Search

The evaluation and quantification of Southern Ocean cloud–radiation interactions simulated by climate models are essential in understanding the sources and magnitude of the radiative bias that persists in climate models for this region. To date, most evaluation methods focus on specific synoptic or cloud-type conditions that do not consider the entirety of the Southern Ocean's cloud regimes at once. Furthermore, it is difficult to directly quantify the complex and non-linear role that different cloud properties have on modulating cloud radiative effect. In this study, we present a new method of model evaluation, using machine learning that can at once identify complexities within a system and individual contributions. To do this, we use an XGBoost (eXtreme Gradient Boosting) model to predict the radiative bias within a nudged version of the Australian Community Climate and Earth System Simulator – Atmosphere-only model, using cloud property biases as predictive features. We find that the XGBoost model can explain up to 55 % of the radiative bias from these cloud properties alone. We then apply SHAP (SHapley Additive exPlanations) feature importance analysis to quantify the role each cloud property bias plays in predicting the radiative bias. We find that biases in the liquid water path are the largest contributor to the cloud radiative bias over the Southern Ocean, though important regional and cloud-type dependencies exist. We then test the usefulness of this method in evaluating model perturbations and find that it can clearly identify complex responses, including cloud property and cloud-type compensating errors. [ABSTRACT FROM AUTHOR]

Large hail events are typically infrequent, with significant time gaps between occurrences at specific locations. However, when these events do happen, they can cause rapid and substantial economic losses within a matter of minutes. Therefore, it is crucial to have the ability to accurately observe and understand hail phenomena to improve the mitigation of this impact. While in situ observations are accurate, they are limited in number for an individual storm. Weather radars, on the other hand, provide a larger observation footprint, but current radar-derived hail size estimates exhibit low accuracy due to horizontal advection of hailstones as they fall, the variability of hail size distributions (HSDs), complex scattering and attenuation, and mixed hydrometeor types. In this paper, we propose a new radar-derived hail product developed using a large dataset of hail damage insurance claims and radar observations. We use these datasets coupled with environmental information to calculate a hail damage estimate (HDE) using a deep neural network approach aiming to quantify hail impact, with a critical success index of 0.88 and a coefficient of determination against observed damage of 0.79. Furthermore, we compared HDE to a popular hail size product (MESH), allowing us to identify meteorological conditions that are associated with biases on MESH. Environments with relatively low specific humidity, high CAPE and CIN, low wind speeds aloft, and southerly winds at the ground are associated with a negative MESH bias, potentially due to differences in HSD, hail hardness, or mixed hydrometeors. In contrast, environments with low CAPE, high CIN, and relatively high specific humidity aloft are associated with a positive MESH bias. [ABSTRACT FROM AUTHOR]

Clouds over the Southern Ocean can contain super-cooled liquid water, which means the water remains liquid below 0°C. This is because the water in these clouds has had little contact with pollution or dust, which can help water freeze. Super-cooled liquid water clouds reflect more sunlight back into space than ice clouds, which can affect the temperature of the ocean. The Southern Ocean is a unique environment with fewer ice nucleating particles, making it difficult to study. Scientists are conducting research expeditions to gather more data and improve climate models. [Extracted from the article]

Weather radars are increasingly being used to study the interaction between wildfires and the atmosphere, owing to the enhanced spatio-temporal resolution of radar data compared to conventional measurements, such as satellite imagery and in situ sensing. An important requirement for the continued proliferation of radar data for this application is the automatic identification of fire-generated particle returns (pyrometeors) from a scene containing a diverse range of echo sources, including clear air, ground and sea clutter, and precipitation. The classification of such particles is a challenging problem for common image segmentation approaches (e.g. fuzzy logic or unsupervised machine learning) due to the strong overlap in radar variable distributions between each echo type. Here, we propose the following two-step method to address these challenges: (1) the introduction of secondary, texture-based fields, calculated using statistical properties of gray-level co-occurrence matrices (GLCMs), and (2) a Gaussian mixture model (GMM), used to classify echo sources by combining radar variables with texture-based fields from (1). Importantly, we retain all information from the original measurements by performing calculations in the radar's native spherical coordinate system and introduce a range-varying-window methodology for our GLCM calculations to avoid range-dependent biases. We show that our method can accurately classify pyrometeors' plumes, clear air, sea clutter, and precipitation using radar data from recent wildfire events in Australia and find that the contrast of the radar correlation coefficient is the most skilful variable for the classification. The technique we propose enables the automated detection of pyrometeors' plumes from operational weather radar networks, which may be used by fire agencies for emergency management purposes or by scientists for case study analyses or historical-event identification. [ABSTRACT FROM AUTHOR]

Large hail events are typically infrequent, with significant time gaps between occurrences at specific locations. However, when these events do happen, they can cause rapid and substantial economic losses within a matter of minutes. Therefore, it is crucial to have the ability to accurately observe and understand hail phenomena to improve the mitigation of this impact. While in-situ observations are accurate, they are limited in number for an individual storm. Weather radars, on the other hand, provide a larger observation footprint, but current radar-derived hail size estimates exhibit low accuracy due to horizontal advection of hailstones as they fall, the variability of hail size distributions (HSD), complex scattering and attenuation, and mixed hydrometeor types. In this paper, we propose a new radar-derived hail product that is developed using a large dataset of hail damage insurance claims and radar observations. We use these datasets coupled with environmental information to calculate a Hail Damage Estimate (HDE) using a deep neural network approach aiming to quantify hail impact, with a critical success index of 0.88 and a coefficient of determination against observed damage of 0.78. Furthermore, we compared HDE to a popular hail size product (MESH), allowing us to identify meteorological conditions that are associated with biases on MESH. Environments with relatively low specific humidity, high CAPE and CIN, low wind speeds aloft and southerly winds at ground are associated with a negative MESH bias, potentially due to differences in HSD or mixed hydrometeors. In contrast, environments with low CAPE, high CIN, and relatively high specific humidity aloft are associated with a positive MESH bias. [ABSTRACT FROM AUTHOR]

