CoChat
Sign Up
Cited by 224Meteorological Phenomena and SimulationsClimate variability and modelsHydrological Forecasting Using AI

The DOI matches its Crossref record; 71 of the 73 references it cites were found in a registry or an index.

Analog forecasting of extreme-causing weather patterns using deep learning

Open access
Chattopadhyay, Ashesh, Nabizadeh, Ebrahim, Hassanzadeh, Pedram·2020·Journal of Advances in Modeling Earth Systems
DOI 10.1029/2019ms001958PMC PMC7375135
Open this paper in CoChat — free account

Abstract

Numerical weather prediction models require ever‐growing computing time and resources but, still, have sometimes difficulties with predicting weather extremes. We introduce a data‐driven framework that is based on analog forecasting (prediction using past similar patterns) and employs a novel deep learning pattern‐recognition technique (capsule neural networks, CapsNets) and an impact‐based autolabeling strategy. Using data from a large‐ensemble fully coupled Earth system model, CapsNets are trained on midtropospheric large‐scale circulation patterns (Z500) labeled 0–4 depending on the existence and geographical region of surface temperature extremes over North America several days ahead. The trained networks predict the occurrence/region of cold or heat waves, only using Z500, with accuracies (recalls) of 69–45% (77–48%) or 62–41% (73–47%) 1–5 days ahead. Using both surface temperature and Z500, accuracies (recalls) with CapsNets increase to 80% (88%). In both cases, CapsNets outperform simpler techniques such as convolutional neural networks and logistic regression, and their accuracy is least affected as the size of the training set is reduced. The results show the promises of multivariate data‐driven frameworks for accurate and fast extreme weather predictions, which can potentially augment numerical weather prediction efforts in providing early warnings.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1029/2019ms001958

Cites

71 of 71 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 Cluster centers of T2m anomalies at the onsets and Z500 patterns of 3 days earlier. The top (bottom) two rows correspond to summers (winters). S0 (W0) shows the average of T2m and Z500 patterns from days with no heat wave (cold spell). S1–S4 and W1–W4 are obtained from K‐means clustering the anomalous T2m patterns at onsets into four classes, which roughly separates the extreme events into four geographical regions: Northern Canada (S1), Western United States‐Canada (S2), Southern United States (S3), and Eastern United States‐Canada (S4) in summers and Northwest United States‐Canada (W1), Alaska (W2), Northeast United States‐Canada (W3), and Northern Canada (W4) in winters. Rows 1 and 3 show the cluster centers, while rows 2 and 4 show the average of Z500 patterns 3 days before the onsets for each cluster.
Figure 1 Cluster centers of T2m anomalies at the onsets and Z500 patterns of 3 days earlier. The top (bottom) two rows correspond to summers (winters). S0 (W0) shows the average of T2m and Z500 patterns from days with no heat wave (cold spell). S1–S4 and W1–W4 are obtained from K‐means clustering the anomalous T2m patterns at onsets into four classes, which roughly separates the extreme events into four geographical regions: Northern Canada (S1), Western United States‐Canada (S2), Southern United States (S3), and Eastern United States‐Canada (S4) in summers and Northwest United States‐Canada (W1), Alaska (W2), Northeast United States‐Canada (W3), and Northern Canada (W4) in winters. Rows 1 and 3 show the cluster centers, while rows 2 and 4 show the average of Z500 patterns 3 days before the onsets for each cluster.CC BY · PubMed Central
Figure 2 Schematic of the data‐driven framework for prediction of cold spells based on Z500 patterns of 3 days earlier. Using the impact‐based autolabeling strategy, Z500 patterns are labeled W0, W1, W2, W3, or W4, depending on the cluster index of T2m 3 days ahead. The panels at the top show examples of T2m patterns at the onset and the corresponding Z500 patterns (from 3 days earlier) for each cluster. Only the Z500 patterns and their labels are inputted into the CapsNet during training. Once trained, the CapsNet can predict, from a given Z500 pattern, the T2m cluster index of 3 days later, thus predicting the occurrence and geographical region of cold spells. For the shown test example, a cold spell in W3 in 3 days is predicted. Note that for winters, Z500 patterns over a larger domain that extends across the Pacific Ocean to 145°E are inputted into the CapsNets, but a smaller domain is shown in this figure for better illustration (see Section 3.2). Separate CapsNets are trained for each season and each prediction lead time.
Figure 2 Schematic of the data‐driven framework for prediction of cold spells based on Z500 patterns of 3 days earlier. Using the impact‐based autolabeling strategy, Z500 patterns are labeled W0, W1, W2, W3, or W4, depending on the cluster index of T2m 3 days ahead. The panels at the top show examples of T2m patterns at the onset and the corresponding Z500 patterns (from 3 days earlier) for each cluster. Only the Z500 patterns and their labels are inputted into the CapsNet during training. Once trained, the CapsNet can predict, from a given Z500 pattern, the T2m cluster index of 3 days later, thus predicting the occurrence and geographical region of cold spells. For the shown test example, a cold spell in W3 in 3 days is predicted. Note that for winters, Z500 patterns over a larger domain that extends across the Pacific Ocean to 145°E are inputted into the CapsNets, but a smaller domain is shown in this figure for better illustration (see Section 3.2). Separate CapsNets are trained for each season and each prediction lead time.CC BY · PubMed Central
Figure 3 Performance of CapsNets in predicting heat waves and cold spells using Z500 patterns at various lead times. (a and b) The symbols show the accuracy at different lead times for each cluster: star (0), triangle (1), square (2), diamond (3), and circle (4). The solid (dashed) lines show the total accuracy (recall). (c and d) Total accuracy at prediction lead times 1, 3, and 5 days versus the size of the training set ( N=750; fractions are rounded to the nearest integer if needed). Results in (a) and (b) are obtained with the largest training set. The symbols show the accuracy averaged over three randomly drawn pairs of training/testing sets. The lines and their shading depict the mean and ±1SD of accuracy or recall computed for the three pairs; the shadings are narrow, demonstrating the small SD and robustness of the results.
Figure 3 Performance of CapsNets in predicting heat waves and cold spells using Z500 patterns at various lead times. (a and b) The symbols show the accuracy at different lead times for each cluster: star (0), triangle (1), square (2), diamond (3), and circle (4). The solid (dashed) lines show the total accuracy (recall). (c and d) Total accuracy at prediction lead times 1, 3, and 5 days versus the size of the training set ( N=750; fractions are rounded to the nearest integer if needed). Results in (a) and (b) are obtained with the largest training set. The symbols show the accuracy averaged over three randomly drawn pairs of training/testing sets. The lines and their shading depict the mean and ±1SD of accuracy or recall computed for the three pairs; the shadings are narrow, demonstrating the small SD and robustness of the results.CC BY · PubMed Central
Figure 4 Same as Figure 3 but for the performance of CapsNets in predicting heat waves and cold spells using both T2m and Z500 patterns. The shadings show ±1SD; the SD values are higher in the multivariate approach. In (c) and (d), the change of accuracies with the size of the training set is small for lead times of 1, 3, and 5 days (labels not shown).
Figure 4 Same as Figure 3 but for the performance of CapsNets in predicting heat waves and cold spells using both T2m and Z500 patterns. The shadings show ±1SD; the SD values are higher in the multivariate approach. In (c) and (d), the change of accuracies with the size of the training set is small for lead times of 1, 3, and 5 days (labels not shown).CC BY · PubMed Central
Figure 5 Same as Figure 3 but for the performance of ConvNets and logistic regression in predicting heat waves and cold spells using Z500 patterns at various lead times. In (a) and (b), the symbols and black lines correspond to ConvNets, while the green lines correspond to logistic regression. In (c) and (d), the solid (dashed) lines correspond to ConvNets (logistic regression).
Figure 5 Same as Figure 3 but for the performance of ConvNets and logistic regression in predicting heat waves and cold spells using Z500 patterns at various lead times. In (a) and (b), the symbols and black lines correspond to ConvNets, while the green lines correspond to logistic regression. In (c) and (d), the solid (dashed) lines correspond to ConvNets (logistic regression).CC BY · PubMed Central
Figure 6 Same as Figure 4 but for the performance of ConvNets and logistic regression in predicting heat waves and cold spells using both T2m and Z500 patterns at various lead times. In (a) and (b), the symbols and black lines correspond to ConvNets, while the green lines correspond to logistic regression. In (c) and (d), the solid (dashed) lines correspond to ConvNets (logistic regression).
Figure 6 Same as Figure 4 but for the performance of ConvNets and logistic regression in predicting heat waves and cold spells using both T2m and Z500 patterns at various lead times. In (a) and (b), the symbols and black lines correspond to ConvNets, while the green lines correspond to logistic regression. In (c) and (d), the solid (dashed) lines correspond to ConvNets (logistic regression).CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 73 references

