Real-world Meteorology Research Methods

The Nature Communications article "How to stop being surprised by unprecedented weather" outlines a comprehensive framework to anticipate and manage the risks of extreme, previously unobserved weather events. The article’s central thesis is that surprise should not be the default response to such events—and that science, policy, and disaster planning can work in concert to build resilience. These methods help anticipate extreme weather events beyond what has occurred in the observational record: a. Conventional Statistical Methods - Use historical weather data and extreme value theory to estimate probabilities of rare events. Limitations: Short observational records, underestimation of extremes, and inability to simulate events beyond past climate conditions. b. Past Events and Proxy Data - Extend the view of climate risk through historical documentation, oral history, and paleoclimate proxies (tree rings, sediments, etc.). Benefits: Reveal long-term variability and past extremes that modern records miss. Limitations: Coarse resolution, dating uncertainty, and difficulty aligning with present-day conditions. c. Event-Based Storylines - Construct physically plausible scenarios of specific high-impact events using counterfactuals and modeling. Useful for local decision-making and public engagement. Limitations: Focused on specific events, often non-probabilistic, and dependent on expert input. d. Weather and Climate Model Data Exploration Mine large ensembles of model outputs (e.g., UNSEEN, SMILEs, CORDEX) for unobserved but plausible extremes. Enables exploration of events outside the observational record using physical consistency. Limitations: Computationally intensive, resolution trade-offs, and model biases.

From Global Forecasts to Street-Level Reality We transform many broad, worldwide weather forecasts into precise, street-level predictions. This is a four-step process 1.    Predicting the Weather (The Foundation) The Concept: Rather than just studying past weather patterns, we focus on the future. We use advanced, "stochastic" climate models that generate thousands of scientifically vetted scenarios for rainfall and temperature. The Science: We analyze data from thousands of forward-looking, scientifically verified pathways to create a view of potential future weather. 2.    Calculating the Runoff (The "Water Budget") The Concept: We divide the earth into large grids that are 10 kilometres wide. Each day, we calculate a "water budget" for every grid square. The Science: Using basic physics with a system called PCR-GLOBWB, we track where every drop of water goes. We estimate how much rain falls, how much is absorbed by the soil, how much evaporates, and exactly how much excess water remains to flow into local streams and rivers. 3.    Building the Flood Wave (The Climate Signal) The Concept: Using our daily river flow data, we identify the extreme worst-case moments—such as a rare 1-in-100-year storm. The Science: We determine a specific timeline (called a hydrograph) that shows exactly how quickly the floodwaters will rise to their peak and how quickly they will decline. 4. Mapping the Spread (Hydrodynamic Routing) ●     The Concept: Knowing how much water is in a river isn’t enough; we need to understand what happens when it escapes the riverbanks. ●     The Science: We take that surging wave of water and run it through a highly detailed 3D digital map of the Earth's surface. We simulate the actual physics of water moving horizontally spilling over banks, filling floodplains, and backing up behind hills (using a system called LISFLOOD-FP). The Result: Pinpoint Accuracy By the end of this pipeline, we have refined a rough, zoomed-out 10-kilometre climate estimate into a detailed local map. For example, by examining our simulations for the St. Lawrence Basin and Montreal, we can use 2025 climate data to accurately depict what a severe 1-in-100-year flood would look like. We focus on a detailed resolution, down to 90 meters or even 10 meters at street level, to show the precise depth of the floodwaters. Naturally, this downscaling is only as valid as the Digital Elevation Model available in the location. For example, in Montreal, it is available at 3-meter resolution, which is the highest resolution justified.

