Search NASA⌕ Search

SEARCH · Search NASA

Results for “Weatherization training”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 145 records · Page 8

Beyond the cockpit: The visual world as a flight instrument

The use of cockpit instruments to guide flight control is not always an option (e.g., low level rotorcraft flight). Under such circumstances the pilot must use out-the-window information for control and navigation. Thus it is important to determine the basis of visually guided flight for several reasons: (1) to guide the design and construction of the visual displays used in training simulators; (2) to allow modeling of visibility restrictions brought about by weather, cockpit constraints, or distortions introduced by sensor systems; and (3) to aid in the development of displays that augment the cockpit window scene and are compatible with the pilot's visual extraction of information from the visual scene. The authors are actively pursuing these questions. We have on-going studies using both low-cost, lower fidelity flight simulators, and state-of-the-art helicopter simulation research facilities. Research results will be presented on: (1) the important visual scene information used in altitude and speed control; (2) the utility of monocular, stereo, and hyperstereo cues for the control of flight; (3) perceptual effects due to the differences between normal unaided daylight vision, and that made available by various night vision devices (e.g., light intensifying goggles and infra-red sensor displays); and (4) the utility of advanced contact displays in which instrument information is made part of the visual scene, as on a 'scene linked' head-up display (e.g., displaying altimeter information on a virtual billboard located on the ground).

Johnson, W. W.↗

Lightning forecasting studies using LDAR, LLP, field mill, surface mesonet, and Doppler radar data

The ultimate goal of this research is to develop rules, algorithms, display software, and training materials that can be used by the operational forecasters who issue weather advisories for daily ground operations and launches by NASA and the United States Air Force to improve real-time forecasts of lightning. Doppler radar, Lightning Detection and Ranging (LDAR), Lightning Location and Protection (LLP), field mill (Launch Pad Lightning Warning System -- LPLWS), wind tower (surface mesonet) and additional data sets have been utilized in 10 case studies of thunderstorms in the vicinity of KSC during the summers of 1994 and 1995. These case studies reveal many intriguing aspects of cloud-to-ground, cloud-to-cloud, in-cloud, and cloud-to-air lightning discharges in relation to radar thunderstorm structure and evolution. They also enable the formulation of some preliminary working rules of potential use in the forecasting of initial and final ground strike threat. In addition, LDAR and LLP data sets from 1993 have been used to quantify the lightning threat relative to the center and edges of LDAR discharge patterns. Software has been written to overlay and display the various data sets as color imagery. However, human intervention is required to configure the data sets for proper intercomparison. Future efforts will involve additional software development to automate the data set intercomparisons, to display multiple overlay combinations in a windows format, and to allow for animation of the imagery. The software package will then be used as a tool to examine more fully the current cases and to explore additional cases in a timely manner. This will enable the formulation of more general and reliable forecasting guidelines and rules.

Forbes, Gregory S.↗

Machine Learning based Aircraft Performance Model Estimation for Trajectory Prediction

