scholarly journals Deep Learning Experiments for Tropical Cyclone Intensity Forecasts

2021 ◽  
Author(s):  
Xu Wenwei ◽  
Balaguru Karthik ◽  
August Andrew ◽  
Lalo Nicholas ◽  
Hodas Nathan ◽  
...  
Keyword(s):  
Deep Learning ◽  
Tropical Cyclone ◽  
Forecast Errors ◽  
Dynamical Models ◽  
Atlantic Basin ◽  
Intensity Prediction ◽  
Operational Forecasts ◽  
Tropical Cyclone Intensity Forecasts ◽  
Using Data ◽  
Mlp Model

AbstractReducing tropical cyclone (TC) intensity forecast errors is a challenging task that has interested the operational forecasting and research community for decades. To address this, we developed a deep learning (DL)-based Multilayer Perceptron (MLP) TC intensity prediction model. The model was trained using the global Statistical Hurricane Intensity Prediction Scheme (SHIPS) predictors to forecast the change in TC maximum wind speed for the Atlantic Basin. In the first experiment, a 24-hour forecast period was considered. To overcome sample size limitations, we adopted a Leave One Year Out (LOYO) testing scheme, where a model is trained using data from all years except one and then evaluated on the year that is left out. When tested on 2010–2018 operational data using the LOYO scheme, the MLP outperformed other statistical-dynamical models by 9-20%. Additional independent tests in 2019 and 2020 were conducted to simulate real-time operational forecasts, where the MLP model again outperformed the statistical-dynamical models by 5-22% and achieved comparable results as HWFI. The MLP model also correctly predicted more rapid intensification events than all the four operational TC intensity models compared. In the second experiment, we developed a lightweight MLP for 6-hour intensity predictions. When coupled with a synthetic TC track model, the lightweight MLP generated realistic TC intensity distribution in the Atlantic Basin. Therefore, the MLP-based approach has the potential to improve operational TC intensity forecasts, and will also be a viable option for generating synthetic TCs for climate studies.

scholarly journals A Novel Multiscale Intensity Metric for Evaluation of Tropical Cyclone Intensity Forecasts

2014 ◽  
Vol 71 (4) ◽  
pp. 1292-1304 ◽  
Author(s):  
Tomislava Vukicevic ◽  
Eric Uhlhorn ◽  
Paul Reasor ◽  
Bradley Klotz
Keyword(s):  
Wind Speed ◽  
Tropical Cyclone ◽  
Microwave Radiometer ◽  
Maximum Speed ◽  
Forecast Errors ◽  
Cyclone Intensity ◽  
Maximum Wind ◽  
Tropical Cyclone Intensity Forecasts ◽  
The Difference ◽  
The Impact

Abstract In this study, a new multiscale intensity (MSI) metric for evaluating tropical cyclone (TC) intensity forecasts is presented. The metric consists of the resolvable and observable, low-wavenumber intensity represented by the sum of amplitudes of azimuthal wavenumbers 0 and 1 for wind speed within the TC vortex at the radius of maximum wind and a stochastic residual, all determined at 10-m elevation. The residual wind speed is defined as the difference between an estimate of maximum speed and the low-wavenumber intensity. The MSI metric is compared to the standard metric that includes only the maximum speed. Using stepped-frequency microwave radiometer wind speed observations from TC aircraft reconnaissance to estimate the low-wavenumber intensity and the National Hurricane Center’s best-track (BT) intensity for the maximum wind speed estimate, it is shown that the residual intensity is well represented as a stochastic quantity with small mean, standard deviation, and absolute norm values that are within the expected uncertainty of the BT estimates. The result strongly suggests that the practical predictability of TC intensity is determined by the observable and resolvable low-wavenumber intensity within the vortex. Verification of a set of high-resolution numerical forecasts using the MSI metric demonstrates that this metric provides more informative and more realistic estimates of the intensity forecast errors. It is also shown that the maximum speed metric allows for error compensation between the low-wavenumber and residual intensities, which could lead to forecast skill overestimation and inaccurate assessment of the impact of forecast system change on the skill.

