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: Correcting Systematic and State‐Dependent Errors in the NOAA FV3‐GFS Using Neural Networks
Abstract Weather forecasts made with imperfect models contain state‐dependent errors. Data assimilation (DA) partially corrects these errors with new information from observations. As such, the corrections, or “analysis increments,” produced by the DA process embed information about model errors. An attempt is made here to extract that information to improve numerical weather prediction. Neural networks (NNs) are trained to predict corrections to the systematic error in the National Oceanic and Atmospheric Administration's FV3‐GFS model based on a large set of analysis increments. A simple NN focusing on an atmospheric column significantly improves the estimated model error correction relative to a linear baseline. Leveraging large‐scale horizontal flow conditions using a convolutional NN, when compared to the simple column‐oriented NN, does not improve skill in correcting model error. The sensitivity of model error correction to forecast inputs is highly localized by vertical level and by meteorological variable, and the error characteristics vary across vertical levels. Once trained, the NNs are used to apply an online correction to the forecast during model integration. Improvements are evaluated both within a cycled DA system and across a collection of 10‐day forecasts. It is found that applying state‐dependent NN‐predicted corrections to the model forecast improves the overall quality of DA and improves the 10‐day forecast skill at all lead times.  more » « less
Award ID(s):
2019625
PAR ID:
10383670
Author(s) / Creator(s):
 ;  ;  ;  ;  ;  
Publisher / Repository:
DOI PREFIX: 10.1029
Date Published:
Journal Name:
Journal of Advances in Modeling Earth Systems
Volume:
14
Issue:
11
ISSN:
1942-2466
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; (, Journal of Atmospheric and Oceanic Technology)
    Abstract Radiative transfer (RT) is a crucial but computationally expensive process in numerical weather/climate prediction. We develop neural networks (NN) to emulate a common RT parameterization called the Rapid Radiative Transfer Model (RRTM), with the goal of creating a faster parameterization for the Global Forecast System (GFS) v16. In previous work we emulated a highly simplified version of the shortwave RRTM only—excluding many predictor variables, driven by Rapid Refresh forecasts interpolated to a consistent height grid, using only 30 sites in the Northern Hemisphere. In this work we emulate the full shortwave and longwave RRTM—with all predictor variables, driven by GFSv16 forecasts on the native pressure–sigma grid, using data from around the globe. We experiment with NNs of widely varying complexity, including the U-net++ and U-net3+ architectures and deeply supervised training, designed to ensure realistic and accurate structure in gridded predictions. We evaluate the optimal shortwave NN and optimal longwave NN in great detail—as a function of geographic location, cloud regime, and other weather types. Both NNs produce extremely reliable heating rates and fluxes. The shortwave NN has an overall RMSE/MAE/bias of 0.14/0.08/−0.002 K day−1for heating rate and 6.3/4.3/−0.1 W m−2for net flux. Analogous numbers for the longwave NN are 0.22/0.12/−0.0006 K day−1and 1.07/0.76/+0.01 W m−2. Both NNs perform well in nearly all situations, and the shortwave (longwave) NN is 7510 (90) times faster than the RRTM. Both will soon be tested online in the GFSv16. Significance StatementRadiative transfer is an important process for weather and climate. Accurate radiative transfer models exist, such as the RRTM, but these models are computationally slow. We develop neural networks (NNs), a type of machine learning model that is often computationally fast after training, to mimic the RRTM. We wish to accelerate the RRTM by orders of magnitude without sacrificing much accuracy. We drive both the NNs and RRTM with data from the GFSv16, an operational weather model, using locations around the globe during all seasons. We show that the NNs are highly accurate and much faster than the RRTM, which suggests that the NNs could be used to solve radiative transfer inside the GFSv16. 