The accurate prediction of aircraft trajectory by ground-based decision support tools is a critical component of air traffic management in the US National Airspace System (NAS). Accurate predictions of where the aircraft will be in the future or when they will arrive at specific locations (e.g., fixes) is a key enabler for sequencing and efficient arrival management of flights. Traditional physics based aircraft trajectory prediction relies on a simplified point-mass total energy model whose parameters are referred to as Aircraft Performance Model (APM) parameters. Even though the performance coefficients and weight of an aircraft are a vital part of the aircraft performance model’s predictions and accuracy, these coefficients are proprietary in nature and therefore, unavailable to decision-support tools. Current approaches freeze some coefficients to default base of aircraft data (BADA) values and optimize others. However, the APM parameters are highly coupled by the flight dynamics and prioritizing one parameter over others leads to bias and skewed predictions. To alleviate this problem, we provide a combined optimization framework to predict all the critical (thrust, drag and weight) APM parameters. This paper is focused on training Machine Learning (ML) models that map historical flights to optimized APM parameters that provide the best fit (in terms of prediction error). Our dataset obtained from NASA’s Sherlock data warehouse is comprised of thousands of historical flights and includes weather and track data collected from 2019. Using different subsets of relevant features (e.g., aircraft type), we trained several ML models to estimate the aircraft’s take off weight, drag polar coefficients (both parasitic and lift induced), and thrust settings (multiplier applied to the maximum engine thrust). The chosen flights are from three of the most common aircraft types (B738, B737, and A320) arriving at four airports (LAX, DEN, MSP, and DFW). Our ML approach is comprised of two different solutions: 1- using a subset of features that are known prior to the flight departure and do not change during flight (such as engine type, current temperature at departure & destination airports, aircraft type) and 2 - using a subset of temporal features of the flight trajectory (such as cruise altitude, Mach, airspeed, and rate of climb) in addition to the pre-departure features from the first solution. The labels or target variables are the APM parameters that were obtained by an optimized ordinary differential equations (ODE) fitting process (applied to individual flights). The ODE-fitting is very time intensive and is therefore performed offline. Thus, training an ML model to learn the relationship between the flight features and ODE-generated labels enables faster estimation of the APM parameters and is therefore amenable to real-time prediction. Various ML models including linear regression, random forest, XGBoost, and neural network were trained, and the results are compared. After model validation and hyperparameter-tuning, we observed that the Random Forest model outperformed the other three models by the overall mean square error (MSE) of 2% for the first solution and 1.5% for the second solution. Finally, the ML-derived parameters are compared against default BADA APM parameters using NASA’s Autonomy Development toolkit (ADK) simulation software. The simulation results for one of each aircraft type is shown and discussed.

Aida Sharif Rohani↗

Real-Time Traffic Information for Emergency Evacuation Operations: Phase A Final Report

There are many instances in which it is possible to plan ahead for an emergency evacuation (e.g., an explosion at a chemical processing facility). For those cases, if an accident (or an attack) were to happen, then the best evacuation plan for the prevailing network and weather conditions would be deployed. In other cases (e.g., the derailment of a train transporting hazardous materials), there may not be any previously developed plan to be implemented and decisions must be made ad-hoc on how to proceed with an emergency evacuation. In both situations, the availability of real-time traffic information plays a critical role in the management of the evacuation operations. To improve public safety during a vehicular emergency evacuation it is necessary to detect losses of road capacity (due to incidents, for example) as early as possible. Once these bottlenecks are identified, re-routing strategies must be determined in real-time and deployed in the field to help dissipate the congestion and increase the efficiency of the evacuation. Due to cost constraints, only large urban areas have traffic sensor deployments that permit access to some sort of real-time traffic information; any evacuation taking place in any other areas of the country would have to proceed without real-time traffic information. The latter was the focus of this SERRI/DHS (Southeast Region Research Initiative/Department of Homeland Security) sponsored project. That is, the main objective on the project was to improve the operations during a vehicular emergency evacuation anywhere by using newly developed real-time traffic-information-gathering technologies to assess traffic conditions and therefore to potentially detect incidents on the main evacuation routes. Phase A of the project consisted in the development and testing of a prototype system composed of sensors that are engineered in such a way that they can be rapidly deployed in the field where and when they are needed. Each one of these sensors is also equipped with their own power supply and a GPS (Global Positioning System) device to auto-determine its spatial location on the transportation network under surveillance. The system is capable of assessing traffic parameters by identifying and re-identifying vehicles in the traffic stream as those vehicles pass over the sensors. The system of sensors transmits, through wireless communication, real-time traffic information (travel time and other parameters) to a command and control center via an NTCIP (National Transportation Communication for ITS Protocol) -compatible interface. As an alternative, an existing NTCIP-compatible system accepts the real-time traffic information mentioned and broadcasts the traffic information to emergency managers, the media and the public via the existing channels. A series of tests, both in a controlled environment and on the field, were conducted to study the feasibility of rapidly deploying the system of traffic sensors and to assess its ability to provide real-time traffic information during an emergency evacuation. The results of these tests indicated that the prototype sensors are reliable and accurate for the type of application that is the focus of this project.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

ORBIT-2 Dataset for Scaling Exascale Vision Foundation Models for Weather and Climate Downscaling