scholarly journals Multidecadal Variability in North Atlantic Tropical Cyclone Activity

2008 ◽  
Vol 21 (15) ◽  
pp. 3929-3935 ◽  
Author(s):  
Philip J. Klotzbach ◽  
William M. Gray
Keyword(s):  
Tropical Cyclone ◽  
North Atlantic ◽  
Tropical Cyclone Activity ◽  
Cyclone Activity ◽  
Sea Level Pressure ◽  
Atlantic Basin ◽  
Multidecadal Variability ◽  
The North ◽  
Sea Surface Temperature Anomalies ◽  
Using Data

Abstract Recent increases in Atlantic basin tropical cyclone activity since 1995 and the associated destructive U.S. landfall events in 2004 and 2005 have generated considerable interest into why there has been such a sharp upturn. Natural variability, human-induced global warming, or a combination of both factors, have been suggested. Several previous studies have discussed observed multidecadal variability in the North Atlantic over 25–40-yr time scales. This study, using data from 1878 to the present, creates a metric based on far North Atlantic sea surface temperature anomalies and basinwide North Atlantic sea level pressure anomalies that shows remarkable agreement with observed multidecadal variability in both Atlantic basin tropical cyclone activity and in U.S. landfall frequency.

scholarly journals Recent Progress in Tropical Cyclone Intensity Forecasting at the National Hurricane Center

2020 ◽  
Vol 35 (5) ◽  
pp. 1913-1922 ◽  
Author(s):  
John P. Cangialosi ◽  
Eric Blake ◽  
Mark DeMaria ◽  
Andrew Penny ◽  
Andrew Latto ◽  
...  
Keyword(s):  
Tropical Cyclone ◽  
Forecast Error ◽  
Forecast Errors ◽  
Atlantic Basin ◽  
Cyclone Intensity ◽  
The Past ◽  
Intensity Forecast ◽  
Past Half Century ◽  
National Hurricane Center ◽  
Past Half

AbstractIt has been well documented that the National Hurricane Center (NHC) has made significant improvements in Atlantic basin tropical cyclone (TC) track forecasting during the past half century. In contrast, NHC’s TC intensity forecast errors changed little from the 1970s to the early 2000s. Recently, however, there has been a notable decrease in TC intensity forecast error and an increase in intensity forecast skill. This study documents these trends and discusses the advancements in TC intensity guidance that have led to the improvements in NHC’s intensity forecasts in the Atlantic basin. We conclude with a brief projection of future capabilities.

scholarly journals Deep learning for bias correction of MJO prediction

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
H. Kim ◽  
Y. G. Ham ◽  
Y. S. Joo ◽  
S. W. Son
Keyword(s):  
Deep Learning ◽  
Bias Correction ◽  
Numerical Models ◽  
Weather Prediction ◽  
Primary Source ◽  
Correction Method ◽  
Prediction Skill ◽  
Convective System ◽  
Forecast Errors ◽  
Potential Predictability

AbstractProducing accurate weather prediction beyond two weeks is an urgent challenge due to its ever-increasing socioeconomic value. The Madden-Julian Oscillation (MJO), a planetary-scale tropical convective system, serves as a primary source of global subseasonal (i.e., targeting three to four weeks) predictability. During the past decades, operational forecasting systems have improved substantially, while the MJO prediction skill has not yet reached its potential predictability, partly due to the systematic errors caused by imperfect numerical models. Here, to improve the MJO prediction skill, we blend the state-of-the-art dynamical forecasts and observations with a Deep Learning bias correction method. With Deep Learning bias correction, multi-model forecast errors in MJO amplitude and phase averaged over four weeks are significantly reduced by about 90% and 77%, respectively. Most models show the greatest improvement for MJO events starting from the Indian Ocean and crossing the Maritime Continent.

scholarly journals Feasibility of Continual Deep Learning-Based Segmentation for Personalized Adaptive Radiation Therapy in Head and Neck Area

