Tropical Cyclone Intensity Estimation Research Articles

Interagency discrepancies in tropical cyclone intensity estimates over the western North Pacific in recent years

AbstractThis study investigates interagency discrepancies among best‐track estimates of tropical cyclone (TC) intensity in the western North Pacific, provided by the Joint Typhoon Warning Center (JTWC), the China Meteorological Administration (CMA), and the Regional Specialized Meteorological Center (RSMC) Tokyo during 2013 to 2019. The results reveal evident differences in maximum wind speed (MSW) estimates, where linear systematic differences are significant. However, the Dvorak parameter (CI) numbers derived from the MSWs reported by the three agencies are internally consistent. Further analysis suggests that the remained CI discrepancies are related to differences in the estimation of intensity trends, initial intensities, and TC positions among these datasets. In addition, the CI estimates provided by the JTWC for TCs over the open ocean are generally higher than those reported by the CMA and RSMC. However, the estimates from CMA and RSMC tend to give higher TC intensities for the TCs in the mainland and coastal areas of China and Japan, respectively, than those over the open ocean with the same intensity in JTWC dataset. This pattern potentially reflects the extensive use of surface observations by these two agencies for landfalling and offshore TCs. These results may help the research community to get more accurate details about the TCs in WNP from the best track datasets of different agencies.

Characterization of Tropical Cyclone Intensity Using the HY-2B Scatterometer Wind Data

The estimation of tropical cyclone (TC) intensity using Ku-band scatterometer data is challenging due to rain perturbation and signal saturation in the radar backscatter measurements. In this paper, an alternative approach to directly taking the maximum scatterometer-derived wind speed is proposed to assess the TC intensity. First, the TC center location is identified based on the unique characteristics of wind stress divergence/curl near the TC core. Then the radial extent of 17-m/s winds (i.e., R17) is calculated using the wind field data from the Haiyang-2B (HY-2B) scatterometer (HSCAT). The feasibility of HSCAT wind radii in determining TC intensity is evaluated using the maximum sustained wind speed (MSW) in the China Meteorological Administration best-track database. It shows that the HSCAT R17 value generally better correlates with the best-track MSW than the HSCAT maximum wind speed, therefore indicating the potential of using the HSCAT data to improve the TC nowcasting capabilities.

Tropical Cyclone Intensity Estimation Using Himawari-8 Satellite Cloud Products and Deep Learning

This study develops an objective deep-learning-based model for tropical cyclone (TC) intensity estimation. The model’s basic structure is a convolutional neural network (CNN), which is a widely used technology in computer vision tasks. In order to optimize the model’s structure and to improve the feature extraction ability, both residual learning and attention mechanisms are embedded into the model. Five cloud products, including cloud optical thickness, cloud top temperature, cloud top height, cloud effective radius, and cloud type, which are level-2 products from the geostationary satellite Himawari-8, are used as the model training inputs. We sampled the cloud products under the 13 rotational angles of each TC to augment the training dataset. For the independent test data, the model shows improvement, with a relatively low RMSE of 4.06 m/s and a mean absolute error (MAE) of 3.23 m/s, which are comparable to the results seen in previous studies. Various cloud organization patterns, storm whirling patterns, and TC structures from the feature maps are presented to interpret the model training process. An analysis of the overestimated bias and underestimated bias shows that the model’s performance is highly affected by the initial cloud products. Moreover, several controlled experiments using other deep learning architectures demonstrate that our designed model is conducive to estimating TC intensity, thus providing insight into the forecasting of other TC metrics.

