ML weather or climate models applied to a sector: renewable energy, agriculture, health, insurance, aviation. The ML-for-weather part must be central.
Search and filter this topic Plain list on GitHub
Position paper surveying how machine learning can help climate change mitigation and adaptation across sectors such as energy and disaster management; not focused on weather or climate science.
ETH ZurichGoogle DeepMindGoogle ResearchLawrence Berkeley National Laboratory / NERSCMassachusetts Institute of Technology (MIT)New York University (M2LInES)Stanford UniversityUniversity of California, Berkeley
Systematic review of machine learning, deep learning and IoT methods for weather and climate prediction in Pakistan, motivated by agricultural and climate risks.
Hybrid WRF-plus-neural-network model forecasts 24-hour-ahead wind power at a northwestern Mexico farm, with 6.97% hourly NMAPE, outperforming decision trees and SVR.
Uses a V-Net to predict leaf area index from climate variables for future vegetation productivity, with outputs consistent with Earth observation LAI product standards.
Combining three NWP sources with ML wind-downscaling reduces Nordic wind power forecast errors by 8-30%, with turbine-level aggregation improving accuracy by 10-15%.
Gradient-boosted hybrid models combining macro-meteorological baselines with local observations cut optical turbulence error by 29% with one day of data and 68% with 180 days.
Compares weather-pattern clustering schemes for day-ahead PV power forecasts in Taiwan; Transformer with weather classification gives the best deterministic and probabilistic accuracy.
Interpretable machine learning projects how climate change under emission scenarios will degrade agricultural land suitability across Central Eurasia, highlighting key climate drivers.
Survey of quantum machine learning for climate change, covering energy, climate data forecasting, monitoring and hazard prediction, with open challenges.
A hybrid dynamic-factor, random forest and neural network model predicts greenhouse temperature, humidity, PAR and CO2 at multiple horizons, beating LSTM and BPNN baselines.
Neural networks emulate mean radiant temperature to guide an optimization of urban tree placement that reduces heat stress during heatwaves and over decades.
ML on CMIP5 projections forecasts crop productivity: rice output falls 10% in Vietnam and 14% in Thailand, but rises 11% in the Philippines by 2026.
Adaptive MRMR feature selection with conditional mutual information builds an optimal graph for GCN wind-farm cluster power forecasts from NWP data, cutting cost while improving accuracy.
Pairs LSTM-RNN weather forecasting with a Random Forest for crop selection, reporting 5-8% RMSE on temperature and rainfall and 97% crop-selection accuracy.
A stacking ensemble ML model projects heat and cold wave days in China to 2060, showing wind power output falls during heat waves and rises during cold waves.
MMST-ViT transformer predicts US county crop yields from satellite images and weather, capturing short-term weather and long-term climate change effects with three metrics improved.
Neural-network forecasts of hybrid wind-solar plant output use NWP feature selection to cut forecast error by about 4% and raise market revenue.
Bayesian deep learning forecasts long-term daily solar irradiance from climate simulation outputs for energy applications.
Machine learning models link meteorological variables to rice yield in Korea, with the best algorithm identified for crop yield prediction.
Modular neural network encodes uncertain seasonal climate forecasts for supply-chain demand forecasting, reducing error by 13% to 17% versus existing methods.
Reviews how AI, including ML and NLP, can improve climate information services and drought early warning in Africa, with sub-Saharan case studies.
Review of machine learning methods for wind flow and wind-power forecasting, covering ensembles, SVR and neural networks, with future directions.
LSTM forecasts hourly floating PV output from KMA weather forecasts at two Korean sites, meeting the 5% error threshold.
LightGBM with SHAP explanations predicts tropical cyclone disaster loss grades in China with 0.86 accuracy, with rainfall and wind as the main drivers.
Nanjing University of Information Science & Technology (NUIST)China Meteorological Administration (CMA)
A Fourier neural operator emulates CFD simulations of 3D urban wind fields, predicting in milliseconds and generalizing to unseen wind directions.
Review of deep learning solar forecasting from sky cameras, satellites and weather stations that tracks cloud displacement, outlining gaps relative to NWP.
A deep neural network predicts typhoon-caused building loss ratios in southern Korea, outperforming multiple regression on MAE, RMSE and R2.
Compares four ML models trained on public weather forecasts versus observations for one-day-ahead reference evapotranspiration; forecast-trained models perform better.
Optimal sizing of a renewable hybrid energy system that uses machine learning weather forecasts as an input; abstract unavailable.
Deep learning converts discrete weather data into continuous fields for ship route optimization; abstract unavailable.
Reviews how machine learning weather and climate forecasts can support anticipatory action in disaster response, identifying gaps between forecasting and impact prediction.
On-site adaptive probabilistic wind power forecasting combined with NWP forecasts improves short-term reliability for grid operation.
Review of AI for climate change adaptation covering vulnerability mapping, climate scenario simulation and risk assessment, plus ethical concerns.
A fully convolutional network forecasts spatially distributed solar radiation over time, targeting renewable energy applications.
Proposes a machine-learning approach to estimate solar radiation from meteorological inputs, aimed at renewable energy planning and climate studies.
Machine learning on Berkeley Earth temperatures and World Bank GDP finds a significant relationship between average temperature and long-term GDP growth.
Hybrid model corrects NWP wind speed with multi-scale deep learning for seven-day wind power forecasts, cutting MSE by 65% versus a state-of-the-art baseline.
A deep network combining satellite imagery with station series improves day-ahead solar irradiance forecasts and outputs per-step distributions, with zero-shot tests at unseen stations.
Multiplicative LSTM models project climate-change effects on paddy rice production in a region of interest; a regional application with limited methodological novelty.
Uses farmland sensor data and ML to predict crop yield and generate local weather forecasts for farm management decisions.
Review of statistical machine learning for fishery weather scenario generation, forecasting and extreme-weather warnings.
A random forest driven by weather-station data predicts wet-snow avalanche activity in the Swiss Alps, with F1 near 0.8 in nowcast and NWP-driven 24-hour modes.
Reviews quantum artificial intelligence for climate change, including pattern identification and parameterization, but offers no quantitative results and little demonstrated weather-ML content.
Random-forest blending of sky-imager, satellite and data-driven forecasts improves 90-minute solar irradiance nowcasts by up to 30% for GHI and 40% for DNI.
Dimensional-analysis gradient boosting parameterizes atmospheric surface-layer optical turbulence C_n^2 with R^2 of 0.958 for free-space optical communication.
Stacked RNN combining SCADA turbine data and NWP maps forecasts wind farm power up to 90 h ahead, with a 25% skill gain over XGBoost.
Multimodal sky-image and meteorological end-to-end learning nowcasts renewable power and cuts system cost 30% versus unimodal baselines in a Dutch case study.
An LSTM driven by ERA5-Land inputs forecasts drinking water demand from 7 days to two months; ERA5-Land is usable but needs bias correction.
Surveys machine learning methods for weather forecasting, framed around marine forecasts for fisherman safety, without new models or quantitative results.
Random forest and geographic random forest models link heatwaves, air pollution and socioeconomic status to emergency department attendances in Perth, with R² of 0.975.
Gradient-boosted models forecast short-term local wind for a Moroccan chemical plant, cutting errors 38-52% versus official forecasts and enabling pollution-aware production decisions.
Describes a Microsoft AI system that predicts on-farm micro-climate conditions from farm sensor data to support precision agriculture decisions.
ANN trained on station weather observations matches NWP-trained PV forecasts closely and beats a physical model at three Finnish sites.
Feed-forward neural networks with ensemble learning predict wind turbine leading-edge erosion from mesoscale NWP weather data and blade inspections in Northern Europe.
Sky-camera images feed ANN and SVM models that nowcast photovoltaic output 3 hours ahead, with errors close to the uncertainty of the solar resource itself.