This dataset release corresponds to the work conducted in ORBIT-2: Scaling Exascale Vision Foundation Models for Weather and Climate Downscaling, where large-scale AI methods were applied to improve climate and weather resolution. The collection integrates four widely used, publicly available datasets: ERA5, PRISM, DAYMET, and IMERG. To prepare the data for ORBIT-2 model training and evaluation, we applied a preprocessing pipeline that generates paired low-resolution and high-resolution samples, enabling supervised downscaling experiments. The transformation from coarse to fine scales was performed using bilinear regridding, consistent with the procedures described in WeatherBench2, a community benchmark for weather and climate AI models. This dataset supports the development and evaluation of foundation models designed for weather and climate downscaling at exascale. Additional details on methodology and applications can be found in Wang et al., ORBIT-2 (arXiv:2505.04802, 2025).

54 ENVIRONMENTAL SCIENCES↗

Smart Methane Emission Detection System Development (Final Report)

Working with the Department of Energy's National Energy Technology Laboratory, Southwest Research Institute® (SwRI®) developed a system to identify methane leaks reliably, accurately, and autonomously at critical midstream sections of the natural gas distribution network in real-time for the purpose of mitigating methane emissions using Optical Gas Imaging (OGI) cameras. SwRI's Smart Leak Detection – Methane (SLED/M) adds a high degree of automation to the process of methane leak detection to minimize sources of human error, minimize response time to a leak event, and maximize midstream visibility. Furthermore, SwRI has been working towards integrating Quantitative OGI (QOGI) capabilities into this existing technology. By leveraging Deep Learning, SwRI now has the capability to estimate fugitive emission leak rates quickly and reliably, which allows operators to detect emissions, quantify leak rate, prioritize repairs, and validate the repairs in a single instrument. The next generation QOGI technology leverages the same cameras used in Leak Detection and Repair (LDAR) programs, with improvements in safety and speed for traditional quantification-based repairs, ultimately leading to less overhead cost for the operators. The goals for this research were to develop two types of models with the following goals: Run in real-time on the edge (≥ 12 Hz), Classification: Achieve less than 5% false positive detection, Classification: Achieve ≥ 95% methane plume detection rate, Regression: achieve ≤ 10 standard cubic feet per hour (scfh) prediction > 70% of the time. In order to achieve these results, multiple infrared (IR) and other sensors were investigated in tandem with the midwave IR (MWIR) OGI to provide additional information to train the underlying models. Information on atmospheric conditions including humidity, temperature, pressure, and solar radiation was provided by a weather station. Several machine learning and deep learning architectures and methods, including looking at quantized classification networks and regressions networks, were explored. As further data was collected, curated, and labeled, it allowed for more refined regressive networks to be adequately trained, leading to better insight into the true flow rates being observed. An important valuable deliverable of this research effort was the development of an advanced network which underwent multiple iterations capable of giving a continuous output. The current network has a predicted mean average percentage error (MAPE) of 12.3% just outside our target goal of 10.00%, but an accuracy of 97.78% at ±50 scfh, well within the overall goal for the Department of Energy (DOE) program. Upon closer inspection, it was observed that more than 10% of datapoints contributing to the MAPE predictions were the result of low flow rate predictions and are beyond the sensitivity of instrument measurement as a result of normal operational variation and noise.

03 NATURAL GAS↗

Smart Methane Emission Detection System Development (Final Report)

