 being 5874, 6408, 2546, and 1800, respectively. Subsequently, two-dimensional kernel density estimation is performed on the data samples in each identification domain. In this embodiment, a Gaussian kernel function is used to generate a scatter plot of the joint probability density of particle depolarization ratio and particle scattering ratio, and the density distribution of the probability density is visualized using color bars.

[0091] Furthermore, the steps for determining the joint probability density distribution of particle scattering ratio and particle depolarization ratio based on the two-dimensional sample set include:

[0092] (1) Preprocess the samples in the two-dimensional sample set;

[0093] Specifically, the particle scattering ratio and particle depolarization ratio of the two-dimensional sample set are paired and verified for validity. Abnormal outlier sample points caused by lidar signal noise and sudden atmospheric disturbances are removed. At the same time, invalid sample data with missing values ​​and calibration failures are screened out. Valid two-dimensional sample data that meet the lidar observation data quality standards are retained to form a preprocessed two-dimensional sample set.

[0094] (2) The kernel density estimation method is used to perform two-dimensional statistical analysis on the preprocessed two-dimensional sample set to obtain the joint probability density distribution.

[0095] Specifically, based on the preprocessed two-dimensional sample set, an appropriate kernel function is selected to perform kernel density estimation calculations on the two-dimensional sample data of particle scattering ratio and particle depolarization ratio. At the same time, the bandwidth parameter of kernel density estimation is adaptively determined according to the distribution characteristics of the sample data. By fitting the global kernel density of the two-dimensional sample data and performing statistical operations, a joint probability density distribution that can reflect the sample distribution characteristics of particle scattering ratio and particle depolarization ratio is obtained.

[0096] Furthermore, the steps for generating a joint probability density scatter plot based on the joint probability density distribution include:

[0097] (1) The numerical magnitude of the joint probability density distribution is mapped through a visual color gradient;

[0098] Specifically, the extreme values ​​of the joint probability density distribution are determined, and the numerical grading intervals of the probability density are defined. Then, a continuous color gradient system is selected to establish a one-to-one mapping relationship between color depth, warmth, and probability density values. Higher values ​​match colors with higher visual recognition, and lower values ​​match colors with lower visual recognition. The color gradient mapping rule is applied to all joint probability density distribution data to complete the conversion from numerical values ​​to visual colors.

[0099] Furthermore, the maximum and minimum values ​​of all joint probability density values ​​are statistically analyzed, and the data distribution is graded into equal intervals or equal frequencies based on the degree of concentration. This allows different levels of values ​​to correspond to different color gradient ranges, avoiding the loss of visual information due to large numerical ranges. The selected color gradients must follow the laws of visual perception, prioritizing monochromatic gradients or multi-color gradients with moderate contrast to ensure that subtle differences in probability density can be reflected through color changes, while maintaining visual harmony. When establishing mapping relationships, areas with high probability density are matched with dark, highly saturated colors, while areas with low probability density are matched with light, low-saturation colors. This allows the core areas where sample points cluster to be quickly identified, laying a visual foundation for the subsequent drawing of the joint probability density scatter plot.

[0100] (2) Construct a coordinate system with the particle scattering ratio as the abscissa and the particle debiasing ratio as the ordinate. Mark the distribution position of the sample points in the coordinate system and assign color features to the sample points that match the probability density to generate a joint probability density scatter plot.

[0101] Specifically, a two-dimensional rectangular coordinate system is constructed by setting the coordinate axis scale and range according to the numerical range of the two parameters, with particle scattering ratio as the horizontal axis and particle depolarization ratio as the vertical axis. A two-dimensional sample set of particle scattering ratio and particle depolarization ratio within the identification domain is extracted, and each sample point is accurately labeled at the corresponding position in the coordinate system according to the horizontal and vertical coordinate values. Based on the previously established color gradient mapping rules, each sample point is assigned a color feature that matches its joint probability density value. The labeled sample points are then visually optimized by adding coordinate labels, color scale bars, and other elements to generate a complete joint probability density scatter plot.

