Shape Classification of Cloud Particles Recorded by the 2D-S Imaging Probe Using a Convolutional Neural Network

The airborne two-dimensional stereo(2D-S) optical array probe has been operating for more than 10 yr, accumulating a large amount of cloud particle image data. However, due to the lack of reliable and unbiased classification tools,our ability to extract meaningful morphological information related to cloud microphysical processes is limited. To solve this issue, we propose a novel classification algorithm for 2D-S cloud particle images based on a convolutional neural network(CNN), named CNN-2DS. A 2D-S cloud particle shape dataset was established by using the 2D-S cloud particle images observed from 13 aircraft detection flights in 6 regions of China(Northeast, Northwest, North,East, Central, and South China). This dataset contains 33,300 cloud particle images with 8 types of cloud particle shape(linear, sphere, dendrite, aggregate, graupel, plate, donut, and irregular). The CNN-2DS model was trained and tested based on the established 2D-S dataset. Experimental results show that the CNN-2DS model can accurately identify cloud particles with an average classification accuracy of 97%. Compared with other common classification models [e.g., Vision Transformer(ViT) and Residual Neural Network(ResNet)], the CNN-2DS model is lightweight(few parameters) and fast in calculations, and has the highest classification accuracy. In a word, the proposed CNN-2DS model is effective and reliable for the classification of cloud particles detected by the 2D-S probe.

Identifying Supercooled Liquid Water in Cloud Based on Airborne Observations: Correlation of Cloud Particle Number Concentration with Icing Probability and Proportion of Spherical Particles

Identifying supercooled liquid water(SLW)in clouds is critical for weather modification,aviation safety,and atmospheric radiation calculations.Currently,aircraft identification in the SLW area mostly depends on emprical estimation of cloud particle number concentration(N_(c))in China,and scientific verification and quantitative identification criteria are urgently needed.In this study,the observations are from the Fast Cloud Droplets Probe,Rosemount ice detector(RICE),and Cloud Particle Imager(CP_(i))onboard a King Air aircraft during seven flights in 2018 and 2019 over central and eastern China.Based on this,the correlation among N_(c),the proportion of spherical particles(P_(s)),and the probability of icing(P_(i))in supercooled stratiform and cumulus-stratus clouds is statistically analyzed.Subsequently,this study proposes a method to identify SLW areas using N_(c) in combination with ambient temperature.The reliability of this method is evaluated through the true skill statistics(TSS)and threat score(TS)methods.Numerous airborne observations during the seven flights reveal a strong correlation among Nc,P_(s),and P_(i)at the temperature from 0 to−18°C.When Nc is greater than a certain threshold of 5 cm^(−3),there is always the SLW,i.e.,P_(i)and P_(s)are high.Evaluation results demonstrate that the TSS and TS values for Nc=5 cm^(−3)are higher than those for Nc<5 cm^(−3),and a larger Nc threshold(>5 cm^(−3))corresponds to a higher SLW identification hit rate and a higher SLW content.Therefore,Nc=5 cm^(−3)can be used as the minimum criterion for identifying the SLW in clouds at temperature lower than 0°C.The SLW identification method proposed in this study is especially helpful in common situations where aircraft are equipped with only Nc probes and without the CP_(i)and RICE.

Modeling of Motion of Particle Clouds Formed by Dumping Dredged Material

The motion of particle clouds formed by dumping dredged material into quiescent waters is experimentally and numerically studied. In the numerical model, the particle phase is modeled by the dispersion model, and turbulence is calculated by the large eddy simulation. The governing equations, including the filtered Navier-Stokes equations and mass transport equation, are solved based on the operator-splitting algorithm and an implicit cubic spline interpolation scheme. The eddy viscosity is evaluated by the modified Smagorinsky model including the buoyancy term. Comparisons of main flow characteristics, including shape, size, average density excess, moving speed and the amount of particles deposited on the bed, between experimental and computational results show that the numerical model well predicts the motion of the cloud from the falling to spreading stage. The effects of silt-fence on the motion of the particle cloud are also investigated.