Working with the Department of Energy’s National Energy Technology Laboratory, Southwest Research Institute® (SwRI®) developed a system to identify methane leaks reliably, accurately, and autonomously at critical midstream sections of the natural gas distribution network in real- time for the purpose of mitigating methane emissions using Optical Gas Imaging (OGI) cameras. SwRI’s Smart Leak Detection – Methane (SLED/M) adds a high degree of automation to the process of methane leak detection to minimize sources of human error, minimize response time to a leak event, and maximize midstream visibility. Furthermore, SwRI has been working towards integrating Quantitative OGI (QOGI) capabilities into this existing technology. By leveraging Deep Learning, SwRI now has the capability to estimate fugitive emission leak rates quickly and reliably, which allows operators to detect emissions, quantify leak rate, prioritize repairs, and validate the repairs in a single instrument. The next generation QOGI technology leverages the same cameras used in Leak Detection and Repair (LDAR) programs, with improvements in safety and speed for traditional quantification-based repairs, ultimately leading to less overhead cost for the operators. The goals for this research were to develop two types of models with the following goals: 1. Run in real-time on the edge (≥ 12 Hz) 2. Classification: Achieve less than 5% false positive detection 3. Classification: Achieve ≥ 95% methane plume detection rate 4. Regression: achieve ≤ 10 standard cubic feet per hour (scfh) prediction > 70% of the time In order to achieve these results, multiple infrared (IR) and other sensors were investigated in tandem with the midwave IR (MWIR) OGI to provide additional information to train the underlying models. Information on atmospheric conditions including humidity, temperature, pressure, and solar radiation was provided by a weather station. Several machine learning and deep learning architectures and methods, including looking at quantized classification networks and regressions networks, were explored. As further data was collected, curated, and labeled, it allowed for more refined regressive networks to be adequately trained, leading to better insight into the true flow rates being observed. An important valuable deliverable of this research effort was the development of an advanced network which underwent multiple iterations capable of giving a continuous output. The current network has a predicted mean average percentage error (MAPE) of 12.3% just outside our target goal of 10.00%, but an accuracy of 97.78% at ±50 scfh, well within the overall goal for the Department of Energy (DOE) program. Upon closer inspection, it was observed that more than 10% of datapoints contributing to the MAPE predictions were the result of low flow rate predictions and are beyond the sensitivity of instrument measurement as a result of normal operational variation and noise.

03 NATURAL GAS↗

Strictly Enforcing Invertibility and Conservation in CNN-Based Super Resolution for Scientific Datasets

Abstract Recently, deep convolutional neural networks (CNNs) have revolutionized image “super resolution” (SR), dramatically outperforming past methods for enhancing image resolution. They could be a boon for the many scientific fields that involve imaging or any regularly gridded datasets: satellite remote sensing, radar meteorology, medical imaging, numerical modeling, and so on. Unfortunately, while SR-CNNs produce visually compelling results, they do not necessarily conserve physical quantities between their low-resolution inputs and high-resolution outputs when applied to scientific datasets. Here, a method for “downsampling enforcement” in SR-CNNs is proposed. A differentiable operator is derived that, when applied as the final transfer function of a CNN, ensures the high-resolution outputs exactly reproduce the low-resolution inputs under 2D-average downsampling while improving performance of the SR schemes. The method is demonstrated across seven modern CNN-based SR schemes on several benchmark image datasets, and applications to weather radar, satellite imager, and climate model data are shown. The approach improves training time and performance while ensuring physical consistency between the super-resolved and low-resolution data. Significance Statement Recent advancements in using deep learning to increase the resolution of images have substantial potential across the many scientific fields that use images and image-like data. Most image super-resolution research has focused on the visual quality of outputs, however, and is not necessarily well suited for use with scientific data where known physics constraints may need to be enforced. Here, we introduce a method to modify existing deep neural network architectures so that they strictly conserve physical quantities in the input field when “super resolving” scientific data and find that the method can improve performance across a wide range of datasets and neural networks. Integration of known physics and adherence to established physical constraints into deep neural networks will be a critical step before their potential can be fully realized in the physical sciences.

