Artificial intelligence (AI) and machine learning (ML) tools are rapidly growing in capability and application across the weather enterprise. Fully AI-based numerical weather prediction (NWP) emulators are beginning to outperform traditional NWP, and many weather agencies have started to adopt ML-derived guidance products into the forecast process. For example, the United States National Weather Service’s Storm Prediction Center (SPC) has implemented a number of ML models to aid in the prediction and detection of tornadoes, severe wind, hail, and wildfires. However, the development of these AI/ML products and their subsequent transition into SPC operations revealed several challenges which potentially slowed their overall adoption into the forecasters’ workflow. This presentation will discuss several factors that impacted the adoption of AI/ML into forecast operations and highlight some best practices used by SPC to help streamline the research-to-operations transition. Case studies of AI/ML projects that were successfully transitioned into SPC operations will help illustrate the application of these best practices and showcase some of the common pitfalls faced by AI/ML development for operational applications.

The 2025 NOAA Hazardous Weather Testbed Spring Forecasting Experiment What: Over 131 forecasters and researchers engaged in real-time severe weather forecasting and evaluation activities to accelerate research-to-operations for next-generation severe weather prediction technologies and prospective operational regional models. Major emphases of Spring Forecasting Experiment (SFE) 2025 included 1) the Rapid Refresh Forecast System (RRFS) and RRFS Ensemble Forecast System (REFS), 2) the Model for Prediction Across Scales, 3) global and regional artificial intelligence (AI)-based NWP emulators, and 4) new visualizations for the Warn-on-Forecast System (WoFS). When: 28 April-30 May 2025 Where: Norman, Oklahoma, and Online

The 2024 NOAA Hazardous Weather Testbed Spring Forecasting Experiment What: Over 160 forecasters and researchers convened in-person and virtually to engage in real-time severe weather forecasting and evaluation activities aimed at accelerating research-to-operations and informing NOAA's Unified Forecast System. Major emphases of SFE 2024 included 1) deterministic and ensemble components of the Rapid Refresh Forecast System, 2) the Model for Prediction Across Scales, 3) global artificial intelligence (AI)-based NWP emulators, 4) the Warn-on-Forecast System, and 5) innovative AI-based postprocessing strategies. When: 29 April-31 May 2024 Where: Norman, OK, and Online

© 2023 American Meteorological Society. This published article is licensed under the terms of the default AMS reuse license. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses). Corresponding author: Adam J. Clark, adam.clark@noaa.gov

Adam J. Clark2,4, Israel L. Jirak1, Burkely T. Gallo1,3, Kent H. Knopfmeier2,3, Brett Roberts1,2,3, Makenzie Krocak1,3,5, Jake Vancil1,3, Kimberly A. Hoogewind2,3, Nathan A. Dahl1,3, Eric D. Loken2,3,4, David Jahn1,3, David Harrison1,3, David Imy2, Patrick Burke2, Louis J. Wicker2,4, Patrick S. Skinner2,3, Pamela L. Heinselman2,4, Patrick Marsh1, Katie A. Wilson2,3, Andrew R. Dean1, Gerald J. Creager2,3, Thomas A. Jones2,3, Jidong Gao2, Yunheng Wang2,3, Montgomery Flora2,3, Corey K. Potvin2,4, Christopher A. Kerr2,3, Nusrat Yussouf2,3,4, Joshua Martin2,3, Jorge Guerra2,3, Brian C. Matilla2,3, and Thomas J. Galarneau2,3,4

This paper explores artificial intelligent training schemes based on multilayer perceptron, considering back propagation and genetic algorithm (GA). The hybrid scheme is compared with the traditional support vector machine approach in the literature to analyze both fault and normal scenarios of a centrifugal pump. A comparative analysis of the performance of the variables was carried out using both schemes. The study used features extracted for three decomposition levels based on wavelet packet transform. In order to investigate the effectiveness of the extracted features, two mother wavelets were investigated. The salient part of this work is the optimization of the hidden layers numbers using GA. Furthermore, this optimization process was extended to the multilayer perceptron neurons. The evaluation of the model system performance used for the study shows better response of the extracted features, and hidden layers variables including the selected neurons. Moreover, the applied training algorithm used in the work was able to enhance the classifications obtained considering the hybrid artificial intelligent scheme been proposed. This work has achieved a number of contributions like GA-based selection of hidden layers and neuron, applied in neural network of centrifugal pump condition classification. Furthermore, a hybrid training method combining GA and back propagation (BP) algorithms has been applied for condition classification of a centrifugal pump. The obtained results have shown the good ability of the proposed methods and algorithms.