Neural networks map reanalysis weather to PV, wind and load, generating 42-year synthetic ensembles for Hungary; Dunkelflaute events and a variance-correction step are analysed.
Unsupervised clustering of ERA5 wind patterns drives a few WRF runs, reproducing a year of wind farm power and wake predictions at under 2% of the compute.
Debiased machine learning estimates adaptation to damaging heat in US corn and soy, finding short-run heat losses are largely offset in the long run.
Random Forest, GAM and LightGBM map landslide susceptibility under CMIP6 warming, finding high-susceptibility zones grow by up to 4.0% at 4 °C warming.
Nanjing University of Information Science & Technology (NUIST)
NWPP-EDH maps regional evaporation duct height from NWP products, reducing RMSE 61–87% versus classical theoretical duct models.
Random forest and XGBoost ensembles predict site-specific solar radiation from station weather forecast reports, then convert it to PV power output.
A CNN-RNN model combining multi-NWP grid inputs with farm histories improves day-ahead wind power forecasts over LightGBM when trained globally across farms.
A neural ML method predicts pentad water-surface evaporation for rice fields from forecast temperature, wind and precipitation, aimed at medium-term agricultural planning.
Hybrid GWO-WNN, VMD-LSTM and Q-learning model combines NWP forecasts with wind power series to improve short-term wind power prediction accuracy.
Reviews feature selection methods such as LASSO and causal selection for climate-economy studies, and applies them to find an optimal heatwave definition for economic outcomes.
Applies neural networks to social-media data to assess typhoon disaster impacts; the abstract is missing, so details are inferred from the title.
Fine-tuned BERT recognizes seven drought impact types in tweets, reaching 0.89 macro-F1 on the Drought Impact Report test set.
Feed-forward neural networks show heat stress and fertilizer drive maize and millet yields in the Sahel, explaining 59-79% of yield variance.
CatBoost best predicts solar radiation in Korea from temperature, humidity, cloudiness and extraterrestrial radiation, outperforming LSTM and GRU variants.
LSTM weather prediction feeds a clustering and nonlinear sizing model for off-grid hybrid renewable systems, giving a 0.1% loss-of-load gain over seasonal naive.
Gradient boosted trees link California rangeland productivity to climate variables and project regional changes in gross primary production under downscaled future climate.
Statistical ML seasonal forecasts of poor US soybean harvests from Pacific SST and soil moisture precursors reach 67% correct at three months before sowing.
Privacy-preserving bi-party framework learns local and global latent features from multi-site NWP inputs for renewable power forecasts, with at least 3% average gains over baselines.
PSO-tuned LSTM hybrids using AROME NWP inputs give the best day-ahead probabilistic solar irradiance forecasts for a Tahiti microgrid.
AutoML models link climate to California almond yields and project yield changes using 17 downscaled climate models under two climate scenarios.
Hybrid CNN-MLP model predicts solar radiation for solar energy; the abstract is missing, so the contribution is inferred from the title only.
Simplified machine learning and CLDAS-based estimates of reference evapotranspiration across China perform well when station meteorological inputs are missing.
XGBoost with SHAP links SPI drought indices to text-based drought impacts in Texas across agriculture, fire, and public health categories.
Genetic-programming AutoML wind power forecasts combining three NWP sources cut errors by 8–22% at a Norwegian wind farm using direct and indirect turbine-level schemes.
Random forest trained on GFS forecasts and radiometer data predicts zenith atmospheric noise temperature at NASA Deep Space Network sites, enabling up to 2.43 dB downlink gain.
NASACalifornia Institute of Technology (Caltech)University of California, Los Angeles (UCLA)
Applies machine learning and nearest-neighbour with genetic algorithms to downscale SSP projections, estimating flight cancellations and losses rising up to 8.29 times by 2100.
Combines CWB WRF-based NWP wind speeds with PCA/EMD preprocessing and XGBoost for wind power forecasts, reaching 5.43% NRMSE at one hour ahead.
Deep reinforcement learning agents probe a simplified World-Earth climate-economy model to find renewable-energy transition policies within planetary boundaries.
Compares low-resource CNN models nowcasting UK solar PV yield from satellite imagery over 1 to 4 hours, weighing carbon costs against benefits.
Proposes supervised SLMVP dimensionality reduction for NWP-based solar irradiance forecasts, achieving about 3-5% lower MAE than PCA and LPP at Seville and Lisbon.
A hybrid NWP-physical model plus LSTM gives intraday single-site PV forecasts for 130 distribution-grid systems, with NRMSE of 8.3% and 8.7% in 2020 and 2021.
Regression models predict annual myocardial infarction counts in Augsburg from weather, air pollution and demographics, reaching adjusted R² of 0.59 to 0.71.
Geospatial data and machine learning model drought-related yield losses in Morocco's Gharb plain; the missing abstract limits assessment beyond this scope.
A machine learning transfer method estimates hub-height wind speed from 10 m wind for wind-energy resource assessment; the abstract is missing, so scope is inferred from the title.
Machine and deep learning back-forecasting models reconstruct multidecadal US electricity demand records to separate temperature-driven natural variability from climate change effects.
Convection-permitting simulations with ML damage models project 128% higher typhoon losses for Southeast China under RCP8.5 warming, with 50-year losses up 20–170%.
Machine learning pipeline to improve NWP-based wind power forecasts in Portugal, targeting under 10% error; results are modest and the method is not novel.
VMD decomposition with an attention-LSTM encoder-decoder improves wind power forecasts from NWP inputs over several benchmark models.
Deep recurrent networks trained on Croatian wind farm output and NWP forecasts are compared with operational tools, with a focus on configuration sensitivity.
RICE-N integrated assessment model with multi-agent reinforcement learning simulates climate-economy negotiations to test cooperation frameworks for climate mitigation agreements.
Differential polynomial neural network with PDE-modular structure predicts intra-day and day-ahead PV output from weather and plant records, compared against statistical and probabilistic baselines.
LSTM-with-attention sonification of Tokyo air temperature from 1876 to 2021 as an interactive tool for exploring climate change data.
Random forest predicts typhoon economic losses for Korea by region, projecting 6.5–14.8 times higher losses by 2100 under IPCC scenarios.
Hybrid deep learning with metaheuristic tuning for solar irradiance forecasting; no abstract available, so the contribution is uncertain.
Deep feedforward networks with LIME explanations assess building-level hurricane wind and surge risk using updated forecasts for a Louisiana case.
Uses a deep RNN weather forecast with Gaussian process uncertainty to schedule seed planting, cutting required capacity by up to 69%.
News summary of machine learning weather forecasts at 15-minute resolution that improve renewable energy harvesting, with potential savings of millions of dollars yearly.
Random Forest-based solar irradiance forecasting at 288 San Antonio sites from weather data, achieving R2 of 0.85 to 0.91.
Random forest classifiers predict Swiss avalanche danger levels from 20 years of station weather and snow simulations, reaching 72-78% accuracy.
Neural networks trained on SMAP soil moisture project Southeast Asian peat drying under RCP8.5, finding decreases about half as strong as recent El Niño droughts.
Massachusetts Institute of Technology (MIT)NASAStanford University
Review introduces Urban Climate Informatics, combining new sensors, crowdsourced data and machine learning to study urban atmospheric states and climate.
Machine learning combines NWP outputs with local observations to forecast wind speed and power 10 minutes to 4 hours ahead, comparing variable selection and linear versus nonlinear models.
A hybrid CNN-SVR model predicts daily global solar radiation at six Queensland solar farms from GCM and station variables, beating eight alternative DL and ML methods.
Random forest quantile regression and Bayesian LSTM surrogates emulate a malaria transmission model driven by seasonal temperature and rainfall ensembles, cutting computation while capturing forecast uncertainty.