54 ENVIRONMENTAL SCIENCES↗

Real-Time Traffic Information for Emergency Evacuations

There are many instances in which it is possible to plan ahead for an emergency evacuation (e.g., a chemical processing facility explosion). For those cases, if an accident were to happen, then the best evacuation plan for the prevailing network and weather conditions would be deployed. In other cases (e.g., the derailment of a train transporting hazardous materials), there may not be any previously developed plan to be implemented, and decisions must be made ad-hoc on how to proceed with an emergency evacuation. In both situations, the availability of real-time traffic information plays a critical role in the management of the evacuation operations. Due to cost constraints, only large urban areas have traffic sensor deployments that permit access to some sort of real-time traffic information; an evacuation taking place in other areas of the country would have to proceed without real-time traffic information. In order to improve operations during a vehicular emergency evacuation anywhere, a system of sensors that uses newly developed real-time traffic-information-gathering technologies to assess traffic conditions and to detect incidents on the main evacuation routes is presented in this paper. A series of tests, both in a controlled environment and in the field, were conducted to study the feasibility of such a system of traffic sensors and to assess its ability to provide real-time traffic information during an emergency evacuation. The results of these tests indicated that the prototype sensors are reliable and accurate for the type of application that is the focus of this paper.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Predicting weather impacts on corn production in a data-limited region using a transfer learning approach

The stability of food supply and prices may depend more on annual changes in yields from year-to-year variability in weather than on longer-term average changes from changing climatic conditions. However, the absence of high-quality data on crop yields at fine spatial resolutions in many regions of the world makes it challenging to statistically model their response to interannual variability in weather patterns. Therefore, there is a need for empirical methods that can project annual crop yield changes even in limited data regions. Here, we propose a transfer learning algorithm that uses high spatial resolution data from one region to project yields in another region with more limited data. The goal of our work is to understand what data types can be beneficial for transferring learning from a source region to a very different target region with more limited data. We utilize Long Short-Term Memory to develop a transfer learning model that is trained on historical county-level corn yield in the United States and predicts district-level corn yield variations in India. Even using smaller amounts of data in India, simulating a data-scarce region, we achieve an average root mean square error of 0.48 bu acre−1 in predicting interannual yield variations. Using Shapley values to interpret results, we explore the contribution of the different weather parameters to interannual yield variability and find a larger influence of precipitation-related variables. Our study demonstrates the usefulness of this method for transferring models of weather impacts on crop yields trained on a data-rich country to one with more limited data. It suggests the potential of applying the transfer learning model to mitigate the need for extensive raw data globally.

Vishwakarma, Srishti [ORNL] (ORCID:000000031674419↗

Medium-Range River Flood Forecasts Using a Long Short-Term Memory Network

River flooding and the impacts are a concern for decision makers throughout the United States. Accurate medium-range forecasts (~3-7 days) are critical for providing advanced outlooks to emergency management officials. Unfortunately, accurately forecasting rainfall-runoff and the subsequent rise and fall within rivers remain a challenge in hydrological modeling. While complex physical modeling systems are the standard for representing the hydrological processes, they are computationally demanding and can require extensive calibration. Further, uncertainties remain in the model parameters and input data. The use of machine learning can reduce some of the computational demand while maintaining high accuracy. Therefore, this project makes use of a Long Short-Term Memory (LSTM) network which explicitly accounts for the time-dependent nature of rainfall-runoff modeling. The developed LSTM was trained to predict river gauge height, or stage height, based on time-lagged input features which include: gauge height to initialize the model, the NASA Short-term Prediction Research and Transition Center’s instance of the Land Information System (SPoRT-LIS) relative soil moisture to describe the rainfall infiltration rate, and 6-hr Multi-Radar Multi-Sensor quantitative precipitation estimate (MRMS QPE). The developed LSTM based system is then used to produce 7-day forecasts with a 6-hr temporal resolution using three different quantitative precipitation forecasts (QPF) from the NWS’s Weather Prediction Center (WPC), the NCEP Global Forecast System (GFS) model and the National Blend of Models (NBM). This trained modeling system has been implemented as an experimental product at over 100 different rivers in collaboration with at multiple National Weather Service (NWS) Forecast Offices and River Forecast Centers (RFC) across the eastern half of the United States. The developed LSTM model achieved average Nash-Sutcliffe efficiency (NSE) 0.89 higher than the equivalent medium-range National Water Model ensemble member forecast over a 7-day forecast. In addition to the initial development and evaluation, this project has continued to expand. While the initial model was developed for precipitation dominated basins, expansion of the project has taken it to basins effected by snow melt. This presentation will provide an overview of the project with focus on recent developments on incorporating snow melt processes into the model.