[0102] Furthermore, by constructing a coordinate system and visually labeling sample points, the correlation features of particle scattering ratio and debiasing ratio, along with the distribution features of joint probability density, are integrated into a visual map, which serves as an important visual basis for subsequent identification stages. When constructing the coordinate system, the coordinate axis scale and range must be adapted to the parameter value range of the sample set. This ensures that all sample points are fully represented in the coordinate system while avoiding excessive concentration of sample points due to an excessively large range, which would affect the observation of distribution features. When labeling sample points, the numerical correspondence between the horizontal and vertical axes is strictly followed to ensure that the sample point positions accurately reflect the combined features of particle scattering ratio and debiasing ratio without positional offset. When assigning color features, the joint probability density value of each sample point is precisely matched with its corresponding color, allowing areas of concentrated color in the coordinate system to intuitively correspond to concentrated sample areas with high probability density, and areas of dispersed color to sparse sample areas with low probability density. Finally, clear coordinate labels and color scale bars are added, allowing users to quickly compare the probability density values ​​corresponding to the colors using the scale bars. This completes the visual optimization of the joint probability density scatter plot, ensuring the readability and usability of the map.

[0103] By transforming joint probability density values ​​into visual colors through color gradients, an intuitive expression of abstract numerical values ​​is achieved, allowing the density characteristics of sample distribution to be directly perceived through color changes. A coordinate system is constructed using particle scattering ratio and debiasing ratio as axes, and sample points are assigned color features matching their probability densities to generate joint probability density scatter plots. This not only accurately visualizes the parameter correlation characteristics of the two-dimensional sample set but also integrates the distribution patterns of probability density, allowing the sample aggregation characteristics at different stages of the aerosol-cloud droplet conversion process to be clearly presented through a combination of color and location. Transforming physical parameter values ​​and joint probability density distributions into easily identifiable visual maps effectively reduces the data analysis difficulty for subsequent stage identification, enabling identifyrs to quickly locate sample concentration areas at different conversion stages. This provides intuitive and clear visualization support for delineating activation, growth, and saturation zones based on distribution characteristics, and for accurately identifying aerosol-cloud droplet conversion stages.

[0104] S104. Based on the microscopic physical process characteristics of each stage of aerosol-cloud droplet conversion, a joint identification criterion is set, and the aerosol-cloud droplet conversion stage is identified based on the particle scattering ratio, the particle depolarization ratio, and the joint identification criterion.

[0105] Specifically, the microscopic physical characteristics of the three transformation stages of aerosol hygroscopic growth, initial cloud droplet formation, and mature cloud droplet development are analyzed. For each stage, a joint identification criterion is set, including the threshold range of particle scattering ratio, the threshold range of particle depolarization ratio, and the changing trends of both. Each two-dimensional sample within the identification domain is traversed, and the particle scattering ratio and particle depolarization ratio values ​​in the sample, as well as the changing trends of the sample in the time or height dimension, are matched one by one with the joint identification criteria of each stage. Based on the matching results, each sample is classified into the corresponding transformation stage, completing the comprehensive identification of the aerosol-cloud droplet transformation stages within the entire identification domain.

[0106] Furthermore, based on physical mechanisms, criteria are set to achieve a refined stage division of the aerosol-cloud droplet conversion process. When setting joint identification criteria, the microscopic physical nature of each stage is closely considered. For the aerosol hygroscopic growth stage, characterized by small particle size and non-spherical shape, the criterion is that the particle scattering ratio is in the first threshold range, and with time, the particle scattering ratio slowly increases while the particle depolarization ratio slowly decreases. For the initial cloud droplet formation stage, characterized by rapid particle size increase and weakened non-spherical characteristics after hygroscopic activation, the criterion is that the particle scattering ratio is in the second threshold range, and both show a significant trend of rapid increase in particle scattering ratio and rapid decrease in particle depolarization ratio. For the mature cloud droplet development stage, characterized by spherical droplets with particle size reaching the cloud droplet level, the criterion is that the particle scattering ratio is in the third threshold range, and both values ​​tend to stabilize without significant change. During the matching and identification process, a sample-by-sample traversal approach is adopted. This not only determines whether the value of a single sample falls within a threshold range but also analyzes the trend of changes in adjacent time periods or at adjacent heights to ensure compliance with the criteria, thus avoiding misjudgments caused by instantaneous fluctuations. For samples in the transition region between two stages, if they simultaneously meet some criteria for both stages, they are classified into the stage with the higher probability density based on the cluster center of the joint probability density distribution. Through this process, each detection point within the identification domain can be precisely mapped to the specific stage of aerosol-cloud droplet transformation, achieving dynamic and refined identification of this continuous physical process.