Feed-forward neural network driven by WRF rainfall output predicts rain attenuation fading on a geostationary satellite downlink during a dust-storm event.
Combines GOES-13 satellite image features with station meteorology in neural networks to estimate solar radiation; rRMSE stays above 20%, rated only fair.
PSO-optimized Bi-LSTM with VMD decomposition predicts typhoon wind speed at offshore wind farms; abstract unavailable, so accuracy is unreported.
Hybrid XGBoost models using S2S dynamical forecasts predict northern China winter wheat yield 3–4 months ahead with R² 0.85, beating observed climate inputs.
Institute of Atmospheric Physics, Chinese Academy of SciencesChina Meteorological Administration (CMA)
Knowledge distillation and transfer learning of forecast errors improve turbine-level wind power predictions at an Arctic wind park by 3.3% to 23.9%.
Open dataset of synthetic hourly PV and wind power for 393 German plants over 500 days, built from ICON-EU weather inputs, with baseline forecasts.
Deep learning probabilistic day-ahead wind power forecasts for an Arctic site improve when measured weather and NWP are used as covariates to correct NWP biases.
Compares kriging, RBF, neural network and tree interpolation of GFS weather to high resolution for flight trajectories, with the choice shaping trajectory optimization.
Builds a regression surrogate of outdoor thermal comfort from WRF climate simulations and a portfolio-based allocation method for urban cooling resources in Singapore.
Generative deep learning turns weather forecasts into multi-step solar irradiation predictions for energy use; the abstract is unavailable, so details are limited.
Neural networks classify atmospheric turbulence encountered in aircraft flight test data, with input parameters optimized for a generic model.
Couples ERA5 and ECMWF HRES inputs with a CatBoost model and grid selection to forecast Turkey's day-ahead national wind power, reaching 7.6% NRMSE.
Compares ML and ensemble methods for day-ahead wind farm output using lagged NWP point forecasts, with a new averaging ensemble giving the lowest error at two farms.
A stacked LSTM autoencoder with feature selection gives daily solar radiation prediction intervals at six Queensland farms, outperforming tree ensembles and persistence.
Hybrid machine learning and S2S dynamical forecasts improve northern China winter wheat yield prediction over observation-based inputs.
Institute of Atmospheric Physics, Chinese Academy of SciencesChina Meteorological Administration (CMA)
Title-only record for ML-inspired weather forecasting applied to solar PV potential; no abstract is available to confirm methods or results.
Compares machine-learning models of heat-related ambulance calls in Japan, examining how heatwave feature definitions affect prediction skill. Abstract not available.
VIIRS nightlight data and simulated hurricane weather feed machine learning that predicts Hurricane Maria power outages in Puerto Rico where outage records are missing.
AutoML comparison finds XGBoost best at predicting persimmon farm fruit-drop damage rates from typhoon information, trained on 2016-2018 and tested on 2019 data.
Machine learning links weather to household air conditioning use, projecting 8–13% higher US residential demand under 1.5–2.0°C warming.
Weather-pattern-split CNN and LSTM models with IOWA combination forecast short-term wind power from NWP variables at a Chinese wind farm.
Random forests with monthly soil moisture and extreme-weather features predict German winter wheat yield anomalies, reaching test R-squared of about 0.70 and explaining 2003, 2014 and 2018 extremes.
Position paper arguing AI in environmental sciences risks bias and injustice, with weather and climate examples; calls for responsible AI practice.
Causal-predictive ML links tropical cyclone exposure to mortality and hospitalization in Medicare recipients, finding substantial respiratory hospitalization increases with wind speed as main driver.
Columbia University (LEAP, Lamont-Doherty)Colorado State University (incl. CIRA)
Machine learning uses typhoon meteorological and grid data to quantify fault risk and generate N-m contingency plans for power grid dispatch, aimed at early warning.
Dropout neural networks with neighboring satellite pixel inputs produce ensemble solar irradiance forecasts with post-processing for renewable energy applications.
Machine learning links daily climate and weather extremes to Ukrainian winter wheat yields, explaining 54% of variability and identifying heat, frost and drought during grain filling as key losses.
Hybrid machine learning with rough-set feature reduction and WRF outputs improves solar radiation prediction in Saudi Arabia, with classification rates up to 8.5% higher.
Differential polynomial neural networks forecast photovoltaic output from recent weather and power data, as an alternative to delayed NWP cloud forecasts.
Gradient boosting and analog ensembles extrapolate NWP wind-speed uncertainty from one year of ensemble runs to 20 years for offshore wind siting, with R2 above 0.75.
An Elman neural network tuned by Mayfly Optimization predicts solar irradiance from weather forecast data, with Hadoop preprocessing. Results beat comparison methods.
SVM and regression models forecast one solar plant's output from meteorological measurements; the case study has limited novelty.
Autoencoder correction of regional NWP feeds a hybrid neural network that forecasts regional wind power, needing only about three months of training data.
Ensemble machine learning projects future South Asian drought and its effect on simulated crop yield under climate scenarios.
Deep learning forecasts satellite rain fade from satellite and radar imagery plus link power, outperforming prior methods in near- and long-term predictions.
Random forest models predict children's emergency department attendances in Western Australia from heatwaves, fire burns and air pollutants, with R² up to 0.95.
Wind and solar resource assessment for Colombia's Caribbean islands pairs semi-empirical models with a neural network that predicts irradiance with RMSE 0.87.
Uses machine learning to model the economic impact of incoming tropical cyclones. The abstract is missing, so this classification is low-confidence.
Random forest and neural networks link weather to flight punctuality at Birmingham Airport, reaching R² of 0.80 for punctuality rates.
Machine learning models with satellite drought indices, weather and NAO predict Moroccan cereal yields up to 4 months before harvest, outperforming single-source models.
Compares two NWP systems with different data assimilation as inputs to a thunderstorm power-outage model, showing simulation errors propagate quantifiably into outage predictions.
Graph neural networks forecast multi-site PV power over Switzerland using PV systems as virtual weather stations, beating single-site NWP methods up to 4 hours ahead.
Linear and logistic regression emulate spatial deposition from a multi-physics atmospheric dispersion ensemble, about 10,000 times faster per run but 24 times faster overall.
Machine learning predicts railway rail temperature from weather forecast and solar features, driving a train-speed-limit alarm map for track safety.
Propagates 1–5 day weather forecast uncertainty into a machine-learning storm outage model for the northeastern US, with precipitation and gusts dominating error.
Generic overview of neural networks and big data for climate resilience, covering modeling, disaster response and resource management, with no new technical results.
Compares decision trees, deep learning, GLM, gradient boosting and random forests to forecast Almus Dam hydropower from GCM projections; gradient boosting best with r=0.717.
Deep learning forecasts of solar power generation; the abstract is missing, so the weather-input role and horizon are unverified.
Neural network uses satellite storm observations and ensemble NWP to predict convection 36 h ahead, yielding a probabilistic indicator that beats existing air traffic convection indicators.
Support vector regression using JMA mesoscale ensemble forecasts improves day-ahead Kanto solar irradiance prediction and reduces the maximum error.
SVM forecasts solar PV output from weather forecasts to support grid balancing; the ML target is power generation rather than the atmosphere.
Compares physics-based and data-driven day-ahead PV power forecasts built on NWP inputs at one site; all ML methods perform comparably.
Deep learning with wavelet transforms nowcasts wind power and wind power ramps in Brazil and Uruguay; no abstract was available, so this is judged from the title.
LSTM and other ML wind forecasts drive a scaffold collapse warning system, giving 1–6 hour advance alerts for typhoon winds at Taichung stations.