Andrew T. White↗

Use of simplifier scenarios for CRM training

Cockpit resource management (CRM) at Metro Airlines is discussed. The process by which the program of CRM training was initiated is mentioned. Management aspects of various flying scenarios are considered. The transfer of training from the classroom to the field is assessed.

Weatherly, D.↗

Probabilistic Forecasting of Surface Ozone with a Novel Statistical Approach

The recent change in the Environmental Protection Agency's surface ozone regulation, lowering the surface ozone daily maximum 8-h average (MDA8) exceedance threshold from 75 to 70 ppbv, poses significant challenges to U.S. air quality (AQ) forecasters responsible for ozone MDA8 forecasts. The forecasters, supplied by only a few AQ model products, end up relying heavily on self-developed tools. To help U.S. AQ forecasters, this study explores a surface ozone MDA8 forecasting tool that is based solely on statistical methods and standard meteorological variables from the numerical weather prediction (NWP) models. The model combines the self-organizing map (SOM), which is a clustering technique, with a step wise weighted quadratic regression using meteorological variables as predictors for ozone MDA8. The SOM method identifies different weather regimes, to distinguish between various modes of ozone variability, and groups them according to similarity. In this way, when a regression is developed for a specific regime, data from the other regimes are also used, with weights that are based on their similarity to this specific regime. This approach, regression in SOM (REGiS), yields a distinct model for each regime taking into account both the training cases for that regime and other similar training cases. To produce probabilistic MDA8 ozone forecasts, REGiS weighs and combines all of the developed regression models on the basis of the weather patterns predicted by an NWP model. REGiS is evaluated over the San Joaquin Valley in California and the northeastern plains of Colorado. The results suggest that the model performs best when trained and adjusted separately for an individual AQ station and its corresponding meteorological site.

Ozone variables↗

Native American Technical Assistance and Training for Renewable Energy Resource Development and Electrical Generation Facilities Management

The Council of Energy Resource Tribes (CERT) will facilitate technical expertise and training of Native Americans in renewable energy resource development for electrical generation facilities, and distributed generation options contributing to feasibility studies, strategic planning and visioning. CERT will also provide information to Tribes on energy efficiency and energy management techniques.This project will provide facilitation and coordination of expertise from government agencies and private industries to interact with Native Americans in ways that will result in renewable energy resource development, energy efficiency program development, and electrical generation facilities management by Tribal entities. The intent of this cooperative agreement is to help build capacity within the Tribes to manage these important resources.

09 BIOMASS FUELS↗

Linking the subseasonal variability of the East Asia winter monsoon and the Madden-Julian Oscillation through wave disturbances along the subtropical jet

Despite an urgent demand for reliable subseasonal-to-seasonal (S2S) predictions to guide disaster preparedness, our current climate models show limited S2S prediction skill, particularly for precipitation, due to an inadequate understanding of the key processes that drive regional S2S variability. Here we demonstrate that the leading subseasonal variability mode of precipitation over the East Asian Winter Monsoon (EAWM) region is not only closely tied to the activity of the Madden-Julian Oscillation (MJO), but also linked to precipitation and temperature extremes worldwide, influenced by a circumglobal Rossby wave-train along the subtropical westerly jet. Despite a close phase-lock relationship between the MJO and subseasonal EAWM precipitation, our findings indicate that the MJO itself may only play a minor role in the subseasonal EAWM variability. Given its significant impact on the S2S variability of global weather extremes, we call for coordinated community efforts to enhance the understanding and prediction of the circumglobal Rossby wave-train.

Atmospheric science↗

A weather pattern responsible for increasing wildfires in the western United States

Abstract The western United States (U.S.) has been experiencing more severe wildfires, in part due to climate change, but the underlying synoptic patterns and their modulation in driving fire weather is unclear. Here we investigated the relationship between weather regimes (WRs) and fire weather indices, specifically vapor pressure deficit (VPD) and the Canadian Forest Fire Weather Index. By identifying five singular WRs using k-means clustering, we found that a particular regime (WR-2), one characterized by a distinct tripolar wave train pattern over the continental U.S., has exhibited an increased frequency since 1980. The ascribed WR-2 regime was found to be mainly responsible for rising trends in the fire weather indices, especially VPD. Further, the average fire indices of the WR-2 regime played a more important role than the frequency in shaping the rising trends in the fire weather indices. The increased frequency of the WR-2 WR was mainly attributed to anthropogenic forcing and, the year-to-year variation of the frequency was associated with sea surface temperature anomalies over the subtropical eastern Pacific. Human-induced climate change might have furthered the exacerbation of wildfire danger in the western U.S. by modulating the behaviors of WRs and fire weather indices.

