Search

Severe rainfall has become increasingly frequent and intense in the Taipei metropolitan area. A complex thunderstorm in the Taipei Basin on 14 June 2015 produced an extreme rain rate (>130 mm h−1), leading to an urban flash flood. This paper presents storms’ microphysical and dynamic features during the organizing and heavy rain stages, mainly based on observed polarimetric variables in a Doppler radar network and ground-based raindrop size distribution. Shallower isolated cells in the early afternoon characterized by big raindrops produced a rain rate > 10 mm h−1, but the rain showers persisted for a short time. The storm’s evolution highlighted the behavior of merged convective cells before the heaviest rainfall (exceeding 60 mm within 20 min). The columnar features of differential reflectivity (ZDR) and specific differential phase (KDP) became more evident in merged cells, which correlated with the broad distribution of upward motion and mixed-phase hydrometeors. The KDP below the environmental 0°C level increased toward the ground associated with the melted graupel and resulted in subsequent intense rain rates, showing the contribution of the ice-phase process. Due to the collision–breakup process, the highest concentrations of almost all drop sizes and smaller mass-weighted mean diameter occurred during the maximum rainfall stage.

Using coastal radar and surface observations of Taiwan, an investigation of intensity and structure variations in the inner core of Typhoon Haitang (0505) is conducted. Within a 3-h period (1933–2233 UTC 17 July 2005), Haitang experienced intensity vacillation with merging of the eyewall with a rainband induced by the coastal barrier jet (CBJ); concentric eyewall breakdown and weakening; and eyewall recovery, contraction, and re-intensification. The northerly flow of the CBJ converged with the southerly flow of a leeside meso-low to form a west–east line of convection south of the storm when the storm was still 100 km offshore. The rainband propagated radially inward and triggered eyewall–rainband interaction. The interaction resulted in approximately 30% amplification of precipitation and 20% decrease in the axisymmetric tangential wind. Barotropic instability is speculated to be the underlying dynamic process. The recovery of the eyewall, following nearshore eyewall axisymmetrization and contraction, resulted in a 40% intensity increase before landfall. Significance Statement The behaviors and underlying physical processes of a landfalling tropical cyclone (TC) under the influence of complex terrain are studied by using coastal radars. The TC inner core structure and intensity change, including concentric eyewall breakdown, weakening, eyewall recovery, eyewall contraction, and re-intensification that occurred 3 h before TC made landfall are documented. The terrain-induced coastal barrier jet and leeside meso-low helped to form an intense line convection near the southern fringe of the eyewall and triggered the rainband–eyewall interaction.

This study investigates the impacts of different types of Cloud Condensation Nuclei (CCN) on the extreme of precipitation and updraft (above the 99th percentile threshold) of a mesoscale convection system occurred over the southwest coast of Taiwan Island under the moist southwesterly monsoon environment using cloudpermitting simulations. The convection system over the Taiwan Strait was observed during the SoWMEX/TiMREX IOP#8 in 2008. Simulations with the maritime type (clean) and the continental type (polluted) CCN are carried out and compared. The polluted simulation exhibits stronger extreme precipitation and updraft. The stronger extreme precipitation is associated with the enhanced size of snow and graupel. The enhanced latent heat release from faster diffusion growth of cloud drops contributes to stronger extreme updraft in the polluted simulation. With smaller rain evaporation rate, cold pool is weaker in the polluted simulation, which also leads to a more optimized convection structure and enhanced extreme updraft. The current results highlight the potential impacts of aerosol loadings on the extreme rainfall of monsoon coastal convection through microphysical, thermodynamics, and dynamical effects.