Machine learning links cacao yields in Sulawesi to ENSO patterns 8–24 months ahead, explaining 75% of yield variation and timing fertilizer use.
Systematic literature review of machine learning for climate prediction and adaptation planning, with a Morocco focus; offers no new model or result.
Applies support vector regression to model precipitation-driven home insurance claims in a Canadian city; low-significance sector application.
Comparison of nine machine learning methods for short-term wind power forecasting at five Arctic farms; NWP wind inputs give an edge over univariate models for 1 to 6 hours.
Random forest and MARS models link TRMM, GPM and GLDAS hydro-climate drivers to Pakistani wheat yield, identifying six phenological phases with strong responses.
Surveys machine learning in geophysics across weather forecasting, wind and wave energy, and oil and gas exploration, with notes on dataflow acceleration.
XGBoost wind power forecaster uses autoencoder-compressed 20-member NWP ensembles from Taiwan CWB, improving short-term deterministic and probabilistic accuracy.
Applies pix2pix GAN and DeepLabv3+ segmentation to live road images to compute a snow-cover hazard index for Japanese highways.
Hybrid NWP and ANN models forecast wind farm power over complex terrain beyond 6 hours, with the direct power model reaching 8.76% normalized MAE.
Combines numerical weather prediction and machine learning to characterise and forecast wind power ramp events; no abstract is available.
Dual-input 1D CNN predicts wind turbine blade icing 24 hours ahead using turbine data and weather forecasts, reaching 97.9% balanced accuracy across three wind farms.
Quantile regression forests give probabilistic icing-related production-loss forecasts for Swedish wind parks, improved by physical icing model inputs.
EALSTM-QR produces interval wind-power forecasts from numerical weather prediction inputs using deep learning; the title suggests an energy-sector application (no abstract).
Conditional inference trees link modelled weather and snowpack variables to avalanche problem types in Glacier National Park, Canada.
Describes an operational ML ensemble that predicts storm-driven customer power outages along the US Eastern Seaboard using weather, land cover and infrastructure proxies.
Uses machine learning to improve winds-aloft forecasts and an uncertainty-aware routing method, aiming to cut flight time and aviation's climate impact.
Neural network maps GDAS numerical weather data to photovoltaic power output at an Italian plant with under 10% error, even without local weather measurements.
An LSTM network maps climate-model meteorology to monthly wind power utilisation hours, giving medium- to long-term generation forecasts for power planning; the case study is limited.
Graph neural network fuses wind farm power history with NWP inputs and multi-task learning to predict ultra-short-term power for all farms in a cluster.
XGBoost combines NWP turbulence indices into a multi-index consensus, raising aviation turbulence forecast skill by 3–17% over Asia-Pacific against pilot reports.
Semi-supervised ConvLSTM encoder-decoder with dual label guessing forecasts aviation air turbulence from weather data, improving accuracy under scarce turbulence labels.
Random forest and ridge regression on GEFS forecast grids predict daily solar energy at 98 stations, with random forest giving the best accuracy.
Compares LSTM, LSTNet and TCN for short-term solar power forecasting; TCN performs best and sunny/cloudy clustering improves accuracy.
Industry day-ahead photovoltaic forecasting system using weather forecast inputs reports over 92% accuracy on sunny days and 83% on cloudy days.
Analysis of machine learning with meta-heuristic optimisation for solar irradiance prediction; no abstract is available to confirm results.
Uses a machine learning approach to generate future weather files under climate change scenarios for building energy simulation; no abstract available.
Gaussian process regression forecasts 48-hour and 4-hour photovoltaic output in the UK, using sky cloud coverage as a predictor.
Deep learning predicts short-term photovoltaic power using weather forecast inputs; abstract not available, so classification is low confidence.
Machine learning models explain up to 65% of historical crop yield anomalies in 17 European countries, linking 2018 extreme temperature and rainfall to losses.
Deep neural networks predict wind power generation from weather forecast data; abstract not available, so classification is low confidence.
Random forests and gradient-boosted regression trees using HRRR forecast features improve upper-level aviation turbulence (EDR) forecasts and reduce false alarms versus regression baselines.
A NARX neural network estimates monthly global solar radiation across Nigeria from meteorological variables, reaching an index of agreement of 0.85 versus 0.77 for linear regression.
Applies AI methods to numerical weather predictions to forecast residential solar power in a smart microgrid; abstract unavailable, so classification is tentative.
LSTM and Markov chain models forecast offshore wind and wave conditions from FINO3 station measurements to support turbine maintenance planning; the study compares the two models.
Uses machine learning to quantify how pre-occurring ENSO signals drive Australian wheat yield variation, linking seasonal climate drivers to agricultural impacts; abstract unavailable.
Neural networks project future hurricane parameters under RCP8.5 for southeastern US coast, showing economic losses rising sharply under climate change.
Compares six ML algorithms for county-level maize yield prediction and finds XGBoost best; the target is crop yield rather than the atmosphere.
Grad-CAM CNN classifies typhoon-damaged areas in Chiba aerial photos, producing explainable damage maps to support faster disaster response.
A deep neural network predicts Chinese rice yields from climate model data, improving correlation about 38.7% over multiple linear regression.
NCAR's updated wind power system combines NWP with ML power conversion, analog ensembles, radar ramp forecasts and fuzzy-logic icing alerts for Xcel Energy.
Machine learning links climate variables from regional and global climate models to European hydropower capacity factors under future scenarios.
Compares ray tracing on numerical weather data with a machine learning model for predicting atmospheric refraction-induced image displacement, consistent with field measurements in overall amplitude and phase.
Differential polynomial neural network predicts intra-day PV power by estimating clear-sky index from 1–9 hour input shifts, applying atmospheric PDE dynamics to solar energy forecasting.
Spatial quantile regression forests map rainfall erosivity in Greece using RCM projections to 2100, with regional climate models validated against gauges.
Gaussian process regression with Matern kernel predicts short-term PV plant output from irradiance, temperature and solar angles, with a 95% error range of about -1.6% to 1.4%.
Extends concentrated solar dispatch planning with probabilistic weather forecasts and an ML calibration step, reporting up to 15% higher revenue than a persistence approach.
Ensemble ML models forecast county-level corn yields in three US states from in-season weather, achieving RRMSE 7.8% with forecasts possible by 1 June.
Bayesian generative models relate US maize yield to monthly temperature and precipitation, projecting a 2°C warming to reduce yield by 8% and triple its variance.
Ranking and combining spatially distributed NWPs with gradient boosting and quantile regression improves day-ahead probabilistic wind farm power forecasts by up to 1.8%.
Kernel-density analogue ensemble forecasts PV output from weather forecasts in Portugal, beating a neural network in most cases and running 32 times faster.
Ultra-short-term wind power forecasting combines day-ahead NWP-based timing of low-accuracy periods with a neural network and persistence, improving accuracy over classical methods.
Clusters lightning-related utility outages and predicts daily outage groups from radar thunderstorm forecasts, handling class imbalance.
Support vector models predict typhoon-driven rice area loss in the Philippines from geographic and typhoon features, with regression MAE of 6.83 percentage points.
A feed-forward backpropagation network predicts wind speed, direction and turbulence from MST radar data to help choose space launch windows.
Evaluates a neural-network hurricane impact-level forecast over 2017 storms, finding best accuracy about 30 h before landfall and wind speed as the key driver.
Reviews deep learning for renewable energy forecasting, with the energy system rather than the atmosphere as main target.
Deep learning forecasts 300 m leaf area index 10 days ahead over Europe using weather inputs, aiming at heat-wave crop warnings; weather prediction itself is not ML.
Reviews wind and solar energy forecasting; ML is not a main contribution and the focus is on energy.
Explores deep learning input and output configurations for short-term solar power forecasts; the target is energy output rather than weather.
