skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: Inverted Transformers for Effective Post‐Calibration in Multisite Wind Power Forecasting
ABSTRACT Accurate wind power forecasts are essential for energy management and resource allocation. However, because of complex weather dynamics and other nonlinearities, it is exceedingly difficult to forecast wind power on the multisite level for dozens of wind farms at once. This paper proposes a hybridized approach that leverages deep learning to predict future forecast errors from physics‐based numerical weather prediction (NWP) model estimates. Utilizing errors from NWP forecasts allows integration of critical atmospheric and meteorological dynamics into the forecasting model, and we demonstrate the importance of post‐calibration based on the physics versus pure data‐driven wind power prediction. This post‐calibration approach is enabled by the inverted transformer architecture, which efficiently and effectively learns meaningful wind farm variate representations, resulting in accurate spatiotemporal corrections to the forecasts. We also investigate modifying the iTransformer with a new embedding approach, named SpaceEmbed, that explicitly encodes spatial distance information into the network. The proposed approach is validated with a case study using real‐world data and forecasts from the Electric Reliability Council of Texas (ERCOT) in 2015 for 74 wind farms in Texas at different time scales. Using the high sustained limit as the metric for power generation, the iTransformer outperforms other state‐of‐the‐art deep learning forecasting methods, succeeding at the post‐calibration task by reducing NWP forecast error by up to 33% on average.  more » « less
Award ID(s):
2226348
PAR ID:
10653486
Author(s) / Creator(s):
 ;  ;  ;  ;  ;  
Publisher / Repository:
Wiley
Date Published:
Journal Name:
Wind Energy
Volume:
28
Issue:
10
ISSN:
1095-4244
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; (, Energies)
    Accurate and uncertainty-aware wind power forecasting is essential for reliable and cost-effective power system operations. This paper presents a novel probabilistic forecasting framework based on diffusion probabilistic models. We adopted a two-stage modeling strategy—a deterministic predictor first generates baseline forecasts, and a conditional diffusion model then learns the distribution of residual errors. Such a two-stage decoupling strategy improves learning efficiency and sharpens uncertainty estimation. We employed the elucidated diffusion model (EDM) to enable flexible noise control and enhance calibration, stability, and expressiveness. For the generative backbone, we introduced a time-series-specific diffusion Transformer (TimeDiT) that incorporates modular conditioning to separately fuse numerical weather prediction (NWP) inputs, noise, and temporal features. The proposed method was evaluated using the public database from ten wind farms in the Global Energy Forecasting Competition 2014 (GEFCom2014). We further compared our approach with two popular baseline models, i.e., a distribution parameter regression model and a generative adversarial network (GAN)-based model. Results showed that our method consistently achieves superior performance in both deterministic metrics and probabilistic accuracy, offering better forecast calibration and sharper distributions. 
    more » « less
  2. (, Energy)
    To make a future run by renewable energy possible, we must design our power system to seamlessly collect, store, and transport the Earth's naturally occurring flows of energy – namely the sun and the wind. Such a future will require that accurate representations of wind and solar resources and their associated variability permeate power systems planning and operational tools. Practically speaking, we must merge weather and power systems modeling. Although many meteorological phenomena that affect wind and solar power production are well-studied in isolation, no coordinated effort has sought to improve medium- and long-term power systems planning using numerical weather prediction (NWP) models. One modern open-source NWP tool – the weather research and forecasting (WRF) model – offers the complexity and flexibility required to integrate weather prediction with a power systems model in any region. However, there are over one million distinct ways to set up WRF. Here, we present a methodology for optimizing the WRF model physics for forecasting wind power density and solar irradiance using a genetic algorithm. The top five setups created by our algorithm outperform all of the recommended setups. Using the simulation results, we train a random forest model to identify which WRF parameters contribute to the lowest forecast errors and produce plots depicting the performance of key physics options to guide energy researchers in quickly setting up an accurate WRF model. 