Deveshwar Singh has successfully defended his Ph.D. and is my 21st doctoral student to earn the Ph.D. degree under my supervision at the University of Houston. His dissertation, titled “Utilization of Deep-Learning Algorithms for Bias Correction and Forecasting of Weather, Air Quality, and Climate Parameters over Regional Scales,” develops and applies deep learning frameworks to enhance prediction and bias correction of key environmental variables across regional scales. In the first study, he addresses systematic biases in Indian Summer Monsoon Rainfall projections from CORDEX-SA regional climate models using two super-resolution methods—Autoencoder-Decoder and Residual Neural Network (ResNet)—which ingest eight meteorological variables plus elevation at 0.25° resolution and generate bias-corrected precipitation at 0.05° resolution; the ResNet model achieves a five-fold increase in spatial resolution and reduces extreme rainfall bias from 21.18 mm to -7.86 mm. The second study proposes the Deep-BCSI framework, which combines CNN-based bias correction with Partial CNN-based spatial imputation to improve 72-hour PM₂.₅ forecasts over South Korea, increasing the Index of Agreement from 0.65–0.68 (CMAQ) to 0.71–0.80 and lowering RMSE by 25%–41% in metropolitan regions, with SHAP analysis confirming behavior consistent with atmospheric chemistry and meteorological processes. The third study introduces the Spatial-Temporal Attention ResNet Transformer (START), which integrates 17 meteorological variables with spatial and temporal encodings to outperform the NASA GEOS baseline over the contiguous United States—reducing RMSE for temperature by 30.3% and MAE for relative humidity by 32.2%—while SHAP-based interpretation and Monte Carlo Dropout with calibrated uncertainty provide reliable, well-explained predictions that support environmental management and policy decisions.

Weather radar has been a cornerstone of operational meteorology for decades, mainly for detecting precipitation and issuing warnings for severe weather such as hail, floods, and tornadoes. But radar data are increasingly becoming valuable not only for real-time monitoring and nowcasting, but also for long-term climate research. A review article in the Bulletin of the American Meteorological Society (2019) explores how weather radar archives can be used for climatological studies of precipitation and convective storms. Radar networks now cover most densely populated regions of the world and generate enormous high-resolution datasets, often reaching terabytes of data per day. This spatial and temporal detail allows scientists to analyze the structure, intensity, and frequency of precipitation in ways that traditional rain-gauge networks cannot capture. Radar data make it possible to study phenomena such as convective storm modes, hail occurrence, and the spatial organization of heavy rainfall with kilometer-scale resolution. At the same time, the paper highlights important challenges. Radar instruments are often replaced or upgraded every 10–20 years, which can introduce inconsistencies in long time series. Without proper metadata, calibration, and quality control, these changes can lead to misleading conclusions when studying climate trends. The key message is clear: the radar observations we archive today will become an essential climate dataset for future generations. Preserving high-quality radar data – together with detailed metadata – is critical for understanding long-term changes in precipitation extremes and convective storms in a warming climate. Figure description: Global map of weather radar coverage shown in Robinson projection. The areas “illuminated” by individual radars were calculated using the open-source Python library wradlib, assuming a maximum radar range of 200 km regardless of bandwidth, polarization, or terrain effects. Most radar locations were retrieved from the WMO radar database, although the database does not include all operational radars. Additional locations were manually added for China, the Philippines, Vietnam, and Myanmar based on publicly available sources and national meteorological service information. Paper: https://lnkd.in/djxw6S-9 #Meteorology #WeatherRadar #ClimateScience #Precipitation #AtmosphericScience

In 2022, I was part of a research team that took on a critical challenge: 𝐡𝐨𝐰 𝐝𝐨 𝐲𝐨𝐮 𝐚𝐬𝐬𝐞𝐬𝐬 𝐫𝐚𝐢𝐧𝐟𝐚𝐥𝐥 𝐢𝐧 𝐫𝐞𝐠𝐢𝐨𝐧𝐬 𝐰𝐡𝐞𝐫𝐞 𝐫𝐚𝐢𝐧 𝐠𝐚𝐮𝐠𝐞𝐬 𝐚𝐫𝐞 𝐬𝐜𝐚𝐫𝐜𝐞? Our focus was Sudan, a country where climate-driven decisions are crucial, but data gaps make it difficult to plan and prepare. We turned to satellite and gridded rainfall datasets, tools that have become essential in modern hydroclimatic research. What we found was both promising and eye-opening. 📍 While most rainfall products showed a tendency to underestimate rainfall, especially on annual and monthly scales, two stood out: 𝐂𝐇𝐈𝐑𝐏𝐒 𝐚𝐧𝐝 𝐂𝐑𝐔 𝐝𝐚𝐭𝐚𝐬𝐞𝐭𝐬 consistently performed best, especially in the western and southern regions of Sudan. 🌧️ We discovered that summer rainfall (the main rainy season) is captured more accurately than annual totals, especially in mountainous areas. And when we explored deeper, we noticed a significant link between rainfall trends and the Atlantic Multidecadal Oscillation (AMO), with some regions showing correlations as high as 90%. This was more than just data and numbers; it was a reflection of how remote sensing, when used wisely, can support 𝐜𝐥𝐢𝐦𝐚𝐭𝐞 𝐫𝐞𝐬𝐢𝐥𝐢𝐞𝐧𝐜𝐞, 𝐟𝐨𝐨𝐝 𝐬𝐞𝐜𝐮𝐫𝐢𝐭𝐲, 𝐚𝐧𝐝 𝐫𝐢𝐬𝐤 𝐩𝐫𝐞𝐩𝐚𝐫𝐞𝐝𝐧𝐞𝐬𝐬 in the very places that need it most. 🔍 The study underscored the value of CHIRPS data for monitoring rainfall variability and extreme events, and why it should be a go-to resource for decision-makers in data-scarce environments. Being part of this work reminded me of the power of science and collaboration in driving evidence-based action. It’s research like this that fuels real-world solutions, and I’m proud to have contributed to it.  Richard Anyah Mohamed Abdallah Ahmed Alriah, PhD Aslak Grinsted Hans Hersbach Lorenz Ewers Victor Ongoma Nixon Mutai #ClimateResearch #RainfallData #RemoteSensing #Sudan #ClimateResilience #CHIRPS #Hydroclimatology #SatelliteData #ClimateScience #DataForDevelopment