The evaluation and quantification of Southern Ocean cloud-radiation interactions simulated by climate models is essential in understanding the sources and magnitude of the radiative bias that persists in climate models for this region. To date, most evaluation methods focus on specific synoptic or cloud type conditions and are unable to quantitatively define the impact of cloud properties on the radiative bias whilst considering the system as a whole. In this study, we present a new method of model evaluation, using machine learning, that can at once identify complexities within a system and individual contributions. To do this, we use an XGBoost model to predict the radiative bias within a nudged version of the Australian Community Climate and Earth System Simulator -- Atmosphere-only Model, using cloud property biases as predictive features. We find that the XGBoost model can explain up to 55% of the radiative bias from these cloud properties alone. We then apply SHapley Additive exPlanations feature importance analysis to quantify the role each cloud property bias plays in predicting the radiative bias. We find that biases in liquid water path is the largest contributor to the cloud radiative bias over the Southern Ocean, though important regional and cloud-type dependencies exist. We then test the usefulness of this method in evaluating model perturbations and find that it can clearly identify complex responses, including cloud property and cloud-type compensating errors. [ABSTRACT FROM AUTHOR]

We use dual-polarization C-band data collected in the Southern Ocean to examine the properties of snow observed during a voyage in the austral summer of 2018. Using existing forward modeling formalisms based on an assumption of Rayleigh scattering by soft spheroids, an optimal estimation algorithm is implemented to infer snow properties from horizontally polarized radar reflectivity, the differential radar reflectivity, and the specific differential phase. From the dual-polarization observables, we estimate ice water content qi, the mass-mean particle size Dm, and the exponent of the mass–dimensional relationship bm that, with several assumptions, allow for evaluation of snow bulk density, and snow number concentration. Upon evaluating the uncertainties associated with measurement and forward model errors, we determine that the algorithm can retrieve qi, Dm, and bm within single-pixel uncertainties conservatively estimated in the range 120%, 60%, and 40%, respectively. Applying the algorithm to open-cellular convection in the Southern Ocean, we find evidence for secondary ice formation processes within multicellular complexes. In stratiform precipitation systems we find snow properties and infer processes that are distinctly different from the shallow convective systems with evidence for riming and aggregation being common. We also find that embedded convection within the frontal system produces precipitation properties consistent with graupel. Examining 5 weeks of data, we show that snow in open-cellular cumulus has higher overall bulk density than snow in stratiform precipitation systems with implications for interpreting measurements from space-based active remote sensors. [ABSTRACT FROM AUTHOR]

Over the remote Southern Ocean, cloud feedbacks contribute substantially to Earth system model (ESM) radiative biases. The evolution of low Southern Ocean clouds (cloud top heights < ~ 3 km) is strongly modulated by precipitation and/or evaporation, which act as the primary sink of cloud condensate. Constraining precipitation processes in ESMs requires robust observations suitable for process-level evaluations. A year-long subset (April 2016 - March 2017) of ground-based profiling instrumentation deployed during the Macquarie Island Cloud and Radiation Experiment (MICRE) field campaign (54.5° S, 158.9° E) combines a 95 GHz (W-band) Doppler cloud radar, two lidar ceilometers, and balloon-borne soundings to quantify the occurrence frequency of precipitation from liquid-phase cloud base. Liquid-based clouds at Macquarie Island precipitate ~ 70 % of the time, with deeper and colder clouds precipitating more frequently and at a higher intensity compared to thinner and warmer clouds. Supercooled cloud layers precipitate more readily than layers with cloud top temperatures > 0 °C, regardless of the geometric thickness of the layer, and also evaporate more frequently. We further demonstrate an approach to employ these observational constraints for evaluation of a 9-year GISS-ModelE3 ESM simulation. Model output is processed through the Earth Model Column Collaboratory (EMC²) radar and lidar instrument simulator with the same instrument specifications as those deployed during MICRE, therefore accounting for instrument sensitivities and ensuring a coherent comparison. Relative to MICRE observations, the ESM produces a smaller cloud occurrence frequency, smaller precipitation occurrence frequency, and greater sub-cloud evaporation. The lower precipitation occurrence frequency by the ESM relative to MICRE contrasts with numerous studies that suggest a ubiquitous bias by ESMs to precipitate too frequently over the SO when compared with satellite-based observations, likely owing to sensitivity limitations of space-borne instrumentation and different sampling methodologies for ground- versus space-based observations. Despite these deficiencies, the ESM reproduces the observed tendency for deeper and colder clouds to precipitate more frequently and at a higher intensity. The ESM also reproduces specific cloud regimes, including near-surface clouds that account for ~ 25 % of liquid-based clouds during MICRE and optically thin, non-precipitating clouds that account for ~ 27 % of clouds with bases higher than 250 m. We suggest that the demonstrated framework, which merges observations with appropriately constrained model output, is a valuable approach to evaluate processes responsible for cloud radiative feedbacks in ESMs. [ABSTRACT FROM AUTHOR]