Estimation of Tropical Cyclone Intensity Using Synthetic Satellite Microwave Temperature Anomaly Structure and a Multifeature Distribution Learning Network

Estimation of Tropical Cyclone Intensity Using Synthetic Satellite Microwave Temperature Anomaly Structure and a Multifeature Distribution Learning Network

Tropical Cyclone Intensity Estimation Using Multidimensional Convolutional Neural Network From Multichannel Satellite Imagery

Tropical Cyclone Intensity Estimation Using Multidimensional Convolutional Neural Network From Multichannel Satellite Imagery

A Conformal Regressor With Random Forests for Tropical Cyclone Intensity Estimation

A Conformal Regressor With Random Forests for Tropical Cyclone Intensity Estimation

Tropical Cyclone Intensity Estimation From Spaceborne Microwave Scatterometry and Parametric Wind Models

Spaceborne microwave sensors, measuring co-/cross-polarization (VV/VH) normalized radar cross section signals, have been widely used for tropical cyclone (TC) monitoring. However, considerable gaps remain to obtain TC intensity since these satellite data either blur inner-core structures (e.g., scatterometer data) or have limited spatial–temporal coverage (e.g., synthetic aperture radar (SAR) data). This study aims to get more accurate TC intensity estimates from scatterometers, which have good global coverage but relatively low spatial resolution. To overcome the blurring effect in scatterometers, we propose a new technique for guidance on TC intensities, with maximum 1-min sustained winds calculated as a function of decay parameters provided by the parametric Rankine-type model. The technique is employed on advanced scatterometer (ASCAT) data acquired between 2016 and 2017, validated with simultaneous SAR VH geophysical model function measurements and best-track (BT) estimates. When validated with BT estimates, the method enhances the blurred maximum winds, where the standard deviation of difference decreased from 6.3 to 3.49 m/s and the coefficient of determination increased from 0.7 to 0.89. Besides, it is noteworthy that the proposed technique performs slightly better than the Mayers–Ruf method. The promising results indicate that the technique can provide more representative TC maximum 1-min sustained wind estimates from ASCAT data, thus contributing to the further exploitation of scatterometer data for TC warnings.

An effective tropical cyclone intensity estimation model using Convolutional Neural Networks

An effective tropical cyclone intensity estimation model using Convolutional Neural Networks

Real-time Tropical Cyclone Intensity Estimation by Handling Temporally Heterogeneous Satellite Data

Analyzing big geophysical observational data collected by multiple advanced sensors on various satellite platforms promotes our understanding of the geophysical system. For instance, convolutional neural networks (CNN) have achieved great success in estimating tropical cyclone (TC) intensity based on satellite data with fixed temporal frequency (e.g., ~3 h). However, to achieve more timely (under 30 min) and accurate TC intensity estimates, a deep learning model is demanded to handle temporally-heterogeneous satellite observations. Specifically, infrared (IR1) and water vapor (WV) images are available within every 15 minute period, while passive microwave rain rate (PMW) is available about every 3 hours. Meanwhile, the visible (VIS) channel is severely affected by noise and sunlight intensity, making it difficult to be utilized. Therefore, we propose a novel framework that combines generative adversarial network (GAN) with CNN. The model utilizes all data during the training phase including VIS and PMW information and eventually uses only the high-frequent IR1 and WV data for providing intensity estimates during the predicting phase. Experimental results demonstrate that the hybrid GAN-CNN framework achieves comparable precision to the state-of-the-art models, while possessing the capability of increasing the maximum estimation frequency from 3 hours to less than 15 minutes. Please visit https://github.com/BoyoChen/CNN-GAN-TC for codes and implementation details.

Physics-augmented Deep Learning to Improve Tropical Cyclone Intensity and Size Estimation from Satellite Imagery

AbstractA deep learning-based method augmented by prior knowledge of tropical cyclones (TCs), called DeepTCNet, is introduced to estimate TC intensity and wind radii from infrared (IR) imagery over the North Atlantic Ocean. While standard deep learning practices have many advantages over conventional analysis approaches and can produce reliable estimates of TCs, the data-driven models informed by machine-readable physical knowledge of TCs could achieve higher performance. To this end, two approaches are explored to develop the physics-augmented DeepTCNet: (i) infusing the auxiliary physical information of TCs into models for single-task learning; (ii) learning auxiliary physical tasks for multi-task learning. More specifically, augmented by auxiliary information of TC fullness (a measure of the radial decay of the TC wind field), the DeepTCNet yielded a 12% improvement in estimating TC intensity over the non-augmented one. By learning TC wind radii and auxiliary TC intensity task simultaneously, the model’s wind radii estimation skill is improved by 6% over only learning four wind radii tasks, and by 9% over separately learning a single wind radii task. The evaluation results showed that the DeepTCNet is in-line with the Satellite Consensus technique (SATCON) but systematically outperforms the Advanced Dvorak Technique (ADT) at all intensity scales with an averaged 39% enhancement in TC intensity estimation. The DeepTCNet also surpasses the Multi-platform Tropical Cyclone Surface Wind Analysis technique (MTCSWA) with an average improvement of 32% in wind radii estimation.