GRU combined with weather forecasts predicts 24-hour-ahead solar irradiance, cutting RMSE 11.9% versus RNN and 23.3% versus backpropagation.
Tracks radar-identified convective storm cells and classifies their power-outage damage potential with random forests and deep networks on imbalanced data.
Position paper surveying how machine learning can help climate change mitigation and adaptation across sectors such as energy and disaster management; not focused on weather or climate science.
ETH ZurichGoogle DeepMindGoogle ResearchLawrence Berkeley National Laboratory / NERSCMassachusetts Institute of Technology (MIT)New York University (M2LInES)Stanford UniversityUniversity of California, Berkeley
CycleGAN transforms street-view house images to depict post-extreme-weather outcomes, paired with climate projections, to visualize 50-year climate change impacts for communication.
Compares two ANN day-ahead PV forecasters, one using daily weather forecasts; the hybrid wins on sunny days (NMAE below 1%) but is less stable.
Clustering NWP-similar historical samples before a neural network improves short-term wind power forecasts at a northeast China farm by about 12%.
Review of statistical and machine learning methods for assessing climate-driven non-catastrophic weather risk in agricultural and home insurance.
Bayesian deep learning gives daily solar irradiance forecasts from weather forecasts with confidence estimates; confident outputs show 5% max error versus 22.8% for non-confident ones.
Hybrid day-ahead PV forecasting switches between clear-sky models and an ANN ensemble via a decision tree driven by the weather forecast, validated on a real plant.
Uses machine learning on sub-national yield data to show climate extremes explain 18-43% of yield-anomaly variance, with temperature extremes dominating global crop losses.
ETH ZurichPotsdam Institute for Climate Impact Research (PIK)
Feed-forward deep learning decomposes day-ahead half-hourly PV power forecasting using weather data and forecasts from two years of Australian solar systems, with competitive accuracy and good scaling.
A two-step model links unannounced weather variables to announced forecasts for one-day-ahead PV output; random forest reaches R-squared 0.705, beating the base approach.
PVNet, a convolutional LSTM over NOAA NWP fields, forecasts aggregated German PV output 24 and 48 hours ahead, outperforming persistence.
Decision tree calibration of greenhouse leaf wetness duration models; the neural network model gives the best estimates, with MAE of 1.3 to 1.85 h.
Neural network and linear models predict Chilean agricultural drought from ENSO, PDO and precipitation indices at 1-6 month lead times.
Applies machine learning to satellite data to nowcast photovoltaic energy output, linking cloud and irradiance nowcasting to solar power.
Machine learning classifiers using NWP and site data raise the economic value of wind turbine icing forecasts over logistic regression and empirical rules.
LSTM-RNN forecasts Sotavento wind farm power 1 to 24 hours ahead using NWP inputs chosen by sensitivity analysis.
Survey of machine learning for weather parameter estimation tied to weather derivatives, with a proposed image-processing and neural network method.
Random forests use regional weather forecasts and on-site data to predict icing-related wind power losses up to 42 hours ahead.
Reviews machine learning for crop yield prediction and climate impact on agriculture, with crops rather than the climate system as the main target.
Compares MLP, CNN and LSTM models using sky images to nowcast one-minute photovoltaic output; LSTM gives a 21% RMSE skill gain over persistence.
Linear regression and a feedforward network predict Chilean agricultural productivity drought anomalies 1–6 months ahead using MODIS, rainfall and ENSO indices.
Maps Tibetan Plateau permafrost at 1 km using logistic regression, SVM and random forest, projecting 25.9% to 43.9% permafrost loss by the 2040s and 2090s under RCP4.5.
Deep learning forecasts solar radiation and PV output during dust storms, capturing aerosol-driven diffuse irradiance effects that degrade standard forecasts.
Hybrid PSR, mRMR, K-means and ANFIS-PSO model combines NWP inputs for multi-step wind farm power prediction, with improved accuracy reported on two real wind farms.
Predictive deep belief network feeds wind speed forecasts into an optimised random forest for wind power prediction at 15 min to 24 h horizons.
Neural network predicts probability distributions over discretised power bins for wind farm forecasts, matching deterministic MAE with lower CRPS and avoiding costly NWP ensembles.
Zero-inflated neural networks predict tornado property damage occurrence (AUROC 0.872) and amount (R² 0.432), producing gridded monthly damage probability maps.
Pre-season soybean and maize yield forecasts use satellite precipitation and seasonal climate forecasts instead of NDVI, with errors comparable to in-season systems.
SOM and LDA dimensionality reduction with a DNN classifier claims 7-23% accuracy gains for weather and crop prediction; methodology is weakly described.
Coupled Gaussian process priors improve multi-site solar power and wind speed forecasts, matching or beating benchmarks with better uncertainty quantification.
Thesis applies hybrid neuro-evolutionary methods to predict solar radiation, wind-farm power variations, wave height and airport low-visibility events for renewable energy and facilities.
Projects 2020–2050 emergency medical visits in Tehran under RCP scenarios by linking downscaled GCM climate and air pollution to health via an ANN.
Random forests and neural networks predict grid power outage damage from convective storm cells, using SMOTE to handle rare extreme events.
Self-adaptive differential evolution tuned extreme learning machines predict long-term solar radiation in solar-rich cities using MODIS and reanalysis predictors.
Gradient boosted decision trees predict short-term photovoltaic power from historical weather and weather forecast inputs for grid operation.
Neural network and data-mining framework for wind power nowcasting and forecasting in semi-complex terrain; abstract not available.
Decision trees classify rice and CGPRT cropping patterns from regional rainfall patterns in Indonesia, with J48 reaching 48% accuracy.
Weighted self-organizing map predicts weather and suitable crops for the Mysore region, reporting 0.5–2% accuracy gains over SOM, KNN and ensemble networks.
Machine learning maps PM concentration to aerosol optical depth, and a dust model is combined with weather forecasts to predict ultra-short-term PV output under fog and haze.
Random Forest, ANN and SVM forecast photovoltaic output from forecast weather and predicted solar radiation, with Random Forest performing best.
LSTM produces hourly day-ahead solar irradiance forecasts from weather forecast inputs for solar energy planning; abstract unavailable, so judged from title.
Applies machine learning to NWP cloud forecasts to support optical communication links; abstract unavailable, so scope is inferred from the title.
Neural-network decision support system calibrates weather forecasts for wind farm orography and predicts power output, improving prediction at ten real wind farms.
Reviews machine learning for solar radiation forecasting, a renewable energy application rather than atmospheric science.
Deep Gaussian process with CNN or LSTM features predicts county-level crop yields from remote sensing; the target is agriculture, not weather or climate.
Applies an autoencoder with LSTM to solar power forecasting; the energy target is not an atmospheric target, so it is out of scope.
Aurora is a large Earth-system foundation model trained on over a million hours of data that beats operational systems for air quality, waves, cyclone tracks and high-resolution weather.
Foundation modelsThe Alan Turing InstituteMicrosoft Research (AI for Science)Silurian AIUniversity of Cambridge (Aardvark)
YingLong is a 3 km AI limited-area model that beats WRF-ARW on surface wind speed in two domains while running much faster, using Pangu-Weather boundaries.
Regional & km-scale modelsBaiduChina Meteorological Administration (CMA)
A variational mode decomposition plus LSTM hybrid forecasts next-hour wind speed in Saudi Arabia with 61–74% skill, outperforming other hybrid and deep models.
Evidential deep learning gives one-model aleatoric and epistemic uncertainty for winter precipitation type and surface flux regression, matching ensemble accuracy; releases the MILES-GUESS package.
ML-assisted physical models extract cloud features from sky-imager and irradiance data for intra-hour solar nowcasting, reducing hardware supervision needs for solar energy.
