Adonis Ferreira Raiol Leal

Past field experiments have demonstrated multiple cloud seeding techniques for modifying the electrical characteristics of developing thunderstorms with the purpose of reducing or eliminating lightning strikes. One motivation for these weather modification research programs was the prospect of preventing wildfires in inaccessible locations or particularly hazardous conditions by temporarily suppressing a prolific ignition source: lightning. It has been nearly 50 years since the last large-scale field effort in lightning suppression, and the ensuing five decades of technological innovation hold the promise of supporting this novel hazard mitigation approach. This study outlines observational and forecasting requirements, as well as an associated remote sensing architecture, that would support an operational wildfire prevention program based on lightning suppression by chaff seeding clouds. Two capabilities enabled by remote sensing observations are highlighted: (1) identifying potential regions of extreme wildfire behavior based on wildland fuels, topography, and weather, and (2) predicting regions where dry lightning strikes are probable. In combination, the proposed nowcasting system would predict the most likely areas of lightning-initiated wildfire danger. Dependent on land management strategies, surrounding infrastructure, firefighting capacity, and risk to human wellbeing, this information could be used to deploy lightning suppression technology to reduce or temporarily prevent wildfire ignitions in these high-risk situations.

Simultaneous observations of ultraviolet (337 nm) and near‐infrared (777 nm) lightning emissions provide insight into streamer and leader processes. However, coordinated ground‐based measurements of these bands during return strokes remain scarce. We present multiband optical and broadband radio (0.1 Hz–5 MHz) observations of 11 cloud‐to‐ground flashes (0.6–4.2 km range). For first return strokes within the photometer field of view, the 337 nm peak precedes the 777 nm peak by 38 μs on average, and 45 μs for subsequent strokes. The 337 nm emission shows faster rise times (∼5 μs) than 777 nm (∼30–40 μs) and correlates more strongly with peak current ( R ² = 0.62 vs. 0.25). Temperature estimates indicate a hotter channel at the 777 nm maximum. Our findings demonstrate that near‐UV emissions provide a sensitive ground‐based diagnostic of the lightning attachment process and may offer improved optical proxies for estimating peak current.

This study presents a methodology for classifying lightning recorded in the Very Low Frequency (VLF) band (7–15 kHz) at low sampling rates, using a Multi-Layer Perceptron (MLP) neural network. The approach aims to minimize data footprint and computational cost while maintaining high classification performance, enabling deployment on compact and energy-efficient hardware. A dataset of lightning events recorded by the Lightning Discharge and Waveform Storage System (LDWSS) in Brazil and the United States was used, encompassing five lightning types: -CG, +CG, –CID, +CID, and IC. Models trained on both wideband (160 Hz–500 kHz) and VLF data achieved comparable accuracy (99.2% and 97.7%, respectively), demonstrating that narrower band systems are suitable for lightning type classification with Machine Learning (ML). Downsampling experiments revealed that performance degradation at lower sampling rates is minimal: at 62.5 kSPS (16× downsampling), accuracy remained above 97%, with only a one-percentage-point decrease in F1-score. These findings indicate that efficient lightning classification can be achieved with reduced sampling rates and reduced time windows, paving the way for low-cost, real-time lightning detection systems with embedded artificial intelligence capabilities.

A Internet das Coisas (IoT) tem dependido de nós de computação em borda para descentralizar a computação e trazer mais poder de processamento próximo aos dispositivos IoT, como sensores e atuadores. Os nós de computação de borda da IoT têm mais poder de processamento de dados e recursos energéticos do que os dispositivos IoT regulares que visam monitorar e atuar no ambiente. No entanto, em geral, os nós de computação de borda não são projetados para treinamento intensivo de Aprendizado de Máquina (ML) ou para hospedar grandes modelos de ML. Nas arquiteturas de rede IoT atuais, existem múltiplos nós de computação de borda estrategicamente localizados perto de um grande número de dispositivos, onde cada um dos nós de computação de borda tem acesso a parte dos dados produzidos por toda a rede IoT. Neste cenário, cada nó de computação de borda executa modelos de ML leves em seu conjunto de dados local. Neste artigo, propomos uma solução, chamada FEderated Decision Tree (FEDT), Árvore de Decisão Federada, que agrega o aprendizado produzido por múltiplas árvores de decisão de nós de borda cooperativos, seguindo os princípios de aprendizado federado. Apresentamos quatro estratégias de aprendizado federado diferentes e demonstramos que o FEDT pode alcançar cerca de 80% de um modelo de ML centralizado em termos de correlação de Pearson.