Tropical Cyclone Intensity Classification and Estimation Using Infrared Satellite Images With Deep Learning

A novel tropical cyclone (TC) intensity classification and estimation model (TCICENet) is proposed using infrared geostationary satellite images from the northwest Pacific Ocean basin in combination with a cascading deep convolutional neural network (CNN). The proposed model consists of two CNN network modules: a TC intensity classification (TCIC) module and a TC intensity estimation (TCIE) module. First, the TCIC module is utilized to divide TC intensity into three categories using infrared satellite images. Next, three TCIE models based on the CNN regression network that combine different intensity types of infrared satellite images with the TC best track data are presented. The three TCIE models consider classification error with the TCIC module in order to improve TCIE accuracy. A total of 1001 TCs from 1981-2019 were used to verify the proposed TCICENet model, with 844 TCs from 1981-2013 employed as training samples, 76 TCs from 2014-2016 used as validation samples, and 81 TCs from 2017-2019 used as testing samples. In order to reduce the computation burden of training the TCICENet model, various input image sizes were explored. An image size of 170 × 170 pixels achieved the best performance, with an overall root mean square error of 8.60 kt and a mean absolute error of 6.67 kt compared to the best track.

Multi-Factor Intensity Estimation for Tropical Cyclones in the Western North Pacific Based on the Deviation Angle Variance Technique

In this paper, the infrared cloud images from Fengyun series geostationary satellites and the best track data from the China Meteorological Administration (CMA-BST) in 2015–2017 are used to investigate the effects of two multi-factor models, generalized linear model (GLM) and long short-term memory (LSTM) model, for tropical cyclone (TC) intensity estimation based on the deviation angle variance (DAV) technique. For comparison, the typical single-factor Sigmoid function model (SFM) with the map minimum value of DAV is also used to produce TC intensity estimation. Sensitivity experiments regarding the DAV calculation radius and different training data groups are conducted, and the estimation precision and optimum calculation radius for DAV in the western North Pacific (WNP) are analyzed. The results show that the root-mean-square-error (RMSE) of the single-factor SFM is 8.79–13.91 m s−1 by using the individual years as test sets and the remaining two years as training sets with the optimum calculation radius of 550 km. However, after selecting and using the high-correlation multiple factors from the same test and training data, the RMSEs of GLM and LSTM models decrease to 5.93–8.68 and 4.99–7.00 m s−1 respectively, with their own optimum calculation radii of 350 and 400 km. All the sensitivity experiments indicate that the SFM results are significantly influenced by the DAV calculation radius and characteristics of the training set data, while the results of multi-factor models appear more stable. Furthermore, the multi-factor models reduce the optimum radius within the process of DAV calculation and improve the precision of TC intensity estimation in the WNP, which can be chosen as an effective approach for TC intensity estimation in marine areas.

A Novel Tensor Network for Tropical Cyclone Intensity Estimation

Tropical cyclone (TC) intensity estimation is an important task in meteorological research. Meanwhile, TC intensity estimation performance can be improved by developing advanced machine learning techniques using the newly emerged high-quality multispectral images (MSIs) acquired by FY-4 meteorological satellite of China. To this end, this article proposes a novel model, tensor-based convolutional neural network (TCNN). Not only being a deep network entirely formulated in tensor algebra, but TCNN also establishes the mathematical connections between tensor decomposition and CNN operations with tensor contraction. Moreover, TCNN adopts a multitask structure, which consists of a classification network for intensity categorization and a regression network for wind speed estimation. It allows the regression network to leverage the outcome of the classification network, thus ensuring intensity estimation accuracy. In addition, two alternative TCNN models, coupled TCNN (C-TCNN) and Tucker TCNN (T-TCNN), are designed to automatically deal with invalid band data in the FY-4 MSIs, which has been a practical issue with the data. Experimental results prove that the proposed network is accurate and functional. They also demonstrate the advantages of TCNN over other TC intensity estimation methods, showing that C-TCNN and T-TCNN outperform several classic models and the state-of-the-art models based on CNN.