Feature engineering and selection for airport wind nowcasts 2–20 min ahead; wind speed MAPE reaches 0.35–0.63%, but direction remains hard.
U-Net models estimate aviation turbulence intensity at three altitude layers from GOES-16 ABI imagery; U-Net3+ matches pilot reports best.
Observations & remote sensingColorado State University (incl. CIRA)
Evaluates Pangu-Weather against ECMWF and MEPS at 183 Norwegian stations; probabilistic post-processing helps global models most, though MEPS remains best.
Post-processing & calibrationMET Norway (Norwegian Meteorological Institute)
LSTM with a Weibull wind-speed distribution, combined with NWP, forecasts Montreal wind 48 h ahead, cutting RMSE 13–31% versus plain LSTM.
Machine learning corrects WRF hub-height wind forecasts for black-start operations, cutting day-2 wind MSE by over 90% with a multioutput network and nowcasting with LSTM.
Machine learning downscales ERA5 wind speed from about 31 km to 1 km in complex terrain using station data and topographic metrics, aimed at energy applications.
CNN plus Bi-GRU classifies weather conditions from on-farm sky images to support farmers' decisions; no quantitative results given.
CNN-LSTM hybrid forecasts aviation visibility, reducing RMSE to 2.68 and MAPE to 13.41% from LSTM alone.
GUMP combines ML sensor predictions with CFD flow fields to give hyper-local urban wind forecasts and a risk map for drone operations.
IoT and fuzzy-logic system predicts 24-hour farm rainfall from sensors and cut water use 55% in a two-hectare maize trial; ML role is modest.
Kazakhstan avalanche bulletin uses a trained neural network on mountain weather station data, reaching 84-90% danger-level classification accuracy.
Wildfire Assessment Model uses a self-supervised CNN on atmospheric and greenness variables to estimate resources and burned area in Spain, improving baselines by up to 21%.
An anticipative vision transformer forecasts solar irradiance 15 minutes ahead from sky image sequences, reaching 21.45% forecast skill over smart persistence.
Uses atmospheric circulation patterns as inputs to ANN models (NARX, PLANN) for monthly to 9-month sea level anomaly forecasts, outperforming post-processed NWP and climatology at US sites.
Sub-seasonal to seasonalNOAA (incl. GFDL, PSL, GSL, NCEP/EMC)
An SVR and MLP ensemble calibrates GFS hourly solar irradiation forecasts in Romania, cutting normalized RMSE by about 16% using sky imagery.
A probabilistic CNN-BiLSTM with multi-head attention forecasts day-ahead wind speed for energy use; the abstract was unavailable, so the claim is limited.
Local & station forecastingUniversity of Cambridge (Aardvark)
A GAN converts parametric typhoon wind and pressure fields into NWP-like forcing, giving storm surge accuracy comparable to NWP and better than parametric models.
LSTM forecasts wintertime air stagnation index in southern China from autumn Niño indices and past ASI, reaching correlation 0.778 at long lead.
Machine-learning post-processing of hourly HARMONIE and ERA5 forecasts predicts 1-minute irradiance variability for Dutch stations, with NWP radiation errors causing about half the inaccuracy.
Post-processing & calibrationKNMI (Royal Netherlands Meteorological Institute)
An explainable ML wildfire model trained on 7 million satellite observations projects a 55% rise in fire exposure for forest carbon projects by 2080 under SSP2-4.5.
Generative ML downscales daily CMIP GCM output to hourly 4 km weather for wind, solar, temperature and humidity, enabling grid studies under climate change.
Reviews physics-informed and data-driven visual anemometry, inferring local wind from the motion of trees and flags for renewable energy and urban applications.
Observations & remote sensingMassachusetts Institute of Technology (MIT)California Institute of Technology (Caltech)
Ensemble ML models forecast fog and dense fog at Patna airport from 6 h to 1 h ahead using historical synoptic station data for aviation early warning.
ML downscaling of gridded precipitation, temperature and soil moisture improves weather index insurance hedging in 70% of Kazakh and Mongolian county cases.
Bayesian-GRU encoder-decoder with multi-head attention forecasts station temperature and humidity for agriculture, reporting R of 0.80–0.90 at three sites.
Combines sky images and satellite observations in one ML framework for deterministic and probabilistic intra-hour solar irradiance nowcasts up to 60 minutes ahead.
Multi-model machine learning estimates that fully developed Fennoscandian forests warm annual mean land surface temperature by 0.26 °C relative to present forest states.
Neural networks forecast Nigerian weather parameters from 19 years of NASA data to assess solar PV siting, reaching 96% training correlation.
Local & station forecastingMassachusetts Institute of Technology (MIT)University of Oklahoma (AI2ES)
Score-based diffusion models super-resolve coarse NWP to satellite-resolution day-ahead solar irradiance, giving probabilistic forecasts with many samples.
Builds a recurrent neural network fuel moisture model for WRF-SFIRE, initialized as a numerical ODE solver and trained on RAWS sensor and RTMA weather data to improve fire-weather inputs.
A CNN classifies six cloud types in ground-based sky images from the SWIMCAT-EXT dataset for use in solar radiation forecasting.
U-Net with ResNet18 segments clouds in horizon-camera images for solar irradiance nowcasting, reaching mean IoU 0.6 with class imbalance limiting performance.
Releases 4 km hourly wind, solar and meteorological fields for the contiguous US under climate scenarios, generatively downscaled from GCMs for energy studies.
Decision-tree and ML systems are proposed for smart-city weather modification forecasting, claiming 35% accuracy gains over traditional models on Shaanxi data with limited methodological detail.
Hybrid CNN-CatBoost predicts solar radiation at 18 Korean locations from weather variables, and adding wind and precipitation improves results at 15 of them.
An LSTM with regression-based feature selection forecasts daily temperature from station variables for aviation safety, cutting RMSE from 4.03 to 2.22.
A 2D/3D CNN with location-aware adaptive normalization improves wildfire danger forecasting on the FireCube dataset over baselines, with the performance gain reported but not quantified in the abstract.
Neural network and random forest classify sky-camera images to nowcast clear-sky index drops 1–5 minutes ahead for PV cloud-induced variations.
CatBoost bias-corrects WRF wind-speed forecasts at a Gobi wind farm, reducing a four-day mean deviation from 3.21 to about 2.50 m/s.
Post-processing & calibrationInstitute of Atmospheric Physics, Chinese Academy of SciencesChina Meteorological Administration (CMA)
Ensemble GRU-LSTM model predicts road surface temperature 1 to 6 hours ahead from station data, lowering MAE versus CNN-LSTM and SVR baselines.
RLMD-BiLSTM hybrid forecasts solar irradiance at two Indian sites, cutting RMSE by 59–89% against benchmark models; the study is station-based and regional.
Compares local, global and transfer-learning strategies for deep learning solar forecasting from sky images across three heterogeneous datasets; transfer needs 80% less data.
Software accompanying a study that uses neural networks to predict hurricane storm surge and assess its sensitivity to storm characteristics.
Compares LSTM, BiLSTM and GRU for long-term offshore wind speed and wave height forecasts from FINO3 station data, reporting better seasonal response and reliability.
Neural network and GLM correct COSMO-REA6 reanalysis wind speeds at 100 m after diagnostic wind-model downscaling, reducing RMSE by almost 30% in Central Europe.
Ensemble of ten regressors forecasts daily rainfall, humidity and temperatures in Senegal, with R² of 0.82–0.93 for the ensemble model.
CNNs on GEFS precipitation fields for 1–14 day Sacramento precipitation occurrence; skill matches bias-corrected ensemble mean but cost-tunable thresholds help water management.
Spatio-temporal neural network predicts Palmer Drought Severity Index across regions, reaching R² 0.90 versus 0.85 for gradient boosting, and projects drought under climate scenarios.
