Search

This paper proposes a new quantitative precipitation estimation (QPE) technique to provide accurate rainfall estimates in complex terrain, where conventional QPE has limitations. The operational radar QPE in Taiwan is mainly based on the simplified relationship between radar reflectivity Z and rain rate R [R(Z) relation] and only utilizes the single-point lowest available echo to estimate rain rates, leading to low accuracy in complex terrain. Here, we conduct QPE using deep learning that extracts features from 3D radar reflectivities to address the above issues. Convolutional neural networks (CNN) are used to analyze contoured frequency by altitude diagrams (CFADs) to generate the QPE. CNN models are trained on existing rain gauges in northern and eastern Taiwan with the 3-yr data during 2015–17 and validated and tested using 2018 data. The weights of heavy rains (≥10 mm h−1) are increased in the model loss calculation to handle the unbalanced rainfall data and improve accuracy. Results show that the CNN outperforms the R(Z) relation based on the 2018 rain gauge data. Furthermore, this research proposes methods to conduct 2D gridded QPE at every pixel by blending estimates from various trained CNN models. Verification based on independent rain gauges shows that the CNN QPE solves the underestimation of the R(Z) relation in mountainous areas. Case studies are presented to visualize the results, showing that the CNN QPE generates better small-scale rainfall features and more accurate precipitation information. This deep learning QPE technique may be helpful for the disaster prevention of small-scale flash floods in complex terrain. [ABSTRACT FROM AUTHOR]

The performance of a numerical weather prediction model using convective-scale ensemble data assimilation with ground-based global navigation satellite systems-zenith total delay (ZTD) and radar data is investigated on a heavy rainfall event that occurred in Taiwan on 10 June 2012. The assimilation of ZTD and/or radar data is performed using the framework of the WRF local ensemble transform Kalman filter with a model grid spacing of 2 km. Assimilating radar data is beneficial for predicting the rainfall intensity of this local event but produces overprediction in southern Taiwan and underprediction in central Taiwan during the first 3 h. Both errors are largely overcome by assimilating ZTD data to improve mesoconvective-scale moisture analyses. Consequently, assimilating both the ZTD and radar data show advantages in terms of the location and intensity of the heavy rainfall. Sensitivity experiments involving this event indicate that the impact of ZTD data is improved by using a broader horizontal localization scale than the convective scale used for radar data assimilation. This optimization is necessary in order to consider more fully the network density of the ZTD observations and the horizontal scale of the moisture transport by the southwesterly flow in this case. [ABSTRACT FROM AUTHOR]