Evaluation of a high‐resolution numerical weather prediction model's simulated clouds using observations from CloudSat, GOES‐13 and <i>in situ</i> aircraft

Qu, Zhipeng; Barker, Howard; Korolev, Alexei; Milbrandt, Jason A.; Heckman, Ivan; Bélair, Stéphane; Leroyer, Sylvie; Vaillancourt, Paul; Wolde, Mengistu; Schwarzenböck, Alfons; Leroy, Delphine; Strapp, J. W.; Cole, Jason N. S.; Nguyen, Louis; Heidinger, A. K.

doi:10.1002/qj.3318

Cited by 18 publications

(23 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The model interest indicates an awareness of the storm, although the break in high model interest from 1500 to 1550 UTC suggests that placement and/or horizontal extent of the convective activity was inaccurate. This result is consistent with findings of Qu et al (2018), who found differences between simulated and observed cloud amounts using the Canadian Global Environmental Multiscale model to analyze an MCS near French Guiana on 16 May 2015 during the Cayenne field campaign.…”

Section: Fig 15supporting

confidence: 92%

Development of a Method to Detect High Ice Water Content Environments Using Machine Learning

Haggerty

Rugg

Potts

et al. 2020

Journal of Atmospheric and Oceanic Technology

Self Cite

View full text Add to dashboard Cite

This paper describes development of a method for discriminating high ice water content (HIWC) conditions that can disrupt jet-engine performance in commuter and large transport aircraft. Using input data from satellites, numerical weather prediction models, and ground-based radar, this effort employs machine learning to determine optimal combinations of available information using fuzzy logic. Airborne in situ measurements of ice water content (IWC) from a series of field experiments that sampled HIWC conditions serve as training data in the machine-learning process. The resulting method, known as the Algorithm for Prediction of HIWC Areas (ALPHA), estimates the likelihood of HIWC conditions over a three-dimensional domain. Performance statistics calculated from an independent subset of data reserved for verification indicate that the ALPHA has skill for detecting HIWC conditions, albeit with significant false alarm rates. Probability of detection (POD), probability of false detection (POFD), and false alarm ratio (FAR) are 86%, 29% (60% when IWC below 0.1 g m−3 are omitted), and 51%, respectively, for one set of detection thresholds using in situ measurements. Corresponding receiver operating characteristic (ROC) curves give an area under the curve of 0.85 when considering all data and 0.69 for only points with IWC of at least 0.1 g m−3. Monte Carlo simulations suggest that aircraft sampling biases resulted in a positive POD bias and the actual probability of detection is between 78.5% and 83.1% (95% confidence interval). Analysis of individual case studies shows that the ALPHA output product generally tracks variation in the measured IWC.

show abstract

Section: Fig 15supporting

confidence: 92%

Development of a Method to Detect High Ice Water Content Environments Using Machine Learning

Haggerty

Rugg

Potts

et al. 2020

Journal of Atmospheric and Oceanic Technology

Self Cite

View full text Add to dashboard Cite

show abstract

“…18. Fragments of frozen drops were also documented through in situ observations reported by Korolev et al (2004) and Rangno (2008).…”

Section: Droplet Fragmentation/shattering During Freezingmentioning

confidence: 99%

“…In these studies, the conclusions about the HM process were obtained based on the observed association with graupel and columnar ice crystals. Fewer studies attributed observations of high ice concentration to drop shattering (e.g., Koenig 1963Koenig , 1965Braham, 1964;Rangno, 2008;Lawson et al, 2017). Ice-ice collisional fragmentation was identified as a source of SIP in natural clouds by Hobbs and Farber (1972), Takahashi (1993), andSchwarzenboeck et al (2009).…”