Neuromorphic sensors have inherently‐fast speeds and low data rates, which potentially make them ideal for the observation of transient sources, such as lightning and sprites. Particularly, for remote observations. In this article, we report the first observations of sprites from the ground with a neuromorphic sensor. These observations are accompanied by measurements with established instruments such as low‐light level and high‐frame rate cameras. We determine that neuromorphic sensors can capture sprites and determine their duration to an accuracy of roughly 6 ms. Average sprite durations were found to be 55 ms within our data set. We have also ascertained that sprites may be too dim for the neuromorphic sensors to resolve the internal spatiotemporal dynamics, at least without the aid of intensifiers.

Lightning Electromagnetic Pulses, or LEMPs, propagate in the Earth–ionosphere waveguide and can be detected remotely by ground-based lightning electric field sensors. LEMPs produced by different types of lightning processes have different signatures. A single thunderstorm can produce thousands of LEMPs, which makes their classification virtually impossible to carry out manually. The lightning classification is important to distinguish the types of thunderstorms and to know their severity. Lightning type is also related to aerosol concentration and can reveal wildfires. Artificial Intelligence (AI) is a good approach to recognizing patterns and dealing with huge datasets. AI is the general denomination for different Machine Learning Algorithms (MLAs) including deep learning and others. The constant improvements in the AI field show us that most of the Lightning Location Systems (LLS) will soon incorporate those techniques to improve their performance in the lightning-type classification task. In this study, we assess the performance of different MLAs, including a SVM (Support Vector Machine), MLP (Multi-Layer Perceptron), FCN (Fully Convolutional Network), and Residual Neural Network (ResNet) in the task of LEMP classification. We also address different aspects of the dataset that can interfere with the classification problem, including data balance, noise level, and LEMP recorded length.

Modeling for rainfall forecasting is widely used in climate forecasting. A high volume of precipitation or scarcity brings serious problems for society. Thus, a methodology for monthly rainfall forecast is presented using Group Method of Data Handling (GMDH) and sea surface temperature (SST). The intelligent model gets the mean monthly SST, in predefined and temporally lagged areas, after a variable selection step. For model training, precipitation data from the Climate Prediction Center were used. The methodology was applied in a certain area of the municipality of Maraba, located in the southeastern region of the Para state. The results obtained with GMDH overcame those got by using a conventional multilayer Artificial Neural those got by using a conventional multilayer Artificial Neural Network (RNA), reaching values for regression coefficient (R) and root mean square error (RMSE) equal to 0.96363 and 30.8100 in the test stage, while the RNA-based model got 0.81712 and 68.2607, respectively. Results show the GMHDs effectiveness for the monthly rainfall prediction, constituting an important tool for the planning and assistance to decision makers.

The importance of lightning has long been recognized from the point of view of climate‐related phenomena. However, the detailed investigation of lightning on global scales is currently hindered by the incomplete and spatially uneven detection efficiency of ground‐based global lightning detection networks and by the restricted spatio‐temporal coverage of satellite observations. We are developing different methods for investigating global lightning activity based on Schumann resonance (SR) measurements. SRs are global electromagnetic resonances of the Earth‐ionosphere cavity maintained by the vertical component of lightning. Since charge separation in thunderstorms is gravity‐driven, charge is typically separated vertically in thunderclouds, so every lightning flash contributes to the measured SR field. This circumstance makes SR measurements very suitable for climate‐related investigations. In this study, 19 days of global lightning activity in January 2019 are analyzed based on SR intensity records from 18 SR stations and the results are compared with independent lightning observations provided by ground‐based (WWLLN, GLD360, and ENTLN) and satellite‐based (GLM, LIS/OTD) global lightning detection. Daily average SR intensity records from different stations exhibit strong similarity in the investigated time interval. The inferred intensity of global lightning activity varies by a factor of 2–3 on the time scale of 3–5 days which we attribute to continental‐scale temperature changes related to cold air outbreaks from polar regions. While our results demonstrate that the SR phenomenon is a powerful tool to investigate global lightning, it is also clear that currently available technology limits the detailed quantitative evaluation of lightning activity on continental scales.