Good afternoon, meteorologists and atmospheric scientists around the world, As some of you may know, NSF NCAR - The National Center for Atmospheric Research #NCAR’s Weather Research and Forecasting model, #WRF, can be installed using the WRF-MOSIT toolkit that my co-authors and I developed. While WRF-MOSIT is not perfect, it does make it easier to install and run the widely used em_real configuration on many platforms with basic nesting support, which remains one of the most common WRF workflows worldwide. What many users may not realize is that WRF supports several compile-time case options beyond em_real, including a range of idealized 1D, 2D, and 3D cases. WRF-MOSIT can be adapted for these, but doing so currently requires modifying the scripts, since em_real is the default target. One simple method is a global sed replacement, though changing nesting behavior is more involved, and I am still working on a cleaner solution. One of the most useful capabilities in the real-data WRF framework is the moving nest, including the vortex-following nest option for tropical cyclones. This allows the highest-resolution inner domain to follow the storm as it moves through the larger parent domain, keeping computational focus on the cyclone without needing to run multiple separate cases. That makes vortex-following nests especially valuable for analyzing tropical cyclone structure, intensity changes, eyewall evolution, rainbands, wind fields, and heavy rainfall. It is an efficient way to maintain storm-centered resolution throughout a simulation. This option is designed for tropical systems, however, and is not generally suitable for extratropical cyclones because of their very different structure and dynamics. Today, Cyclone Maila in the western Pacific is a good example of why this matters. With the storm expected to strengthen and threaten parts of the region with damaging winds, heavy rainfall, flooding, and infrastructure impacts, a vortex-following nest helps keep the model focused on the cyclone as it evolves and moves across the domain. This is one of the reasons WRF remains such a powerful tool, not only for operational-style forecasting, but also for research into tropical cyclones, mesoscale dynamics, convection, and nested storm-scale processes. #WRF #NCAR #NSFNCAR #WRFMOSIT #Meteorology #AtmosphericScience #NumericalWeatherPrediction #WeatherModeling #MovingNest #VortexFollowingNest #TropicalCyclone #CycloneMaila #MesoscaleModeling #Forecasting #TropicalMeteorology

IITM new Study: Estimation of location-specific precipitation using Deep Neural Networks Research By: Bipin Kumar, Bhvisy Kumar Yadav, Soumyodeep Mukhopadhyay, Rakshit Rohan, Bhupendra Bahadur Singh, Rajib Chattopadhyay, Nagraju Chilukoti & Atul Kumar Sahai IITM's latest research on hyperlocal weather forecasting using Deep Neural Networks (DNN), published in Theoretical and Applied Climatology (TAAC), Volume 157, article number 223, (2026) The developed model provides high-precision precipitation data based on specific coordinates (latitude and longitude). This location specific forecast technique has significant real-world applications: from helping farmers optimize their crop management to providing urban citizens with street-level weather updates. 📖 Read the full paper here: https://lnkd.in/dhuGc__P #Meteorology #DeepLearning #AI #ClimateScience #Hyperlocal Ministry of Earth Sciences