WISRnet combines a CNN autoencoder with k-means to identify Dunkelflaute periods from reanalysis-type wind and solar patterns, finding most events fall in 5 of 25 clusters.
Compares threshold, neural network and belief-function methods for 0–24 h lightning strike risk over the sea for aviation, with WWLLN ground truth.
Machine learning forecasts residuals of NWP wind speed and direction at an Arctic wind farm, enabling bias correction of NWP inputs for wind power forecasting.
U-ConvLSTM wildfire prediction using environmental and meteorological data reaches over 97% accuracy and predicts 2018 California fires 5-14 days ahead, versus about 76% for CNNs.
Machine learning trajectory and intensity models trained on HURDAT2 generate synthetic Atlantic hurricanes whose landfall locations and wind speeds match historical records.
Compares nine machine-learning site-adaptation methods to calibrate the STRÅNG gridded solar irradiance product in Sweden using three stations; no single method wins everywhere.
XGBoost with SHAP predicts forest-interior temperatures from open-field weather data for Brandenburg sites, supporting climate-change ecosystem assessment.
Combines Twitter sentiment with Global Fire Atlas satellite data to predict wildfire attributes, treating social media as a human-sensor for wildfire nowcasting.
Two-step neural post-processing of ensemble forecasts for 100m wind and irradiance, improving on raw ensembles up to 48 h at the Hungarian Met Service.
Bayesian neural networks forecast offshore wind speed and direction at FINO1 from ocean variables, giving calibrated uncertainty for North Sea wind-resource prediction.
Combines all-sky camera images with satellite observations in one ML framework, improving intra-hour solar irradiance forecasts up to 60 minutes ahead.
Hybrid BA-KNEA post-processing of GEFSv12 forecasts improves 1–3 day solar radiation forecasts in Xinjiang, with RMSE 3.2–3.7 MJ m-2 d-1.
Random forest post-processing of CFSv2 precipitation forecasts over Iran (1982-2017) reaches correlation above 0.87, packaged as a GUI decision support tool.
UT-GLOBUS uses machine learning on spaceborne altimetry to map building heights for 1200+ cities, improving WRF-Urban intra-urban air temperature RMSE by 55% over a table-based approach.
Hybrid CNN-REGST model predicts daily solar radiation at six Queensland farms from GCM and observation features, reported as more accurate than DL and ML baselines.
Study of exogenous atmospheric variables as machine learning predictors of wind speed for energy applications; abstract unavailable, so scope is inferred from the title.
Neural network retrieves ocean surface wind, SST and near-surface air temperature and humidity from AMSR2, with heat flux RMSE of 18-40 W/m2 against buoys.
Observations & remote sensingNanjing University of Information Science & Technology (NUIST)
Neural network post-processing of ECMWF ensemble mean corrects Brazilian wind-speed bias to within ±0.1 m/s and cuts RMSE by 10–30% for 1–46 day lead times.
Post-processing & calibrationNOAA (incl. GFDL, PSL, GSL, NCEP/EMC)
Random forests best downscale NWP wind speed to 100 m over South Korea for agriculture, with 48-hour RMSE near 0.8 m/s.
Downscaling & super-resolutionKorea Meteorological Administration (KMA)
MARS regression fusing ERA5 and Meteosat cloud data estimates surface downward longwave radiation with 18.8 W/m² RMSE, beating the LSA-SAF operational formula.
CNN-based statistical downscaling of NWP fields to turbine hub height matches physical-model accuracy for wind power forecasting at one farm.
Conditional GAN turns webcam images into photorealistic future weather views conditioned on COSMO-1 forecasts, matching forecast conditions in at least 89% of cases.
Uses a machine learning framework to observationally constrain 13 CMIP6 Earth system models, projecting higher socioeconomic wildfire exposure in Africa despite lower fire carbon emissions.
Extremes & hazardsOak Ridge National LaboratoryColorado State University (incl. CIRA)
SSA-tuned least-squares SVM forecasts Amman temperature, pressure and soil humidity 24 to 72 h ahead for smart irrigation; a single-city study.
CNN on infrared all-sky images forecasts clear-sky index ranges 5 to 15 min ahead, reaching about 65% accuracy at 5 min for solar irradiance.
Multi-site collaborative deep learning for wind speed forecasting in valley complex terrain; abstract unavailable, judged from title only.
Local & station forecastingChina Meteorological Administration (CMA)
Compares analog ensemble and CNN empirical-statistical downscaling for 10 m winds over complex terrain, finding the analog ensemble has smaller errors than the CNN.
Hybrid deep learning corrects numerical weather forecasts of short-term wind speed; the abstract is unavailable, so the gain is unquantified.
Machine learning with new topography-derived variables downscales Weibull wind resource maps over South Korea, cutting RMSE by about 15% versus baseline inputs.
Vision Transformers with spatial self-attention detect rain and road surface wetness from freeway CCTV cameras, reaching 96.7% and 98.1% F1.
Support vector quantile regression post-processes mesoscale ensemble solar irradiance forecasts, reducing serious overestimation at five Kanto sites.
Post-processing & calibrationJapan Meteorological Agency (JMA)
A deep belief network driven by relative humidity, wind speed and drought index estimates monthly forest fire occurrence in South Korea with R² of 0.84.
A ConvLSTM treats GOES satellite-based solar irradiance nowcasting as next-frame prediction, reaching 120 W/m2 RMSE for North America up to 3 hours ahead without a GPU.
A deep learning model trained on image-like wind fields around PIREP turbulence reports predicts Japanese airspace turbulence with accuracy comparable to existing grid-point indices.
Extremes & hazardsNanjing University of Information Science & Technology (NUIST)
Uses artificial neural networks on INMET station climate records to forecast temperature and precipitation, linking forecasts to crop planting schedules.
Systematic review of 65 articles (2011-2020) on sky-camera cloud nowcasting for solar power, finding a shift from classical statistics toward neural networks and CNNs.
Self-supervised CNN-LSTM models forecast GOES-R satellite irradiance to nowcast solar PV output, reaching errors close to ground-truth-based models at 25 sites.
Uses an LSTM to predict temperature, humidity, sunshine and rainfall for Köppen-class agricultural yield estimation, reporting 3.29% MAPE.
Compares regression with NARX-based neural models to link climate variables to tree-ring growth in NE Korea and reconstruct the growth index for 1999–2019.
Deep MLP corrects RACMO2 Antarctic surface melt using automatic weather station and MODIS albedo data, outperforming random forest and XGBoost on cross-validation.
ML error correction of WRF solar irradiance forecasts in the tropics cuts NRMSE by 40-50% and makes results across four radiation schemes comparable.
Random forest on dynamically downscaled weather produces a daily gridded chamise fuel moisture product for California, reaching R2 0.79 for new growth, to support wildfire danger.
Ensemble neural networks trained on perturbed observations nowcast wind speed and direction for wind farm output, with better accuracy and runtime than alternatives.
Fitness-tuned adaptive neural network forecasts Karnataka rainfall and temperature for 2021-2030 and maps effects on five crops; validation is limited.
Online sequential-learning combinations of NWP and statistical extended-range forecasts improve probabilistic skill by 5-15% over standard multimodel methods up to lead week 5.
Post-processing & calibrationColumbia University (LEAP, Lamont-Doherty)University of Reading
Spatio-temporal deep learning corrects day-ahead wind speed forecasts built on WRF predictions for energy use; abstract missing, judged from title.
Urban energy balance-guided machine learning predicts nocturnal surface urban heat island during a Naples heatwave; abstract missing, judged from title.
UKF-aided LSTM improves short-term aviation wind nowcasts over plain LSTM and quantifies uncertainty with Monte Carlo simulation.