[0107] Furthermore, the implementation steps for setting joint identification criteria based on the microscopic physical process characteristics of each stage of aerosol-cloud droplet conversion include:

[0108] (1) When the particle scattering ratio is within the first threshold range and the particle depolarization ratio decreases as the particle scattering ratio increases, it is determined to be in the aerosol stage;

[0109] Specifically, a preset first threshold range for particle scattering ratio is retrieved, and the particle scattering ratio value of each sample point in the identification domain is compared with this range to screen out candidate sample points whose values ​​fall within this range; continuous sequence extraction is performed on the candidate sample points to analyze the changing trend of particle depolarization ratio as the particle scattering ratio value increases; finally, when it is confirmed that the particle scattering ratio meets the value range requirements and the particle depolarization ratio shows a continuous decreasing trend, this type of sample point is determined to be in the aerosol stage.

[0110] Furthermore, based on the initial physical characteristics of aerosol hygroscopic growth, stage determination is completed to identify stages where particle size is small and non-spherical characteristics gradually weaken. The first threshold range is a low-value interval pre-defined based on the scattering characteristics of aerosol particles, corresponding to the scattering contribution of unactivated or initially hygroscopic aerosol particles in the atmosphere. When screening candidate sample points, the particle scattering ratio is checked point by point to ensure that only samples that meet the scattering characteristics are included. In trend analysis, candidate sample points are sorted by particle scattering ratio from smallest to largest. By calculating the difference in particle depolarization ratio between adjacent sample points, the overall direction of change is determined. If the difference is consistently negative, it is considered a downward trend. This trend corresponds to the microscopic physical process of aerosol particles slowly increasing in size and gradually transitioning to a spherical shape during hygroscopic growth. Only when both the numerical range and the trend conditions are met can the aerosol stage be finally determined, avoiding misjudgment caused by a single numerical value.

[0111] (2) When the particle scattering ratio is within the second threshold range and the particle depolarization ratio increases with the increase of the particle scattering ratio, it is determined to be the initial cloud stage;

[0112] Specifically, a preset second threshold range for particle scattering ratio is retrieved, which is higher than the first threshold range. The particle scattering ratio values ​​of unjudged sample points within the identification domain are compared with this range to filter out candidate sample points. Continuous sequence extraction is performed on the candidate sample points to analyze the changing trend of particle depolarization ratio as the particle scattering ratio value increases. When it is confirmed that the particle scattering ratio meets the value range requirements and the particle depolarization ratio shows a continuous upward trend, such sample points are judged as the nascent cloud stage.

[0113] Furthermore, focusing on the critical transition stage of aerosol to cloud droplets, the key is to identify the physical processes of rapid particle activation and the formation of nascent cloud droplets. The second threshold range is a pre-defined median range based on the scattering characteristics of nascent cloud droplets, corresponding to the stage where aerosol particles rapidly increase in size and significantly enhance their scattering contribution after hygroscopic activation. During the screening process, sample points not identified in previous steps are prioritized to avoid overlapping stages. In trend analysis, candidate sample points are sorted by particle scattering ratio, and the difference between adjacent particle depolarization ratios is calculated. If the difference remains consistently positive, it is considered an upward trend. This specific trend corresponds to the microscopic physical phenomenon of a transient enhancement of non-spherical characteristics due to particle aggregation in the early stages of nascent cloud droplet formation. Through dual verification of the numerical range and the upward trend, the sample characteristics of this transition stage can be accurately captured, achieving effective identification of the nascent cloud stage.