Cancers ◽  
2021 ◽  
Vol 13 (4) ◽  
pp. 702
Author(s):  
Nalee Kim ◽  
Jaehee Chun ◽  
Jee Suk Chang ◽  
Chang Geol Lee ◽  
Ki Chang Keum ◽  
...  
Keyword(s):  
Deep Learning ◽  
Head And Neck ◽  
Organs At Risk ◽  
Subjective Assessment ◽  
Turing Test ◽  
Similar Rate ◽  
Dice Similarity Coefficient ◽  
Adaptive Planning ◽  
Test Set ◽  
Using Data

This study investigated the feasibility of deep learning-based segmentation (DLS) and continual training for adaptive radiotherapy (RT) of head and neck (H&N) cancer. One-hundred patients treated with definitive RT were included. Based on 23 organs-at-risk (OARs) manually segmented in initial planning computed tomography (CT), modified FC-DenseNet was trained for DLS: (i) using data obtained from 60 patients, with 20 matched patients in the test set (DLSm); (ii) using data obtained from 60 identical patients with 20 unmatched patients in the test set (DLSu). Manually contoured OARs in adaptive planning CT for independent 20 patients were provided as test sets. Deformable image registration (DIR) was also performed. All 23 OARs were compared using quantitative measurements, and nine OARs were also evaluated via subjective assessment from 26 observers using the Turing test. DLSm achieved better performance than both DLSu and DIR (mean Dice similarity coefficient; 0.83 vs. 0.80 vs. 0.70), mainly for glandular structures, whose volume significantly reduced during RT. Based on subjective measurements, DLS is often perceived as a human (49.2%). Furthermore, DLSm is preferred over DLSu (67.2%) and DIR (96.7%), with a similar rate of required revision to that of manual segmentation (28.0% vs. 29.7%). In conclusion, DLS was effective and preferred over DIR. Additionally, continual DLS training is required for an effective optimization and robustness in personalized adaptive RT.

scholarly journals Deep learning for cephalometric landmark detection: systematic review and meta-analysis

2021 ◽  
Author(s):  
Falk Schwendicke ◽  
Akhilanand Chaurasia ◽  
Lubaina Arsiwala ◽  
Jae-Hong Lee ◽  
Karim Elhennawy ◽  
...  
Keyword(s):  
Systematic Review ◽  
Deep Learning ◽  
Meta Analysis ◽  
High Accuracy ◽  
Risk Of Bias ◽  
Automated Detection ◽  
Reference Test ◽  
Landmark Detection ◽  
Future Studies ◽  
Using Data

Abstract Objectives Deep learning (DL) has been increasingly employed for automated landmark detection, e.g., for cephalometric purposes. We performed a systematic review and meta-analysis to assess the accuracy and underlying evidence for DL for cephalometric landmark detection on 2-D and 3-D radiographs. Methods Diagnostic accuracy studies published in 2015-2020 in Medline/Embase/IEEE/arXiv and employing DL for cephalometric landmark detection were identified and extracted by two independent reviewers. Random-effects meta-analysis, subgroup, and meta-regression were performed, and study quality was assessed using QUADAS-2. The review was registered (PROSPERO no. 227498). Data From 321 identified records, 19 studies (published 2017–2020), all employing convolutional neural networks, mainly on 2-D lateral radiographs (n=15), using data from publicly available datasets (n=12) and testing the detection of a mean of 30 (SD: 25; range.: 7–93) landmarks, were included. The reference test was established by two experts (n=11), 1 expert (n=4), 3 experts (n=3), and a set of annotators (n=1). Risk of bias was high, and applicability concerns were detected for most studies, mainly regarding the data selection and reference test conduct. Landmark prediction error centered around a 2-mm error threshold (mean; 95% confidence interval: (–0.581; 95 CI: –1.264 to 0.102 mm)). The proportion of landmarks detected within this 2-mm threshold was 0.799 (0.770 to 0.824). Conclusions DL shows relatively high accuracy for detecting landmarks on cephalometric imagery. The overall body of evidence is consistent but suffers from high risk of bias. Demonstrating robustness and generalizability of DL for landmark detection is needed. Clinical significance Existing DL models show consistent and largely high accuracy for automated detection of cephalometric landmarks. The majority of studies so far focused on 2-D imagery; data on 3-D imagery are sparse, but promising. Future studies should focus on demonstrating generalizability, robustness, and clinical usefulness of DL for this objective.

