Operational Application of Optical Flow Techniques to Radar-Based Rainfall Nowcasting

Woo, Wang-chun; Wong, W.C.

doi:10.3390/atmos8030048

Cited by 155 publications

(139 citation statements)

References 22 publications

Supporting

Mentioning

139

Contrasting

Order By: Relevance

“…With convolutional filters replacing the matrix multiplications, ConvLSTM can preserve and exploit the spatial correlations while representing the temporal correlations simultaneously. Thanks to that, a sequence-to-sequence model built from ConvLSTM cells outperforms the Real-time Optical flow by Variational methods for Echoes of Radar (ROVER) algorithm, which is considered as the state of the art optical flow based model [19].…”

Section: Convolutional Rnn-based Modelsmentioning

confidence: 99%

Computer Vision in Precipitation Nowcasting: Applying Image Quality Assessment Metrics for Training Deep Neural Networks

Tran

Song

2019

Atmosphere

View full text Add to dashboard Cite

This paper presents a viewpoint from computer vision to the radar echo extrapolation task in the precipitation nowcasting domain. Inspired by the success of some convolutional recurrent neural network models in this domain, including convolutional LSTM, convolutional GRU and trajectory GRU, we designed a new sequence-to-sequence neural network structure to leverage these models in a realistic data context. In this design, we decreased the numbers of channels in high abstract recurrent layers rather than increasing them. We formulated the task as a problem of encoding five radar images and predicting 10 steps ahead at the pixel level, and found that using only the common mean squared error can misguide the training and mislead the testing. Especially, the image quality of last predictions usually degraded rapidly. As a solution, we employed some visual image quality assessment techniques including Structural Similarity (SSIM) and multi-scale SSIM to train our models. Experimental results show that our structure was more tolerant to increasing uncertainty in the data, and the use of image quality metrics can significantly reduce the blurry image issue. Moreover, we found that using SSIM was very effective and a combination of SSIM with mean squared error and mean absolute error yielded the best prediction quality.

show abstract

Section: Convolutional Rnn-based Modelsmentioning

confidence: 99%

Computer Vision in Precipitation Nowcasting: Applying Image Quality Assessment Metrics for Training Deep Neural Networks

Tran

Song

2019

Atmosphere

View full text Add to dashboard Cite

show abstract

“…Weather radars and satellite images are important tools for monitoring severe weather such as thunderstorms, squall lines, and rainstorms in the vicinity of Hong Kong as well as the movement and structure of tropical cyclones. The real-time weather radar information is also utilized together with data of rain-gauge networks in nowcasting of rainstorms and support of relevant warning services (Woo and Wong 2017). In particular, the Hong Kong International Airport (HKIA) is among the first airports to implement LIDAR systems for the detection of low-level windshear for safeguarding aviation safety (Shun and Chan 2008).…”

Section: (D) Remote-sensing Measurementsmentioning

confidence: 99%

“…In addition to the NWP products from major global NWP models, by adopting JMA-NHM (Saito et al 2006), HKO developed a rapid update non-hydrostatic modeling system for mesoscale weather forecasting with horizontal resolutions of 10 km initialized by 3-hourly analysis, and of 2 km using hourly update cycle incorporating weather radar data for convective-allowing simulations (Wong 2011). Moreover, HKO's Aviation Model (AVM; Wong et al 2013;Chan and Hon 2016), which is based on WRF-ARW system (Skamarock et al 2008), providing hourly updated model forecasts at horizontal resolutions of 600 and 200 m. Regarding quantitative precipitation estimation (QPE) and quantitative precipitation forecast (QPF) for the next few hours, an in-house developed rainstorm nowcasting system called Short-range Warnings of Intense Rainstorm of Localized Systems (SWIRLS) has been operating to support rainstorm-related high-impact weather services for public and specialized users (Li et al 2014;Woo and Wong 2017). Using post-processed outputs from state-of-the-art NWP models, location-specific weather forecast for the next 9 days has been made available to the public in recent years through the Automatic Regional Weather Forecast (ARWF) service.…”