[0114] (3) When the scattering ratio is in the third threshold range and the particle depolarization ratio shows a stable or slightly changing trend as the particle scattering ratio increases, it is determined to be a mature cloud stage. The third threshold range is greater than the second threshold range, and the second threshold range is greater than the first threshold range.

[0115] Specifically, a preset third threshold range for particle scattering ratio is retrieved. Since the third threshold range is higher than the second threshold range, the particle scattering ratio values ​​of the remaining undetermined sample points within the identification domain are compared with this range to filter out candidate sample points. Continuous sequence extraction is performed on the candidate sample points, and a fluctuation threshold for particle depolarization ratio is set to analyze the change in particle depolarization ratio as the particle scattering ratio value increases. When it is confirmed that the particle scattering ratio meets the numerical range requirements and the change in particle depolarization ratio is less than the preset fluctuation threshold, this type of sample point is determined to be in the mature cloud stage.

[0116] Furthermore, the final stage determination is completed based on the steady-state physical characteristics of mature cloud droplets, with the core being the identification of the stage where particles have formed spherical droplets and achieved morphological stability. The third threshold range is a high-value interval pre-defined based on the scattering characteristics of mature cloud droplets, corresponding to the stage where cloud droplet particle size has reached a stable level and scattering contribution dominates. During screening, only the remaining sample points not determined in the previous steps are considered to ensure the uniqueness of the stage division. In trend analysis, after sorting the candidate sample points by particle scattering ratio, the absolute change in particle depolarization ratio is calculated and compared with a preset small fluctuation threshold. If the change in all adjacent sample points is less than this threshold, it is considered a stable or small-scale change trend. This trend corresponds to a microscopic physical state where mature cloud droplets have formed regular spheres, non-spherical characteristics have disappeared, and particle depolarization ratio tends to be constant. Through the dual determination of high-value range and low-fluctuation trend, samples in the mature cloud stage can be accurately identified, completing the stage division of the entire aerosol-cloud droplet conversion process.

[0117] By progressively filtering particle scattering ratios across three threshold ranges from low to high, and combining this with a dual assessment of the differentiated trends in particle depolarization ratios within each range, a multi-dimensional stage identification system based on physical characteristics was constructed. This system not only utilizes the numerical gradient of particle scattering ratios to differentiate the strength of particle scattering contributions, but also precisely captures key nodes in particle morphology evolution through the changing trends of particle depolarization ratios, perfectly matching the microscopic physical essence of the three stages: aerosol hygroscopic growth, initial cloud droplet formation, and mature cloud droplet development. This progressive assessment and conditional coupling approach effectively avoids the limitations of single-parameter assessment, achieving precise division and boundary definition of different stages in the aerosol-cloud droplet transformation process. This ensures the scientific validity and uniqueness of the identification results, providing standardized and refined judgment criteria for in-depth research into the continuous evolution mechanism of aerosol-cloud interactions.

[0118] Furthermore, the steps for identifying the aerosol-cloud droplet conversion stage based on the particle scattering ratio, the particle depolarization ratio, and the joint identification criterion include:

[0119] (1) Match the particle scattering ratio and the particle depolarization ratio with the position of the joint probability density scatter plot;

[0120] Specifically, the particle scattering ratio and particle depolarization ratio of the particulate matter sample to be judged are extracted, and these values ​​are combined as two-dimensional coordinate points. The points are located in the two-dimensional rectangular coordinate system of the generated joint probability density scatter plot. Based on the scale of the horizontal and vertical axes, the geometric position of the two-dimensional coordinate point is found in the coordinate system. The color features of this position in the joint probability density scatter plot and the distribution and clustering of the surrounding sample points are confirmed to complete the accurate position matching between the numerical values ​​and the visualization map.

[0121] Furthermore, when extracting numerical values, it is ensured that the particle scattering ratio and particle depolarization ratio of the sample to be judged are valid data after correction and calculation, and that both belong to the same detection spatiotemporal node. When locating the geometric position, the coordinate scale rules of the joint probability density scatter plot are strictly followed. The particle scattering ratio value determines the horizontal coordinate position, and the particle depolarization ratio value determines the vertical coordinate position, ensuring that the sample point forms a unique spatial position in the map. When confirming color and clustering features, the probability density level to which the location belongs is determined by the color intensity at that location. By observing whether it falls into a high-concentration color cluster area, it is determined whether it belongs to the core sample group of a certain transformation stage. This position matching not only achieves the visualization of numerical values ​​but also provides intuitive auxiliary references for subsequent judgments using the clustering features of the map, reducing the one-sidedness of single numerical judgments.