    more » « less
  3. ; ; ; ; (, Geoscientific Model Development)
    With rising global temperatures, urban environments are increasingly vulnerable to heat stress, often exacerbated by the Urban Heat Island (UHI) effect. While most UHI research has focused on large metropolitan areas around the world, relatively smaller-sized cities (with a population 100 000–300 000) remain understudied despite their growing exposure to extreme heat and meteorological significance. In particular, urban heat advection (UHA), the transport of heat by mean winds, remains a key but underexplored mechanism in most modeling frameworks. High-resolution numerical weather prediction (NWP) models are essential tools for simulating urban hydrometeorological conditions, yet most prior evaluations have focused on retrospective reanalysis products rather than forecasts. In this study, we assess the performance of a widely used operational weather forecast model, the High-Resolution Rapid Refresh (HRRR), as a representative example of current NWP systems. We investigate its ability to predict spatial and temporal patterns of urban heat and UHA within and around Lubbock, Texas, a small-sized city located in a semi-arid environment in the southwestern US. Using data collected between 1 September 2023, and 31 August 2024 from the Urban Heat Island Experiment in Lubbock, Texas (U-HEAT) network and five West Texas Mesonet stations, we compare 18 h forecasts against in situ observations. HRRR forecasts exhibit a consistent nighttime cold bias at both urban and rural sites, a daytime warm bias at rural locations, and a pervasive dry bias across all seasons. The model also systematically overestimates near-surface wind speeds, further limiting its ability to accurately predict UHA. Although HRRR captures the expected slower nocturnal cooling in urban areas, it does not well capture advective heat transport under most wind regimes. Our findings reveal both systematic biases and urban representation limitations in current high-resolution NWP forecasts. Our forecast–observation comparisons underscore the need for improved urban parameterizations and evaluation frameworks focused on forecast skill, with important implications for heat-risk warning systems and forecasting in small and mid-sized cities. 
    more » « less
  4. ; ; (, Geophysical Research Letters)
    Abstract The development of deep learning (DL) weather forecasting models has made rapid progress and achieved comparable or better skill than traditional Numerical Weather prediction (NWP) models, which are generally computationally intensive. However, applications of these DL models have yet to be fully explored, including for severe convective events. We evaluate the DL model Pangu‐Weather in forecasting tornadic environments with one‐day lead times using convective available potential energy (CAPE), 0–6 bulk wind difference (BWD6), and 0–3 km storm‐relative helicity (SRH3). We also compare its performance to the National Centers for Environmental Prediction (NCEP)'s Global Forecast System (GFS), a traditional NWP model. Pangu‐Weather generally outperforms GFS in predicting BWD6 and SRH3 at the closest grid point and hour of the storm report. However, Pangu‐Weather tends to underpredict the maximum values of all convective parameters in the 1–2 hr before the storm across the surrounding grid points compared to the GFS. 
    more » « less
  5. ; ; (, Weather and Forecasting)
    Abstract Pure artificial intelligence (AI)-based weather prediction (AIWP) models have made waves within the scientific community and the media, claiming superior performance to numerical weather prediction (NWP) models. However, these models often lack impactful output variables such as precipitation. One exception is Google DeepMind’s GraphCast model, which became the first mainstream AIWP model to predict precipitation, but performed only limited verification. We present an analysis of the ECMWF’s Integrated Forecasting System (IFS)-initialized (GRAPIFS) and the NCEP’s Global Forecast System (GFS)-initialized (GRAPGFS) GraphCast precipitation forecasts over the contiguous United States and compare to results from the GFS and IFS models using 1) grid-based, 2) neighborhood, and 3) object-oriented metrics verified against the fifth major global reanalysis produced by ECMWF (ERA5) and the NCEP/Environmental Modeling Center (EMC) stage IV precipitation analysis datasets. We affirmed that GRAPGFSand GRAPIFSperform better than the GFS and IFS in terms of root-mean-square error and stable equitable errors in probability space, but the GFS and IFS precipitation distributions more closely align with the ERA5 and stage IV distributions. Equitable threat score also generally favored GraphCast, particularly for lower accumulation thresholds. Fractions skill score for increasing neighborhood sizes shows greater gains for the GFS and IFS than GraphCast, suggesting the NWP models may have a better handle on intensity but struggle with the location. Object-based verification for GraphCast found positive area biases at low accumulation thresholds and large negative biases at high accumulation thresholds. GRAPGFSsaw similar performance gains to GRAPIFSwhen compared to their NWP counterparts, but initializing with the less familiar GFS conditions appeared to lead to an increase in light precipitation. Significance StatementPure artificial intelligence (AI)-based weather prediction (AIWP) has exploded in popularity with promises of better performance and faster run times than numerical weather prediction (NWP) models. However, less attention has been paid to their capability to predict impactful, sensible weather like precipitation, precipitation type, or specific meteorological features. We seek to address this gap by comparing precipitation forecast performance by an AI model called GraphCast to the Global Forecast System (GFS) and the Integrated Forecasting System (IFS) NWP models. While GraphCast does perform better on many verification metrics, it has some limitations for intense precipitation forecasts. In particular, it less frequently predicts intense precipitation events than the GFS or IFS. Overall, this article emphasizes the promise of AIWP while at the same time stresses the need for robust verification by domain experts. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Text Encoded Conversion
  • XML
  • Save / Share this Record
  • Send to Email