Experiment Study of the Evolution of Coral Sand Particle Clouds in Water

The motion of particle clouds(i.e.,sediment clouds)usually can be found in engineering applications such as wastewater discharge,land reclamation,and marine bed capping.In this paper,a series of laboratory tests are conducted on coral sand to investigate the shape feature of the single particle and the mixing processes of the coral sand particle clouds.The shape of coral sand particle is measured and quantified.The experimental results demonstrate that the shape of coral sand particles tends to be spherical as the particle size decreases,and empirical equations were established to explain the variation of D50 and fS,50 of coral sand.Compared with the silica sand,the evolution of the coral sand particle cloud still experiences three stages,but the threshold for the Reynolds number of particle clouds entering the next stage changes.Further,the normalized axial distance of the coral sand particle clouds is 58%smaller.The frontal velocity exhibits similar varying tendency for the coral sand particle cloud.Considering the difference in shape between coral sand particles and silica sand particles,a semi-empirical formula was proposed based on the original silica sand prediction formula by adding the shape factor and the experimental data of 122μm≤D_(50)≤842μm.It can predict the frontal velocity of the coral sand particle clouds.

Service composition based on discrete particle swarm optimization in military organization cloud cooperation

This paper addresses the problem of service composition in military organization cloud cooperation（MOCC）. Military service providers（MSP） cooperate together to provide military resources for military service users（MSU）. A group of atom services, each of which has its level of quality of service（QoS）, can be combined together into a certain structure to form a composite service. Since there are a large number of atom services having the same function, the atom service is selected to participate in the composite service so as to fulfill users＇ will. In this paper a method based on discrete particle swarm optimization（DPSO） is proposed to tackle this problem. The method aims at selecting atom services from service repositories to constitute the composite service, satisfying the MSU＇s requirement on QoS. Since the QoS criteria include location-aware criteria and location-independent criteria, this method aims to get the composite service with the highest location-aware criteria and the best-match location-independent criteria. Simulations show that the DPSO has a better performance compared with the standard particle swarm optimization（PSO） and genetic algorithm（GA）.

Study on Physical Characteristics of a Precipitation Cloud System in Hebei Province in Spring by Aircraft Observation

Using data of airborne particle measurement system, weather radar and Ka-band millimeter wave cloud-meter, physical structure characteristics of a typical stable stratiform cloud in Hebei Province on February 27, 2018 was analyzed. Research results showed that the detected cloud system was the precipitation stratiform cloud in the later stage of development. The cloud layer developed stably, and the vertical structure was unevenly distributed. The concentration of small cloud particles in high-level clouds was low, and it fluctuated greatly in space, and presented a discontinuous distribution state. The concentration of large cloud particles and precipitation particles was high, which was conducive to the growth of cloud droplets and the aggregation of ice crystals. The concentration of small cloud particles and the content of supercooled water were high in the middle and low-level clouds. The precipitation cloud system had a significant hierarchical structure, which conformed to the "catalysis-supply" mechanism. From the upper layer to the lower layer, the cloud particle spectrum was mainly in the form of single peak or double peak distribution, which showed a monotonic decreasing trend in general. The spectral distribution of small cloud particles in the cloud was discontinuous, and the high-value areas of spectral concentration of large cloud particles and precipitation particles were concentrated in the upper part of the cloud layer, and the particle spectrum was significantly widened. There was inversion zone at the bottom of the cloud layer, which was conducive to the continuous increase of particle concentration and the formation of large supercooled water droplets.

Revisiting the size of nonspherical particles recorded by optical array probes with a new method based on the convex hull

In recent years,the Cloud Imaging Probe(CIP)and Precipitation Imaging Probe(PIP)produced by Droplet Measurement Technologies(DMT)have been introduced by a number of meteorological research and operation centers in China.The supporting software provided by DMT,i.e.,PADS(Particle Analysis and Display System),cannot output detailed information on each individual particle,which definitely limits the in-depth utilization of cloud and precipitation particle image data in China.In this paper,particle-by-particle information was extracted by decompressing the CIP and PIP original particle image data,based on which a new definition of the dimension for nonspherical particles is proposed by using the area of the convex hull enclosing a particle to obtain the equivalent diameter of a circle with equal area.Based on the data detected during one flight in Inner Mongolia,the particle size distribution obtained using the new particle size definition and that used by the other four existing definitions are compared.The results show that the particle number concentration calculated using different particle size definitions can vary by up to an order of magnitude.The result obtained based on the new particle size definition is closest to that calculated with the area-equivalent diameter definition.