scholarly journals A Precise Benchmark for Cluster Scaling Relations: Fundamental Plane, Mass Plane and IMF in the Coma Cluster from Dynamical Models

2020 ◽  
Author(s):  
Shravan Shetty ◽  
Michele Cappellari ◽  
Richard M McDermid ◽  
Davor Krajnović ◽  
P T de Zeeuw ◽  
...  
Keyword(s):  
Galaxy Formation ◽  
Scaling Relations ◽  
Coma Cluster ◽  
Dynamical Models ◽  
Fundamental Plane ◽  
Cluster Galaxies ◽  
Dynamical Scaling ◽  
Small Uncertainty ◽  
Cluster Distance ◽  
Using Data

Abstract We study a sample of 148 early-type galaxies in the Coma cluster using SDSS photometry and spectra, and calibrate our results using detailed dynamical models for a subset of these galaxies, to create a precise benchmark for dynamical scaling relations in high-density environments. For these galaxies, we successfully measured global galaxy properties, modeled stellar populations, and created dynamical models, and support the results using detailed dynamical models of 16 galaxies, including the two most massive cluster galaxies, using data taken with the SAURON IFU. By design, the study provides minimal scatter in derived scaling relations due to the small uncertainty in the relative distances of galaxies compared to the cluster distance. Our results demonstrate low (≤55% for 90th percentile) dark matter fractions in the inner 1Re of galaxies. Owing to the study design, we produce the tightest, to our knowledge, IMF-σe relation of galaxies, with a slope consistent with that seen in local galaxies. Leveraging our dynamical models, we transform the classical Fundamental Plane of the galaxies to the Mass Plane. We find that the coefficients of the mass plane are close to predictions from the virial theorem, and have significantly lower scatter compared to the Fundamental plane. We show that Coma galaxies occupy similar locations in the (M* - Re) and (M* - σe) relations as local field galaxies but are older. This, and the fact we find only three slow rotators in the cluster, is consistent with the scenario of hierarchical galaxy formation and expectations of the kinematic morphology-density relation.

Prediction of August Atlantic Basin Hurricane Activity

10.1175/814.1 ◽  
2004 ◽  
Vol 19 (6) ◽  
pp. 1044-1060 ◽  
Author(s):  
Eric S. Blake ◽  
William M. Gray
Keyword(s):  
Tropical Cyclone ◽  
Subtropical High ◽  
Northwest Pacific ◽  
Seasonal Forecasts ◽  
Reanalysis Dataset ◽  
Atlantic Basin ◽  
The North ◽  
Hurricane Activity ◽  
Environmental Prediction ◽  
The Bering Sea

Abstract Although skillful seasonal hurricane forecasts for the Atlantic basin are now a reality, large gaps remain in our understanding of observed variations in the distribution of activity within the hurricane season. The month of August roughly spans the first third of the climatologically most active part of the season, but activity during the month is quite variable. This paper reports on an initial investigation into forecasting year-to-year variability of August tropical cyclone (TC) activity using the National Centers for Environmental Prediction–National Center for Atmospheric Research reanalysis dataset. It is shown that 55%–75% of the variance of August TC activity can be hindcast using a combination of 4–5 global predictors chosen from a 12-predictor pool with each of the predictors showing precursor associations with TC activity. The most prominent predictive signal is the equatorial July 200-mb wind off the west coast of South America. When this wind is anomalously strong from the northeast during July, Atlantic TC activity in August is almost always enhanced. Other July conditions associated with active Augusts include a weak subtropical high in the North Atlantic, an enhanced subtropical high in the northwest Pacific, and low pressure in the Bering Sea region. The most important application of the August-only forecast is that predicted net tropical cyclone (NTC) activity in August has a significant relationship with the incidence of U.S. August TC landfall events. Better understanding of August-only TC variability will allow for a more complete perspective of total seasonal variability and, as such, assist in making better seasonal forecasts.

Sign in / Sign up

Export Citation Format

Share Document