    more » « less
  2. ; ; ; (, Journal of Geophysical Research: Atmospheres)
    We estimate the effect of model deficiencies in the Global Forecast System that lead to systematic forecast errors, as a first step toward correcting them online (i.e., within the model) as in Danforth & Kalnay (2008a, 2008b). Since the analysis increments represent the corrections that new observations make on the 6 h forecast in the analysis cycle, we estimate the model bias corrections from the time average of the analysis increments divided by 6 h, assuming that initial model errors grow linearly and first ignoring the impact of observation bias. During 2012–2016, seasonal means of the 6 h model bias are generally robust despite changes in model resolution and data assimilation systems, and their broad continental scales explain their insensitivity to model resolution. The daily bias dominates the submonthly analysis increments and consists primarily of diurnal and semidiurnal components, also requiring a low dimensional correction. Analysis increments in 2015 and 2016 are reduced over oceans, which we attribute to improvements in the specification of the sea surface temperatures. These results provide support for future efforts to make online correction of the mean, seasonal, and diurnal and semidiurnal model biases of Global Forecast System to reduce both systematic and random errors, as suggested by Danforth & Kalnay (2008a, 2008b). It also raises the possibility that analysis increments could be used to provide guidance in testing new physical parameterizations. 
    more » « less
  3. ; ; ; (, Journal of Climate)
    Abstract The predictability of sea ice during extreme sea ice loss events on subseasonal (daily to weekly) time scales is explored in dynamical forecast models. These extreme sea ice loss events (defined as the 5th percentile of the 5-day change in sea ice extent) exhibit substantial regional and seasonal variability; in the central Arctic Ocean basin, most subseasonal rapid ice loss occurs in the summer, but in the marginal seas rapid sea ice loss occurs year-round. Dynamical forecast models are largely able to capture the seasonality of these extreme sea ice loss events. In most regions in the summertime, sea ice forecast skill is lower on extreme sea ice loss days than on nonextreme days, despite evidence that links these extreme events to large-scale atmospheric patterns; in the wintertime, the difference between extreme and nonextreme days is less pronounced. In a damped anomaly forecast benchmark estimate, the forecast error remains high following extreme sea ice loss events and does not return to typical error levels for many weeks; this signal is less robust in the dynamical forecast models but still present. Overall, these results suggest that sea ice forecast skill is generally lower during and after extreme sea ice loss events and also that, while dynamical forecast models are capable of simulating extreme sea ice loss events with similar characteristics to what we observe, forecast skill from dynamical models is limited by biases in mean state and variability and errors in the initialization. Significance Statement We studied weather model forecasts of changes in Arctic sea ice extent on day-to-day time scales in different regions and seasons. We were especially interested in extreme sea ice loss days, or days in which sea ice melts very quickly or is reduced due to diverging forces such as winds, ocean currents, and waves. We find that forecast models generally capture the observed timing of extreme sea ice loss days. We also find that forecasts of sea ice extent are worse on extreme sea ice loss days compared to typical days, and that forecast errors remain elevated following extreme sea ice loss events. 
    more » « less
  4. ; (, Geophysical Research Letters)
    Abstract The traditional method for estimating weather forecast sensitivity to initial conditions uses adjoint models, which are limited to short lead times due to linearization around a control forecast. The advent of deep‐learning frameworks enables a new approach using backpropagation and gradient descent to iteratively optimize initial conditions, minimizing forecast errors. We apply this approach to the June 2021 Pacific Northwest heatwave using the GraphCast model, yielding over 90% reduction in 10‐day forecast errors over the Pacific Northwest. Similar improvements are found for Pangu‐Weather model forecasts initialized with the GraphCast‐derived optimal, suggesting that model error is an unimportant part of the perturbations. Eliminating small scales from the perturbations also yields similar forecast improvements. Extending the length of the optimization window, we find forecast improvement to about 23 days, suggesting atmospheric predictability at the upper end of recent estimates. 
    more » « less
  5. ; ; (, Geophysical Research Letters)
    Abstract A simple and efficient Bayesian machine learning (BML) training algorithm, which exploits only a 20‐year short observational time series and an approximate prior model, is developed to predict the Niño 3 sea surface temperature (SST) index. The BML forecast significantly outperforms model‐based ensemble predictions and standard machine learning forecasts. Even with a simple feedforward neural network (NN), the BML forecast is skillful for 9.5 months. Remarkably, the BML forecast overcomes the spring predictability barrier to a large extent: the forecast starting from spring remains skillful for nearly 10 months. The BML algorithm can also effectively utilize multiscale features: the BML forecast of SST using SST, thermocline, and windburst improves on the BML forecast using just SST by at least 2 months. Finally, the BML algorithm also reduces the forecast uncertainty of NNs and is robust to input perturbations. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email