[0122] (2) Based on the scattering ratio threshold range of the location and the trend of depolarization ratio with scattering ratio, the conversion stage of particulate matter is determined by referring to the judgment conditions in the joint identification criteria.

[0123] Specifically, based on the horizontal coordinate position of the sample point to be judged in the joint probability density scatter plot, the threshold range of its particle scattering ratio is determined; combined with the distribution trend of the surrounding adjacent sample points, the changing trend of particle depolarization ratio with the increase of particle scattering ratio is analyzed; the determined threshold range and changing trend are simultaneously compared with the dual judgment conditions of aerosol stage, nascent cloud stage and mature cloud stage preset in the joint identification criterion, and the unique corresponding condition item is matched to finally determine the transformation stage to which the particulate matter belongs.

[0124] Furthermore, when determining the threshold range, the horizontal axis of the joint probability density scatter plot is divided according to the numerical gradient. The horizontal axis position of the sample point directly corresponds to the first, second, or third threshold range, quickly locking its scattering contribution level. When analyzing the trend, it is no longer limited to the value of a single sample point, but combines the distribution trajectory of surrounding samples in the joint probability density scatter plot. If the sample point falls into the low scattering ratio region and the surrounding points show a downward trend to the right, the corresponding depolarization ratio decreases; if it falls into the medium scattering ratio region and the surrounding points show an upward trend to the right, the corresponding depolarization ratio increases; if it falls into the high scattering ratio region and the surrounding points are horizontally distributed, the corresponding depolarization ratio is stable. Finally, the features of these two dimensions are checked one by one with the joint identification criteria to ensure that the scattering ratio range and trend characteristics of the sample point completely match the judgment conditions of a certain stage, thereby obtaining a unique and accurate transformation stage judgment result, realizing the deep integration of quantitative calculation, visualization analysis, and physical criteria.

[0125] By matching the particle scattering ratio and depolarization ratio with the joint probability density scatter plot, a precise mapping from quantitative data to a visual spectrum was achieved. The color clustering features of the spectrum provided an intuitive spatial reference for stage determination. The scattering ratio threshold range was determined by location, and the depolarization ratio trend was analyzed in conjunction with surrounding distribution trajectories. Finally, the transformation stage was determined by comparing with joint identification criteria, constructing a full-link determination process of numerical positioning, trend analysis, and criterion comparison. This approach retains the accuracy of quantitative parameters while leveraging the spatial distribution characteristics of the visual spectrum to overcome the limitations of single numerical analysis. It achieves rapid, accurate, and unique determination of particulate matter transformation stages, effectively improving the efficiency and reliability of aerosol-cloud droplet transformation process identification and providing an efficient technical path for refined research on atmospheric aerosol-cloud interactions.

[0126] Furthermore, this application will illustrate the solution through a complete embodiment, including the following steps:

[0127] (a) Raw signal processing: The echo signals of the main channel, depolarization channel and molecular channel are acquired synchronously by lidar and processed by background denoising, distance square correction and gain ratio calibration.

[0128] (ii) Scattering-depolarization inversion: Based on this signal, the particle backscattering coefficient and the molecular backscattering coefficient are inverted, and the particle scattering ratio R is calculated. p (Defined as the ratio of particle backscattering to molecular backscattering) and particle depolarization ratio δ p (Defined as the ratio of vertical backscattering to parallel backscattering).

[0129] (III) Probability density estimation: Select the spatiotemporal range of lidar observation with typical "aerosol → cloud" characteristics as the identification domain, and extract the scattering ratio R of all particles in this region. p And particle debias ratio δ p Sample data; and δ within the identification domain p -R p Two-dimensional kernel density estimation is performed on the samples to generate a joint probability density scatter plot. The color bars in the plot reflect the number density of the sample distribution characteristics.