In this study, the structure and evolution of a coastal barrier jet (CBJ) along the east coast of Taiwan is documented using operational Doppler radars. The formation of the CBJ was controlled by the flow regime associated with the approaching Typhoon Haitang (2005). The CBJ persisted for 6 h and was approximately 140 km long and 25 km wide. The northern branch of the CBJ had stronger winds with maximum wind speed 49–52 m s−1, a greater vertical extent with jet core between 1.0 and 2.5 km in height, and a more persistent jet signal than the southern branch with maximum wind speed 43–46 m s−1 and jet core between 1.0 and 2.0 km. We investigated the terrain blocking effect leading to the CBJ formation using an idealized simulation. A vortex resembling Haitang is constructed based on circulation retrieved from generalized velocity track display (GVTD) technique. The result of a no-terrain simulation reveals wind speed 10–22 m s−1 lower than the observed Doppler velocity. The difference suggests the enhanced wind speed along the coast was most likely due to the terrain blocking effect.

Extreme weather events, such as typhoons, have occurred more frequently in the last few decades in the Philippines. The heavy precipitation caused by typhoons is difficult to measure with traditional instruments, such as rain gauges and ground-based radar, because these instruments have an uneven distribution in remote areas. Satellite precipitation datasets (SPDs) provide integrated spatial coverage of rainfall measurements, even for remote areas. This study performed subdaily (3-hour) assessments of SPDs (i.e., the Integrated Multi-satellitE Retrievals for Global Precipitation Measurement [IMERG], Global Satellite Mapping of Precipitation [GSMaP], and Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks datasets) during five typhoon-related heavy precipitation events in the Philippines between 2016 and 2018. The aforementioned assessments were performed through a point-to-grid comparison by using continuous and volumetric statistical validation indices for the 34-knot wind radii of the typhoons, rainfall intensity, the terrain, and wind velocity effects. The results revealed that the IMERG exhibited good agreement with rain gauge measurements and exhibited high performance in detecting rainfall during five typhoon events, whereas the GSMaP exhibited high agreement during peak rainfall. All the SPDs tended to overestimate rainfall during light to moderate rainfall events and underestimate rainfall during heavy to extreme events. The IMERG exhibited a strong ability to detect moderate rainfall events (5–15 mm/3 hours), whereas the GSMaP exhibited superior performance in detecting heavy to extreme rainfall events (15–25, 25–50, and >50 mm/3 hours). The GSMaP exhibited the best performance for detecting heavy rainfall at high elevations, whereas the IMERG exhibited the best performance for rainfall detection at low elevations. The IMERG exhibited a strong ability to detect heavy rainfall under various wind speeds. A strong ability to detect heavy rainfall events for different wind speeds in the western and eastern parts of the mountainous region of Luzon were found for the GSMap and IMERG, respectively. This study demonstrated that the IMERG and GSMaP datasets exhibit promising performance in detecting heavy precipitation caused by typhoon events.

Four types (2D-video disdrometer: 2DVD; precipitation occurrence sensor system: POSS; micro-rain radar: MRR; and Joss–Waldvogel disdrometer: JWD) of sixteen instruments were collocated within a square area of 400 m2 from 16 April to 8 May 2008 for intercomparison of drop size distribution (DSD) of rain. This unique dataset was used to study the inherent measurement uncertainty due to the diversity of the measuring principles and sampling sizes of the four types of instruments. The DSD intercomparison shows generally good agreement among them, except that the POSS and MRR had higher concentrations of small raindrops (5.2 mm). The measurement uncertainty ( σ ) was obtained quantitatively after considering the zero or non-zero measurement error covariance between two instruments of the same type. The results indicate the measurement uncertainties were found to be neither independent nor identical among the same type of instruments. The MRR is relatively accurate (lower σ ) due to large sampling volumes and accurate measurement of the Doppler power spectrum. The JWD is the least accurate due to the small sampling volumes. The σ decreases rapidly with increasing time-averaging window. The 2DVD shows the best accuracy of R in longer averaging time, but this is not true for Z due to the small sampling volume. The MRR outperformed other instruments for Z for entire averaging time due to its measuring principle.