Section: (A) Weather Forecast and Warningmentioning

confidence: 99%

Urban-focused weather and climate services in Hong Kong

2018

Self Cite

View full text Add to dashboard Cite

Under the relentless pace of population growth and urban development, weather and climate services provided by the Hong Kong Observatory (HKO) since its establishment in 1883 have been evolving continuously with an increasing focus on the delivery of tailored information to meet the needs of special users and various stakeholders in the community. With meteorological observations at the headquarters of the Observatory dating back more than 130 years to 1884, the advent of automatic weather station network in the mid-1980s has significantly expanded the data coverage both in terms of spatial and temporal resolution. The establishment of the Community Weather Information Network in 2007 further extended the meteorological measurement to the school and community levels. Regular upper air soundings and the increasing availability of remote-sensing observations also enable the monitoring of meteorological parameters beyond the near-surface levels through the depth of the atmosphere. By integrating comprehensive weather observations and numerical weather prediction products, HKO now provides a wide range of forecasts covering multi-time scales and warnings and advisories for high-impact weather such as tropical cyclone, thunderstorm, heavy rain, landslide, flooding, and cold and very hot weather. Riding on the advances in communication technology in recent decades, members of the public have gained improved access to the latest weather warnings, advisories, weather information, and forecast for the next couple of hours to 9 days ahead, as well as forecasts near users' locations through HKO's website and mobile app. In addition to information for public consumption, HKO's services and products have been tailored in support of weather-sensitive operations such as those provided to users in the aviation and marine communities. In the process of such diversified development, HKO has successfully cultivated close partnerships with various stakeholders to enhance its weather and climate services by embracing the spirit of the Big Data and smart city concept in recent years, in particular in areas related to energy, water, health, and disaster risk reduction. In this paper, a review of the weather and climate services focusing on the urban aspects in Hong Kong will be presented, with examples illustrating specific applications in different sectors. Future thrusts in enhancing and integrating the urban weather and climate services for Hong Kong in the Big Data era under the Climate Friendly Demonstration City initiative and impact-based forecast approach promulgated by the World Meteorological Organization will also be discussed.

show abstract

“…Traditional extrapolation methods, including the centroid tracking (Dixon and Wiener, 1993), cross-correlation (Rinehart and Garvey, 1978) and optical flow (Woo and Wong, 2017), mainly rely on extrapolating the echo linearly with the calculated motion vectors, where the motion vectors for centroid tracking are cell-wise and for the other two are regionwise. However, their extrapolation capacity and application ability are both limited considering the following two issues.…”

Section: Introductionmentioning

confidence: 99%

Aenn: A Generative Adversarial Neural Network for Weather Radar Echo Extrapolation

Jing

Ding

et al. 2019

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. Weather radar echo is one of the fundamental data for meteorological workers to weather systems identification and classification. Through the technique of weather radar echo extrapolation, the future short-term weather conditions can be predicted and severe convection storms can be warned. However, traditional extrapolation methods cannot offer accurate enough extrapolation results since their modeling capacity is limited, the recent deep learning based methods make some progress but still remains a problem of blurry prediction when making deeper extrapolation, which may due to they choose the mean square error as their loss function and that will lead to losing echo details. To address this problem and make a more realistic and accurate extrapolation, we propose a deep learning model called Adversarial Extrapolation Neural Network (AENN), which is a Generative Adversarial Network (GAN) structure and consist of a conditional generator and two discriminators, echo-frame discriminator and echo-sequence discriminator. The generator and discriminators are trained alternately in an adversarial way to make the final extrapolation results be realistic and accurate. To evaluate the model, we conduct experiments on extrapolating 0.5h, 1h, and 1.5h imminent future echoes, the results show that our proposed AENN can achieve the expected effect and outperforms other models significantly, which has a powerful potential application value for short-term weather forecasting.

show abstract

Operational Application of Optical Flow Techniques to Radar-Based Rainfall Nowcasting

Cited by 155 publications

References 22 publications

Computer Vision in Precipitation Nowcasting: Applying Image Quality Assessment Metrics for Training Deep Neural Networks

Computer Vision in Precipitation Nowcasting: Applying Image Quality Assessment Metrics for Training Deep Neural Networks

Urban-focused weather and climate services in Hong Kong

Aenn: A Generative Adversarial Neural Network for Weather Radar Echo Extrapolation

Contact Info

Product

Resources

About