[0130] (iv) Identification of the transformation stage: Based on the sample distribution pattern in the scatter plot, divide the area into "activation zone", "growth zone" and "saturation zone", and establish the following criteria:

[0131] a. If the sample distribution is δ p With R p Increase decrease, R p Lower (usually <10) 0 The region (activation zone) is identified as the aerosol hygroscopic growth stage;

[0132] b. If the sample distribution is δ p With R p Increase remains unchanged, R p The value increases rapidly (usually 10). 0 ~10 2 The area (growth zone) is identified as the initial cloud formation stage;

[0133] c. If the sample distribution is δ p With R p Increase, R p High (usually >10) 2 The region (saturation zone) is identified as the mature cloud multiple scattering stage.

[0134] (v) Results output: Based on the above criteria, output the results of the aerosol-cloud droplet conversion stage;

[0135] Furthermore, in step (i), the lidar is a high-spectral-resolution lidar or a Raman scattering lidar, which has the ability to separate aerosol / cloud droplet scattering and molecular scattering signals;

[0136] In step (ii), the volume scattering ratio R should be strictly distinguished during the inversion stage. v(Defined as the ratio of bulk backscattering to molecular backscattering), bulk depolarization ratio δ v (defined as the ratio of vertical body backscattering to parallel body backscattering) and the R described in this invention p δ p The difference must be clear, otherwise the identification may fail;

[0137] In step (iii), during the identification stage, the identification domain can be cross-analyzed by combining prior meteorological data (such as relative humidity and temperature profiles); the kernel function used for the kernel density estimation has no special restrictions, and the bandwidth parameter can be adaptively determined according to the sample distribution to improve the robustness of the probability density estimation.

[0138] In step (four), R in the "activation zone", "growth zone" and "saturation zone" p δ p The range can be dynamically set according to actual observation needs, supporting real-time online processing and phased trend analysis. The physical basis for dividing the three regions is: ① the typical characteristic of aerosols gradually transforming from non-spherical to spherical during the hygroscopic growth process (activation zone); ② the rapid growth phenomenon of activated particles that have fully absorbed moisture and become spherical after breaking through the critical radius (growth zone); ③ the phenomenon of multiple scattering and depolarization that occurs from the cloud base upwards after the liquid water content of cloud droplets reaches a certain level (saturation zone).

[0139] In step (v), the output results may exist in multiple decision stages in step (iv) at the same time; at the same time, there should be a significant difference in the number of density scatter points in different stages, otherwise the decision domain may be too large.

[0140] Furthermore, Figure 3 This is a schematic diagram illustrating the aerosol-cloud conversion identification result, which is an exemplary embodiment of this application. The method provided in this embodiment specifically includes:

[0141] S1: The lidar in this embodiment is a high-spectral-resolution lidar. The lidar system will receive three signals with different characteristics, and its lidar equations are represented by formulas (1), (2), and (3). Note B || B ⊥ The signal should be calibrated to the system gain ratio. Figure 2 The lidar signal shown is the calibrated complete component signal B=B || +B ⊥ The time-height map is obtained by stitching together different height profiles from lidar information.

[0142] S2: Solving the lidar equations for the above three channels yields β||p and β⊥p, and the particle scattering ratio R observed by the lidar is obtained. p With deflection ratio δ p It is obtained by calculation using formulas (4) and (5).

[0143] S3: To accurately identify the continuous transformation process from aerosol to cloud, it is necessary to select an observational spatiotemporal region with typical dynamic evolution characteristics as the "discrimination domain." This embodiment is based on... Figure 2 The lidar time-elevation map shown selects four regions with different macroscopic characteristics for comparative analysis. This region ranges in altitude from approximately 4.5 to 6 km and covers the time period from 23:30 to 02:00 local time. Four different types of identification domains are selected in the map:

[0144] • Identification domain 1: The signal is relatively strong overall and the structure is relatively uniform.

[0145] • Identification Domain 2: The signal shows a clear trend from weak to strong.

[0146] • Identification domain 3: The signal has obvious local enhancement spots.