Research works printed but not linked: 1

  1. 10.1034/j.1600-0870.1994.t01-2-00006.x · DOI

Cited by

Sorted by citations

See all 224

Related

Metadata

Authors
Chattopadhyay, A., Nabizadeh, E., & Hassanzadeh, P.
Affiliations
Rice University (US)
Corresponding author
Hassanzadeh, Pedram
ORCID
Venue
Journal of Advances in Modeling Earth Systems
Publisher
American Geophysical Union (AGU)
Published
2020-02
Volume
12
Issue
2
Pages
e2019MS001958-e2019MS001958
ISSN
1942-2466 (ISSN-L)
DOI
10.1029/2019ms001958
PMC
PMC7375135
OpenAlex
W2965580012
Cited by
224
References
73
Citations by year
2026: 21 · 2025: 30 · 2024: 42 · 2023: 34 · 2022: 35 · 2021: 46 · 2020: 16
FWCI
10.73
Type
journal article
Language
en
Open access
Licence
Locations
  • Journal of Advances in Modeling Earth Systems (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • EarthArXiv (California Digital Library) (submittedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • EarthArXiv (OSF Preprints) (submittedVersion)
  • Zenodo (CERN European Organization for Nuclear Research) (submittedVersion, cc-by)
  • arXiv (Cornell University)
Funders
  • Gulf Research Program
  • National Aeronautics and Space Administration (80NSSC17K0266)
Topics
  • Physical Sciences › Earth and Planetary Sciences › Atmospheric Science › Meteorological Phenomena and Simulations
  • Physical Sciences › Environmental Science › Global and Planetary Change › Climate variability and models
  • Physical Sciences › Environmental Science › Environmental Engineering › Hydrological Forecasting Using AI
Keywords
analog forecasting, extreme weather prediction, cold waves, heat waves, capsule networks, Earth system model ensembles, early warning

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review