Surrogate model for time-dependent metocean conditions during hurricanes. No abstract is available, so this classification is low-confidence.
Compares linear regression, Gaussian process, SVM and deep learning for statistical downscaling of Thai temperature and precipitation, with deep learning performing best.
Hybrid LSTM and genetic programming model forecasts short-term global solar radiation from weather station data, with improved accuracy over stacking baselines.
Cloud-deployed machine learning compares classifiers for predicting hurricane and cyclone occurrence over Pacific and Atlantic coasts, with map visualization; the methodological novelty is limited.
XGBoost downscales NWP outputs to medium-range daily reference evapotranspiration across China, a weather-derived quantity for water and agriculture planning.
Twelve AI models predict wind speed from meteorological data, with LSTM best at 97.8% accuracy, and copulas model joint wind speed dependence across sites.
XGBoost model estimates hourly 1-km air temperature across 13 northeastern US states, with RMSE about 1.6 K versus 2.5 K for NLDAS-2.
Observations & remote sensingColumbia University (LEAP, Lamont-Doherty)
A multivariable denoising-autoencoder network corrects hub-height wind speed NWP forecasts at 15 Japanese wind farm sites, cutting RMSE using extra atmospheric variables.
Improved CNN estimates hurricane intensity from infrared satellite imagery with 7.6-knot RMSE and adds VGG19-based post-disaster damage assessment.
Deep neural networks on UAS orthomosaics detect and classify tornado damage, producing georeferenced damage heatmaps for post-event assessment after the 2018 Eureka tornado.
EarthNet2021 dataset and challenge forecasts Sentinel-2 surface imagery conditioned on mesoscale future weather, enabling model intercomparison at 20 m resolution.
A neural-network similarity metric replaces hand-tuned distances in analog ensembles, improving wind speed and solar irradiance forecasts over Euclidean-distance analogs.
6-km regional simulations show local SST warming shifts US West Coast snowpack, with neural networks attributing the response to precipitation amount and temperature.
Supervised machine learning applied to wind speed forecasting at one site in Uruguay to support grid energy management; the abstract gives no quantitative results.
Recurrent neural networks generate localized synthetic weather data from airport and on-site measurements, improving test accuracy over standard weather files by over 100%.
Local & station forecastingMassachusetts Institute of Technology (MIT)
LM-BP neural network estimates clear-sky surface shortwave radiation from ASTER imagery with RMSE 27.34 W/m2 against site measurements.
Introduces EarthNet2021, a 32,000-sample dataset and challenge for forecasting 20 m Sentinel-2 imagery conditioned on 1.28 km weather forecasts, with a new EarthNetScore ranking.
U-Net variants with residual, parallel and asymmetric convolutions forecast multi-step coastal Netherlands sea elements from hourly Copernicus data, with the best model winning on frame prediction.
Estimates short-term solar irradiance from sky images with low computational cost, forecasting up to 4 h ahead on 16 years of data.
Compares four deep learning models on nearly a decade of remote sensing, weather and fire records, reaching 83% AUC for wildfire likelihood.
Adversarial deep learning super-resolves wind and solar outputs from global climate models by up to 50x, producing physically consistent fields for renewable resource assessment.
Validates GFS day-ahead solar irradiance at 18 China stations; a gradient-boosting correction of GFS gives the best accuracy, beating persistence.
Deep CNN real-time system for evaluating extreme rainfall to support construction-industry decisions; the abstract is missing, so details are limited.
Compares twelve AI algorithms for wind speed prediction from meteorological station data, with LSTM reaching 97.8% accuracy for wind farm planning.
Deep learning first classifies marine atmospheric duct type from sparse propagation factor measurements, then estimates refractivity parameters for real-time radar-performance applications.
A k-nearest-neighbour model conditioned on reanalysis environment generates stochastic tropical cyclone event sets for the South West Pacific for cyclone risk assessment.
Road weather station network with ML that flags faulty sensor data and corrects physical road-ice forecasts, aiming to improve winter road safety.
DEL-Jet ensemble of CNNs, autoencoder and nonlinear PCA forecasts wind speed for renewable energy, offering a more robust ensemble than single models.
Fuzzy-logic ALPHA algorithm fuses satellite, NWP and radar data to detect high ice water content for aviation, with AUC 0.85 against in situ measurements.
Observations & remote sensingNASANSF NCAR / UCARBureau of Meteorology (Australia)
Multistage ensemble of linear regression, ANN and SVM refines NWP wind speed for wind-farm forecasts, reaching 0.13% MAPE at 72 hours.
Machine learning corrects a coupled climate model's seasonal forecasts of daily air temperature for field-scale agricultural management.
Sub-seasonal to seasonalKorea Meteorological Administration (KMA)
Lightweight perturbed-observation ensemble neural network nowcasts wind speed at masts, cutting NRMSE by 20.9% and running an order of magnitude faster than rivals.
A location-generalizing deep network forecasts short-term solar irradiance from satellite and NWP inputs without local ground data, matching local models (31.31% vs 32.01% rRMSE) across 25 Dutch sites.
Thesis-style work applying machine learning to tropical cyclone impact assessment; the abstract is unavailable, so scope and results are unclear.
Pattern-recognition algorithm identifies mixed wind hazard climate types from station records, then recalculates extreme wind speeds and design return periods for southeast China.
GRU-based correction of wind speed forecast errors to improve short-term wind power forecasting.
Stacked denoising autoencoder features with LSTM improve South African wind speed forecasts by 17% over benchmarks, with NWP data cleaned by clustering.
Frames stratospheric aerosol injection as deep reinforcement learning, using a deep-learning emulator of HadCM3 as the training environment; presented as a first DRL application in climate science.
LSTM sequence-to-sequence models forecast solar irradiance 1 to 24 hours ahead, outperforming GBRT and feed-forward baselines using neighbouring-site wind and irradiance features.
Deep networks on sky-camera videos estimate and forecast solar irradiance up to 4 hours ahead, reducing error relative to satellite-based approaches.
Neural network classifies typhoon low-altitude turbulence from Hong Kong terminal Doppler radar, validated against commercial aircraft turbulence measurements.
Observations & remote sensingNOAA (incl. GFDL, PSL, GSL, NCEP/EMC)University of Oklahoma (AI2ES)
An LSTM forecasts 5-minute solar irradiance minutes ahead from one year of plant radiometer data, beating standard time-series baselines.
Bayesian network learned from four years of station data forecasts wind speed and solar irradiance classes for renewable energy resource evaluation.
CNN trained on past wind, humidity, temperature and visibility predicts runway visual range at a local airport from station time series.
Ensemble of ANN, SVR and Gaussian process models blends NWP wind forecasts with weather-station data via Bayesian model averaging for wind power forecasting at ERCOT.
MARS regression on Korean emergency visits and meteorological data yields heatwave thresholds, such as 32.58°C maximum temperature for two consecutive days.
A neural network tuned by a meta-heuristic predicts monthly and seasonal precipitation in southern Brazil; the abstract does not match the title.
Backpropagation neural network forecasts 12 and 24 h wind speed at La Serena, Chile, with RMSE about 0.7 m/s, and links it to wind turbine power output.
Artificial neural networks replace expert steps in a seasonal rainfall forecasting workflow for Zambia, with results reported only qualitatively.
Hybrid EEMD and wavelet neural network for wind speed prediction; abstract not available, classified from title.
Clustering-based multi-model machine learning forecasts one-hour-ahead solar irradiance, outperforming single-model baselines by about 20%.
Shows AI methods, combined with physical understanding, improve forecasts of storm duration, severe wind and hail, precipitation type, renewable energy and aviation turbulence.
Extremes & hazardsNSF NCAR / UCARNOAA (incl. GFDL, PSL, GSL, NCEP/EMC)University of Oklahoma (AI2ES)