[0147] • Identification domain 4: The signal has no enhancement spots, but is slightly stronger than the signal at higher levels.

[0148] Based on macroscopic signal characteristic analysis, the number of cloud droplets in identification domains 1 to 4 shows a decreasing trend. Valid data points are extracted from each identification domain, forming (δ... p , R p The data point set is configured such that the number of data points in identification domains 1 to 4 are 5874, 6408, 2546, and 1800, respectively. Subsequently, a two-dimensional kernel density estimation is performed on the data samples within each identification domain. In this embodiment, a Gaussian kernel function is used to generate δ... p -R p A joint probability density scatter plot is generated, and the density distribution of the probability density is visualized using color bars.

[0149] S4: Transformation Stage Identification: such as Figure 3 As shown, based on the sample distribution pattern and density characteristics in the scatter plot, the following identification results are obtained sequentially:

[0150] • Identification domain 1: Concentrated in the growth region, fully distributed in the saturation region, and only partially distributed in the activation region;

[0151] • Identification Domain 2: Concentrated in the activation region, with a complete distribution in the growth region, and only a partial distribution in the saturation region;

[0152] • Identification domain 3: Concentrated in the activation region, with only partial distribution in the growth region and no distribution in the saturation region;

[0153] • Identification domain 4: Concentrated in the activation region, with no growth or saturation regions distributed there;

[0154] S5: Output of results: as shown Figure 3 As shown, the final output is the conclusion of the aerosol-cloud droplet conversion stage corresponding to each identification domain:

[0155] • Identification Domain 1: The result is nascent cloud → mature cloud;

[0156] • Identification Domain 2: The result is aerosol → nascent cloud → a small amount of mature cloud;

[0157] • Identification Domain 3: The result is aerosols → a small amount of nascent clouds;

[0158] • Identification domain 4: The result is only aerosols.

[0159] The aerosol-cloud droplet transformation stage identification method based on lidar provided in this embodiment achieves objective and quantitative identification of cloud formation stages. It is less affected by local outliers and has clear physical meaning. It can provide effective technical support for the study of cloud and fog microphysical processes and artificial weather modification operations. Through the accurate acquisition and correction calculation of multi-channel echo signals, the core parameters of particle scattering ratio and depolarization ratio that can truly characterize particle scattering and morphological characteristics are obtained. Then, a high-quality identification domain is determined by physical law delimitation and signal-to-noise ratio screening. Combined with kernel density estimation, a joint probability density scatter plot of fused probability density distribution is generated. Based on the microphysical characteristics of each stage of aerosol-cloud droplet transformation, a joint identification criterion coupled with the scattering ratio threshold range and the depolarization ratio change trend is set to achieve accurate stage determination by combining quantitative data and visualized maps. This breakthrough overcomes the limitations of traditional lidar's binary classification of aerosols and clouds, enabling the identification of the complete transformation chain from aerosol hygroscopic growth to nascent cloud formation and mature cloud formation. It provides more refined atmospheric process observation information. Based on two-dimensional joint probability density analysis, it effectively suppresses interference from single-point noise and local outliers, making the identification results more objective, quantitative, and physically meaningful. Furthermore, it can be integrated into existing lidar data processing systems, suitable for various operational and research scenarios such as real-time monitoring, climate change research, and cloud physics parameterization improvement. This provides key technical support for revealing the indirect climate effects of aerosols, improving cloud precipitation forecasting, and enhancing the accuracy of climate model parameterization. It also bridges the technical gap between insufficient spatial coverage of in-situ observations and the coarseness of traditional remote sensing identification, providing an effective technical means for studying the microphysical processes of clouds and fog and for weather modification operations.