Section: Introductionmentioning

confidence: 99%

A new look at the environmental conditions favorable to secondary ice production

et al. 2020

Self Cite

View full text Add to dashboard Cite

Abstract. This study attempts a new identification of mechanisms of secondary ice production (SIP) based on the observation of small faceted ice crystals (hexagonal plates or columns) with typical sizes smaller than 100 µm. Due to their young age, such small ice crystals can be used as tracers for identifying the conditions for SIP. Observations reported here were conducted in oceanic tropical mesoscale convective systems (MCSs) and midlatitude frontal clouds in the temperature range from 0 to −15 ∘C and heavily seeded by aged ice particles. It was found that in both MCSs and frontal clouds, SIP was observed right above the melting layer and extended to higher altitudes with colder temperatures. The roles of six possible mechanisms to generate the SIP particles are assessed using additional observations. In most observed SIP cases, small secondary ice particles spatially correlated with liquid-phase, vertical updrafts and aged rimed ice particles. However, in many cases, neither graupel nor liquid drops were observed in the SIP regions, and therefore, the conditions for an active Hallett–Mossop process were not met. In many cases, large concentrations of small pristine ice particles were observed right above the melting layer, starting at temperatures as warm as −0.5 ∘C. It is proposed that the initiation of SIP above the melting layer is stimulated by the recirculation of large liquid drops through the melting layer with convective turbulent updrafts. After re-entering a supercooled environment above the melting layer, they impact with aged ice, freeze, and shatter. The size of the splinters generated during SIP was estimated as 10 µm or less. A principal conclusion of this work is that only the freezing-drop-shattering mechanism could be clearly supported by the airborne in situ observations.

show abstract

“…Conversely, and the focus of the present study, large‐eddy simulations, cloud system‐resolving models (CSRMs), and limited‐area numerical weather prediction (NWP) models often use Δ x ≪ 1 km with domains containing well over 10 6 columns (Stevens et al, ; Bretherton and Khairoutdinov, ; Milbrandt et al, ). Their 1D RTMs operate directly on well‐resolved cloud fields and make relatively few, if any, assumptions about subgrid‐scale conditions (Qu et al, ). This approach to modelling radiative transfer is known as the Independent Column Approximation (ICA: e.g.…”

Section: Introductionmentioning

confidence: 99%

Accelerating radiative transfer calculations for high‐resolution atmospheric models

Barker

2019

Quart J Royal Meteoro Soc

Self Cite

View full text Add to dashboard Cite

This study presents a method for calculating all‐sky, broadband radiative fluxes for high‐resolution atmospheric model domains, such as limited‐area numerical weather prediction models and Cloud System‐Resolving Models, that have many columns (i.e. N ≫ 105). The aim is to replace the Independent Column Approximation (ICA) which applies 1D radiative transfer models (RTMs) to all N columns; but usually after skipping many dynamical time steps due to the ICA's computational burden. The proposed method, referred to as Partitioned Gauss–Legendre Quadrature (PGLQ), begins by partitioning a domain into M sub‐domains. This study is concerned just with cloudy columns so partitioning was done according to cloud‐top altitudes. Then, for each sub‐domain, sort columns from smallest to largest cloud water path W, assign each column a sequence number, identify nG GLQ points, and apply 1D RTMs to the nG columns with resulting flux profiles assigned to nearby columns in the sorted sequence. Last, reposition columns back to the full domain. Marine boundary‐layer and deep convective cloud fields were used to assess PGLQ. For ratios of RTM calculations between ICA and PGLQ of fRT = N/(nG · M)−1 ∈ [103, 104], MBEs and RMSEs for net short‐wave (SW) surface fluxes FnormalSWNETSFC scale close to (nG · M)−1, RMSEs being typically ∼200× smaller than domain‐average FnormalLWNETSFC. Structure function analyses of FnormalSWNETSFC indicate that PGLQ errors resemble white noise. Long‐wave (LW) fluxes perform less well due to sorting on W but seem likely to be acceptable, RMSEs being ∼5–10× smaller than domain‐average FnormalSWNETSFC. All MBEs are typically <±0.5 W/m2. Estimates of radiative heating rate profiles are unbiased, and while noisy at small scales, they are portrayed well at scales larger than typical individual clouds. Overall speed‐up of PGLQ relative to ICA, for serial computation as consider here, is expected to be >300×; likely to be comparable to parallel computation. This should allow application of RTMs at every dynamical time step.

show abstract

Evaluation of a high‐resolution numerical weather prediction model's simulated clouds using observations from CloudSat, GOES‐13 and in situ aircraft

Cited by 18 publications

References 38 publications

Development of a Method to Detect High Ice Water Content Environments Using Machine Learning

Development of a Method to Detect High Ice Water Content Environments Using Machine Learning

A new look at the environmental conditions favorable to secondary ice production

Accelerating radiative transfer calculations for high‐resolution atmospheric models

Contact Info

Product

Resources

About