This paper presents a comparative study of two artificial intelligent systems, namely; Multilayer Perceptron (MLP) and support vector machine (SVM), to classify six fault conditions and the normal (nonfaulty) condition of a centrifugal pump. A hybrid training method for MLP is proposed for this work based on the combination of Back Propagation (BP) and Genetic Algorithm (GA). The two training algorithms are tested and compared separately as well. Features are extracted using Discrete Wavelet Transform (DWT), both approximations, details, and two mother wavelets were used to investigate their effectiveness on feature extraction. GA is also used to optimize the number of hidden layers and neurons of MLP. In this study, the feature extraction, GA-based hidden layers, neurons selection, training algorithm, and classification performance, based on the strengths and weaknesses of each method, are discussed. From the results obtained, it is observed that the DWT with both MLP-BP and SVM produces better classification rates and performances.

General comments: I support publication of this manuscript in AMT. The research aligns well with the scope of AMT. The reviewer finds the application of release of latent heat for detecting a freezing event in immersion mode ice spectrometer unique. The authors successfully present the applicability of their technique (IR-NIPI) to characterize immersion freezing efficiencies of three different forms of the sample (incl. chips, powder and ambient particles collected on the filter and scrubbed with water) at T > -22 ◦C. In particular, its applicability to the atmospheric sample seems promising

Abstract. Low concentrations of ice-nucleating particles (INPs) are thought to be important for the properties of mixed-phase clouds, but their detection is challenging. Hence, there is a need for instruments where INP concentrations of less than 0.01 L−1 can be routinely and efficiently determined. The use of larger volumes of suspension in drop assays increases the sensitivity of an experiment to rarer INPs or rarer active sites due to the increase in aerosol or surface area of particulates per droplet. Here we describe and characterise the InfraRed-Nucleation by Immersed Particles Instrument (IR-NIPI), a new immersion freezing assay that makes use of IR emissions to determine the freezing temperature of individual 50 µL droplets each contained in a well of a 96-well plate. Using an IR camera allows the temperature of individual aliquots to be monitored. Freezing temperatures are determined by detecting the sharp rise in well temperature associated with the release of heat caused by freezing. In this paper we first present the calibration of the IR temperature measurement, which makes use of the fact that following ice nucleation aliquots of water warm to the ice–liquid equilibrium temperature (i.e. 0 ∘C when water activity is ∼1), which provides a point of calibration for each individual well in each experiment. We then tested the temperature calibration using ∼100 µm chips of K-feldspar, by immersing these chips in 1 µL droplets on an established cold stage (µL-NIPI) as well as in 50 µL droplets on IR-NIPI; the results were consistent with one another, indicating no bias in the reported freezing temperature. In addition we present measurements of the efficiency of the mineral dust NX-illite and a sample of atmospheric aerosol collected on a filter in the city of Leeds. NX-illite results are consistent with literature data, and the atmospheric INP concentrations were in good agreement with the results from the µL-NIPI instrument. This demonstrates the utility of this approach, which offers a relatively high throughput of sample analysis and access to low INP concentrations.

— Centrifugal pumps are complex machines which can experience different types of fault. Condition monitoring can be used in centrifugal pump fault detection through vibration analysis for mechanical and hydraulic forces. Vibration analysis methods have the potential to be combined with artificial intelligence systems where an automatic diagnostic method can be approached. An automatic fault diagnosis approach could be a good option to minimize human error and to provide a precise machine fault classification. This work aims to introduce an approach to centrifugal pump fault diagnosis based on artificial intelligence and genetic algorithm systems. An overview of the future works, research methodology and proposed experimental setup is presented and discussed. The expected results and outcomes based on the experimental work are illustrated.

The present study provides an overview of the ice nucleation activity of different feldspars varying in their chemical composition and crystal structure. In total 15 different feldspars including plagioclase and alkali feldspars were analyzed for one weight percentage using a special freezing array method called μl-NIPI (microliter Nucleation by Immersed Particle Instrument). Additionally, the samples were characterized with respect to their BET surface area and mineralogical structure at least for the dominating feldspar phase. It was found that K-feldspars generally nucleate ice more efficiently than other types of feldspar except for the two most efficient so-called “hyper-active” alkali feldspars (one of microcline and albite). One of the main statements of the study is that the observed ice nucleation efficiency of different feldspars varies. This effect was less pronounced in the case of K-feldspars. I consider this as the key information of this study. The influence of particle aging in water on the ice nucleation ability of

fate of fluorescence labeled stem cells was also followed in colon tissues.Results: Lethargy was significantly decreased in the group receiving stem cells in comparison with IEC-6 or saline controls.The area of colonic lesions was 50% lower than controls and colon wet weight were 30% less than controls.The body weight loss and severity of colitis in these groups also showed significant improvement.Colon dilation, thickness and pericolonic adhesions were significantly decreased after stem cell treatment.Conclusions: 1) Local i.c.administration of bone marrow-derived stem cells accelerated the healing of ulcerative colitis induced by iodoacetamide.2) This beneficial effect is specific to stem cells, as intestinal epithelial cell IEC-6 had no effect.3) Thus, local stem cell therapy seems to be a new option to achieve a rapid healing of experimental ulcerative colitis, possibly by restoring epithelial barrier integrity and promoting angiogenesis.