[0160] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for identifying the aerosol-cloud droplet conversion stage based on lidar, characterized in that, The method includes: Acquire echo signals from the main channel, depolarization channel, and molecular channel; The particle scattering ratio and particle depolarization ratio are calculated based on the echo signal and the atmospheric molecular scattering model. Determine the identification domain, collect samples of particle scattering ratio and particle depolarization ratio in the identification domain to form a two-dimensional sample set, determine the joint probability density distribution of particle scattering ratio and particle depolarization ratio based on the two-dimensional sample set, and generate a joint probability density scatter plot based on the joint probability density distribution. A joint identification criterion is set based on the microscopic physical process characteristics of each stage of aerosol-cloud droplet conversion. The aerosol-cloud droplet conversion stage is identified by combining the particle scattering ratio and the particle depolarization ratio with the joint identification criterion.

2. The method according to claim 1, characterized in that, The calculation of particle scattering ratio and particle depolarization ratio based on the echo signal and atmospheric molecular scattering model includes: Calculate the particle backscattering coefficient and the molecular backscattering coefficient based on the echo signal; The particle scattering ratio is calculated based on the ratio of the particle backscattering coefficient to the molecular backscattering coefficient; The particle depolarization ratio is calculated based on the ratio of the particle backscattering coefficient of the depolarization channel to the particle backscattering coefficient of the main channel.

3. The method according to claim 2, characterized in that, The calculation of particle backscattering coefficients and molecular backscattering coefficients based on the echo signal includes: Using the echo signal of the molecular channel as a standard reference, the echo signals of the main channel and the depolarization channel are normalized and corrected. Based on the inversion principle of the lidar system and the atmospheric molecular scattering model, the corrected echo signal is solved to obtain the particle backscattering coefficient and the molecular backscattering coefficient.

4. The method according to claim 1, characterized in that, The determined identification domain includes: Determine the evolution of particle properties with altitude and time during the aerosol-cloud droplet conversion process; The initial range is divided according to the evolution law, and the data intervals of the initial range are filtered according to the signal-to-noise ratio threshold to obtain the identification domain.

5. The method according to claim 1, characterized in that, The step of generating a joint probability density scatter plot based on the joint probability density distribution includes: The numerical magnitude of the joint probability density distribution is mapped through a visual color gradient; A coordinate system is constructed with the particle scattering ratio as the abscissa and the particle debiasing ratio as the ordinate. The distribution positions of the sample points are marked in the coordinate system and the sample points are assigned color features that match the probability density to generate a joint probability density scatter plot.

6. The method according to claim 1, characterized in that, The joint identification criteria established based on the microscopic physical process characteristics of each stage of aerosol-cloud droplet conversion include: When the particle scattering ratio is within a first threshold range and the particle depolarization ratio decreases as the particle scattering ratio increases, it is determined to be in the aerosol stage. When the particle scattering ratio is within the second threshold range and the particle depolarization ratio shows an increasing trend with the increase of the particle scattering ratio, it is determined to be the initial cloud stage. When the scattering ratio is within the third threshold range, and the particle depolarization ratio shows a stable or slightly changing trend as the particle scattering ratio increases, it is determined to be a mature cloud stage. The third threshold range is greater than the second threshold range, and the second threshold range is greater than the first threshold range.

7. The method according to claim 1, characterized in that, The identification of the aerosol-cloud droplet conversion stage based on the particle scattering ratio, the particle depolarization ratio, and the joint identification criterion includes: The particle scattering ratio and the particle debiasing ratio are matched with the joint probability density scatter plot for positional matching. Based on the scattering ratio threshold range of the location and the trend of depolarization ratio with scattering ratio, the conversion stage of particulate matter is determined by comparing it with the judgment conditions in the joint identification criteria.

8. The method according to claim 1, characterized in that, The samples that statistically analyze the particle scattering ratio and particle depolarization ratio in the identification domain form a two-dimensional sample set, including: Extract sample data of all particle scattering ratios and particle depolarization ratios in the identification domain; The data are combined into two-dimensional data pairs according to a one-to-one correspondence, and the two-dimensional sample set is constructed based on all the two-dimensional data pairs.

9. The method according to claim 1, characterized in that, The determination of the joint probability density distribution of particle scattering ratio and particle depolarization ratio based on the two-dimensional sample set includes: The samples in the two-dimensional sample set are preprocessed; Two-dimensional statistical analysis was performed on the preprocessed two-dimensional sample set using the kernel density estimation method to obtain the joint probability density distribution.

Images