Deep learning estimation of tropical cyclone intensity along the southeast coast of China using FY-4A satellite

Deep learning estimation of tropical cyclone intensity along the southeast coast of China using FY-4A satellite

A CNN-Based Hybrid Model for Tropical Cyclone Intensity Estimation in Meteorological Industry

Accurate estimation of tropical cyclone (TC) intensity is the key to understanding and forecasting the behavior of TC and is crucial for initialization in forecast models and disaster management in the meteorological industry. TC intensity estimation is a challenge because it requires domain knowledge to manually extract TC cloud structure features and form various sets of parameters obtained from satellites. In this paper, a novel hybrid model is proposed based on convolutional neural networks (CNNs) for TC intensity estimation with satellite remote sensing. According to the intensity of TCs, we divide them into three types and use three different models for intensity regression, respectively. The results show that the use of piecewise thinking can improve the model's fitting speed on small samples. A classification network is provided to classify unlabeled TC samples before TC regression, whose results would determine which regression network to estimate these samples. Finally, the estimation values are sent to the backpropagation (BP) neural network to fit the suitable intensity values. Experimental results demonstrate that our model achieves high accuracy and low root-mean-square error (RMSE up to 8.91 kts) by just using inferred images.

Tropical Cyclone Intensity Estimation Using Multi-Dimensional Convolutional Neural Networks from Geostationary Satellite Data

For a long time, researchers have tried to find a way to analyze tropical cyclone (TC) intensity in real-time. Since there is no standardized method for estimating TC intensity and the most widely used method is a manual algorithm using satellite-based cloud images, there is a bias that varies depending on the TC center and shape. In this study, we adopted convolutional neural networks (CNNs) which are part of a state-of-art approach that analyzes image patterns to estimate TC intensity by mimicking human cloud pattern recognition. Both two dimensional-CNN (2D-CNN) and three-dimensional-CNN (3D-CNN) were used to analyze the relationship between multi-spectral geostationary satellite images and TC intensity. Our best-optimized model produced a root mean squared error (RMSE) of 8.32 kts, resulting in better performance (~35%) than the existing model using the CNN-based approach with a single channel image. Moreover, we analyzed the characteristics of multi-spectral satellite-based TC images according to intensity using a heat map, which is one of the visualization means of CNNs. It shows that the stronger the intensity of the TC, the greater the influence of the TC center in the lower atmosphere. This is consistent with the results from the existing TC initialization method with numerical simulations based on dynamical TC models. Our study suggests the possibility that a deep learning approach can be used to interpret the behavior characteristics of TCs.

Tropical Cyclone Intensity Estimation Using Two-Branch Convolutional Neural Network From Infrared and Water Vapor Images

This article proposes a two-branch convolutional neural network model (TCIENet) to estimate the intensity of tropical cyclone (TC) from infrared and water vapor images in the northwest Pacific basin. Three different sizes of input images are explored to train the TCIENet model, and the size of $60\times 60$ pixels (radius 450 km) achieves the best performance with an overall root mean square error (RMSE) of 5.13 m/s and mean absolute error (MAE) of 4.03 m/s. TCs are divided into six categories whose RMSEs range from 4.07 to 6.05 m/s. In addition, the TCs in the year 2017 are used to analyze the correlation between the rainfall intensity from the global precipitation measurement (GPM) mission and the estimation errors of the TCIENet model. Preliminary results suggest that the model performs the best at the categories of tropical storm and super typhoon, but it degrades in performance for moderate intense categories and the weakest category of the tropical depression. The correlation coefficient between the estimation error and the rainfall intensity is 0.19. It is far from certain that the rainfall intensity accounts for the error achieved by the TCIENet model.