Observational Evidence of High Ice Concentration in a Shallow Convective Cloud Embedded in Stratiform Cloud over North China

In this study we observed the microphysical properties, including the vertical and horizontal distributions of ice particles,liquid water content and ice habit, in different regions of a slightly supercooled stratiform cloud. Using aircraft instrument and radar data, the cloud top temperature was recorded as higher than -15℃, behind a cold front, on 9 September 2015 in North China. During the flight sampling, the high ice number concentration area was located in the supercooled part of a shallow convective cloud embedded in a stratiform cloud, where the ambient temperature was around -3℃. In this area,the maximum number concentrations of particles with diameter greater than 100 μm and 500 μm（N_（100） and N_（500）） exceeded 300 L-（-1） and 30 L-（-1）, respectively, and were related to large supercooled water droplets with diameter greater than 24 μm derived from cloud–aerosol spectrometer probe measurements. The ice particles types in this region were predominantly columnar, needle, graupel, and some freezing drops, suggesting that the occurrence of high ice number concentrations was likely related to the Hallett–Mossop mechanism, although many other ice multiplication processes cannot be totally ruled out.The maximum ice number concentration obtained during the first penetration was around two to three orders of magnitude larger than that predicted by the Demott and Fletcher schemes when assuming the cloud top temperature was around-15℃.During the second penetration conducted within the stratiform cloud, N_（100） and N_（500） decreased by a factor of five to ten, and the presence of columnar and needle-like crystals became very rare.

EXPERIMENTAL STUDY ON INSTANTANEOUS DISCHARGE OF UNSORTED PARTICLE CLOUD IN CROSS-FLOW

A series of extensive laboratory experiments were conducted to investigate the transport and mixing of instantaneous discharge of unsorted particle cloud in cross-flow. The following experimental results were obtained： （1） the vertical frontal position and the longitudinal width of the particle cloud in the cross-flow is much larger than those in stagnant water, （2） the smaller cross-flow velocity will normally cause the larger non-dimensional frontal position for the same particle size range and initial volume, （3） the non-dimensional longitudinal width of the particle cloud in the cross-flow increases with the increase of cross-flow velocity. The empirical constants （α1, α3 and α2） and their variance range, which can be used to determine the relationships of non-dimensional vertical frontal position and time, longitudinal width and time, and vertical frontal position and longitudinal width were also obtained through the analysis of experimental data.

Detailed modeling of aluminum particle combustion——From single particles to cloud combustion in Bunsen flames

A numerical model for aluminum cloud combustion which includes the effects of interphase heat transfer,phase change,heterogeneous surface reactions,homogeneous combustion,oxide cap growth and radiation within the Euler–Lagrange framework is proposed.The model is validated in single particle configurations with varying particle diameters.The combustion process of a single aluminum particle is analyzed in detail and the particle consumption rates as well as the heat release rates due to the various physical/chemical sub-models are presented.The combustion time of single aluminum particles predicted by the model are in very good agreement with empirical correlations for particles with diameters larger than 10μm.The prediction error for smaller particles is noticeably reduced when using a heat transfer model that is capable of capturing the transition regime between continuum mechanics and molecular dynamics.The predictive capabilities of the proposed model framework are further evaluated by simulating the aluminum/air Bunsen flames of Mc Gill University for the first time.Results show that the predicted temperature distribution of the flame is consistent with the experimental data and the double-front structure of the Bunsen flame is reproduced well.The burning rates of aluminum in both single particle and particle cloud configurations are calculated and compared with empirical correlations.Results show that the burning rates obtained from the present model are more reasonable,while the correlations,when embedded in the Euler–Lagrange context,tend to underestimate the burning rate in the combustion stage,particularly for the considered fuel-rich flames.

THE MICROPHYSICAL STRUCTURE OF SNOW CLOUDS AND THE GROWTH PROCESS OF SNOW PARTICLES IN WINTER IN XINJIANG

The microphysical structure of snow clouds and the growth process of snow crystals were observed by means of instrumented aircraft, weather radar, snow crystal observations etc. in Urumqi region during the winter of 1982. The analysis of three cases show that about 70% of snow mass growth is produced in the lower layer below 2000 m under the cold front, and that the concentration of ice crystals is as high as 60 L^(-1) and the supercooled water is absent in lower clouds. We may infer that the deposition of ice crystals and the aggregation of snow crystals are important processes for the snow development. The microphysical structure of the snow band near the front aloft and its characteristics as a seeder cloud are also described in this paper.