Zhang, Wei (ORCID:0000000221698749)↗

Enhancing Short-Range Weather Forecasts through Temporal Variation Encoding: A Multiperiod Embedding Approach

Machine learning (ML) techniques have emerged as promising approaches to improve regional weather forecast accuracy and reliability through data-driven methods. We propose a novel ML-based weather forecasting model, the Multiperiod Embed Net (MPENet). A key distinguishing feature of MPENet is its explicit utilization of the inherent cyclic nature in weather dynamics, unlike the autoregressive strategies commonly used in other ML weather forecasting approaches. Critical cyclic structures are identified via Fourier analyses of dynamic time series. Cyclicity in the convolutional representation is achieved by transforming one-dimensional time series of meteorological variables into two-dimensional tensors based on identified periods. This approach enables the model to leverage intrinsic weather patterns, enhancing regional forecast performance. To demonstrate the effectiveness of MPENet, we conduct a comparative analysis with Nvidia’s FourCastNet. Both models are trained on High-Resolution Rapid Refresh (HRRR) data from 2015 to 2022, over a 192 km × 192 km region in Tennessee. The comparisons are performed locally at two specific locations known to have different weather dynamics due to orographic effects: Crossville, on the relatively flat Cumberland Plateau with fewer topographic airflow disruptions, and Oak Ridge, in the ridge-and-valley region, where airflow is heavily influenced by surrounding valleys and mountains. Our results indicate that FourCastNet achieves strong accuracy at very short lead times, while MPENet maintains competitive skill and shows advantages in capturing temporal evolution over longer periods. Cross-correlation analyses of MPENet and FourCastNet predictions with the HRRR data suggest that encoding critical cyclicity into the network architecture leads to improvements in the forecasting skill.

Artificial intelligence↗

Towards physics-inspired data-driven weather forecasting: integrating data assimilation with a deep spatial-transformer-based U-NET in a case study with ERA5

Abstract. There is growing interest in data-driven weather prediction (DDWP), e.g., using convolutional neural networks such as U-NET that are trained on data from models or reanalysis. Here, we propose three components, inspired by physics, to integrate with commonly used DDWP models in order to improve their forecast accuracy. These components are (1) a deep spatial transformer added to the latent space of U-NET to capture rotation and scaling transformation in the latent space for spatiotemporal data, (2) a data-assimilation (DA) algorithm to ingest noisy observations and improve the initial conditions for next forecasts, and (3) a multi-time-step algorithm, which combines forecasts from DDWP models with different time steps through DA, improving the accuracy of forecasts at short intervals. To show the benefit and feasibility of each component, we use geopotential height at 500 hPa (Z500) from ERA5 reanalysis and examine the short-term forecast accuracy of specific setups of the DDWP framework. Results show that the spatial-transformer-based U-NET (U-STN) clearly outperforms the U-NET, e.g., improving the forecast skill by 45 %. Using a sigma-point ensemble Kalman (SPEnKF) algorithm for DA and U-STN as the forward model, we show that stable, accurate DA cycles are achieved even with high observation noise. This DDWP+DA framework substantially benefits from large (O(1000)) ensembles that are inexpensively generated with the data-driven forward model in each DA cycle. The multi-time-step DDWP+DA framework also shows promise; for example, it reduces the average error by factors of 2–3. These results show the benefits and feasibility of these three components, which are flexible and can be used in a variety of DDWP setups. Furthermore, while here we focus on weather forecasting, the three components can be readily adopted for other parts of the Earth system, such as ocean and land, for which there is a rapid growth of data and need for forecast and assimilation.

54 ENVIRONMENTAL SCIENCES↗