An Index to Better Estimate Tropical Cyclone Intensity Change in the Western North Pacific

AbstractA revised predictor called the net energy gain rate (NGR) is suggested by considering wind‐dependent drag coefficient based on the existing maximum potential intensity theory. A series of wind speed‐dependent NGR, known as NGR‐w, is calculated based on pretropical cyclone (TC) averaged ocean temperatures from the surface down to 120 m (at 10‐m intervals) to include the TC‐induced vertical mixing for 13 years (2004–2016) in the western North Pacific. It turns out that NGR50‐w (NGR‐w based on temperature averaged over top 50 m) has the highest correlation with 24‐hr TC intensity change compared with the commonly used sea surface temperature‐based intensification potential (POT), depth‐averaged temperature‐based POT (POTDAT), and constant drag coefficient in the NGR. To demonstrate the effectiveness of NGR50‐w, we designed and conducted experiments for training (2004–2014) and testing (2015–2016). The model with NGR50‐w shows greater skill than does the model with POTDAT or POT by reducing prediction errors by about 16%.

Observed Characteristics Change of Tropical Cyclones During Rapid Intensification Over Western North Pacific Using CloudSat Data

Rapid intensification (RI) progress is the main challenge that precludes the improvement of tropical cyclone (TC) intensity estimates. In this paper, a composite study of the observable characteristics of 11-year TCs undergoing RI and non-RI in the Western North Pacific were conducted using CloudSat tropical cyclone (CSTC) dataset. 2B-GEOPROF, 2B-CLDCLASS, and 2B-CWC-RO products in the CSTC dataset were used to construct radar reflectivity, cloud class, and cloud water-content distributions for each category, respectively. The results show that radar echo statistics have an arc-like contour profile for all five categories, with two distinct modes of reflectivity distributions separated by the melting layer. RI has the highest frequency expanding through the whole reflectivity range along with the broadest coverage in the upper branch, and a “continuity” of reflectivity distribution from −10 to 10 dBZ at 6–11 km could be a vital indicator of TC intensification. A sharp slope can be observed in the lower branch due to heavy precipitation attenuation. Vertical distributions of cloud types show that all categories have similar cloud compositions, and deep convective clouds are apparently to play an important role in maintaining the TC intensification. Liquid water content (LWC) curve exhibits a bimodal distribution with two peaks at around 1.5 and 5 km, while ice water content (IWC) curve is much smoother and has one peak at around 9 km. It is suggested that TCs with larger LWC around freezing level or/and larger IWC at upper level tend to intensify within the next 24 hours.

Using Deep Learning to Estimate Tropical Cyclone Intensity from Satellite Passive Microwave Imagery

AbstractA deep learning convolutional neural network model is used to explore the possibilities of estimating tropical cyclone (TC) intensity from satellite images in the 37- and 85–92-GHz bands. The model, called “DeepMicroNet,” has unique properties such as a probabilistic output, the ability to operate from partial scans, and resiliency to imprecise TC center fixes. The 85–92-GHz band is the more influential data source in the model, with 37 GHz adding a marginal benefit. Training the model on global best track intensities produces model estimates precise enough to replicate known best track intensity biases when compared to aircraft reconnaissance observations. Model root-mean-square error (RMSE) is 14.3 kt (1 kt ≈ 0.5144 m s−1) compared to two years of independent best track records, but this improves to an RMSE of 10.6 kt when compared to the higher-standard aircraft reconnaissance-aided best track dataset, and to 9.6 kt compared to the reconnaissance-aided best track when using the higher-resolution TRMM TMI andAquaAMSR-E microwave observations only. A shortage of training and independent testing data for category 5 TCs leaves the results at this intensity range inconclusive. Based on this initial study, the application of deep learning to TC intensity analysis holds tremendous promise for further development with more advanced methodologies and expanded training datasets.