Rainfall intensity-duration (I-D) characteristics are widely used to study landslide triggering. Recent studies also propose that rainfall kinetic energy (ek) can be used to quantify the rate of soil erosion induced by strong precipitation events. In the present study, the Joss-Waldvogel Disdrometers (JWD) was utilized to measure ek data, and the relationship between rainfall kinetic energy and rainfall intensity was determined as the regression results of ekN = 32.19 × (1–0.725e (− 0.029I)) for northern Taiwan and ekS = 32.23 × (1–0.643e(− 0.022I)) for southern Taiwan. Additionally, by examining data from 76 landslide events from the landslide inventory for the period of 2006–2012, the study established an empirical power law I-D relationship, which is crucial for determining the rainfall threshold needed in forecasting landslides. Two rainfall thresholds, i.e., functions of rainfall intensity (I) and rainfall duration (D), were established for study areas in northern and southern Taiwan to be IN = 13.81 D− 0.31 and IS = 66.44 D− 0.57 respectively. Landslides in northern Taiwan usually occurred when D was over 10 h and such events in southern Taiwan took place in two different D groups whose gap was between 16 h and 20 h. Two mechanisms were operative in triggering landslides in the two areas—water infiltration and high rainfall intensity. However, ek also proved to be an important rainfall parameter, and the present study further established that not only rainfall intensity-duration but also kinetic energy can provoke landslide triggering, thus allowing the compiling of more comprehensive information for more accurate landslide forecasting.

A long-lived mesoscale convective system (MCS) with extreme rainfall over the western coastal region of Guangdong on 10 May 2013 during the Southern China Monsoon Rainfall Experiment (SCMREX) is studied. The environmental conditions are characterized by little convective inhibition, low-lifting condensation level, moderate convective available potential energy and precipitable water, and lack of low-level jets from the tropical ocean. Repeated convective back building and subsequent northeastward “echo training” of convective cells are found during the MCS's development stages. However, the initiation/maintenance factors and organization of convection differ significantly during the earlier and later stages. From midnight to early morning, convection is continuously initiated as southeasterly flows near the surface impinge on the east side of mesoscale mountains near the coastal lines and then moves northeastward, leading to formation of two quasi-stationary rainbands. From early morning to early afternoon, new convection is repeatedly triggered along a mesoscale boundary between precipitation-induced cold outflows and warm air from South China Sea and Gulf of Tokin, resulting in the formation of “band training” of several parallel rainbands that move eastward in a later time, i.e., two scales of “training” of convective elements are found. As the MCS dissipates, a stronger squall line moves into the region from the west and passes over within about 3.5 h, contributing about 10%–15% to the total rainfall amount. It is concluded that terrain, near-surface winds, warm advection from the upstream ocean in the boundary layer, and precipitation-generated cold outflows play important roles in initiating and maintaining the extreme rain-producing MCS.

During the presummer rainy season (April–June), southern China often experiences frequent occurrences of extreme rainfall, leading to severe flooding and inundations. To expedite the efforts in improving the quantitative precipitation forecast (QPF) of the presummer rainy season rainfall, the China Meteorological Administration (CMA) initiated a nationally coordinated research project, namely, the Southern China Monsoon Rainfall Experiment (SCMREX) that was endorsed by the World Meteorological Organization (WMO) as a research and development project (RDP) of the World Weather Research Programme (WWRP). The SCMREX RDP (2013–18) consists of four major components: field campaign, database management, studies on physical mechanisms of heavy rainfall events, and convection-permitting numerical experiments including impact of data assimilation, evaluation/improvement of model physics, and ensemble prediction. The pilot field campaigns were carried out from early May to mid-June of 2013–15. This paper: i) describes the scientific objectives, pilot field campaigns, and data sharing of SCMREX; ii) provides an overview of heavy rainfall events during the SCMREX-2014 intensive observing period; and iii) presents examples of preliminary research results and explains future research opportunities.

The concept and mathematical framework of the distance velocity–azimuth display (DVAD) methodology is presented. DVAD uses rVd (Doppler velocity scaled by the distance from the radar to a gate, r) as the basis to display, interpret, and extract information from single Doppler radar observations. Both linear and nonlinear wind fields can be represented by the same Cartesian polynomial with different orders. DVAD is mathematically concise and superior to the velocity–azimuth display (VAD) in interpreting and deducing flow characteristics. The rVd pattern of a two-dimensional linear wind field is exclusively in the form of a bivariate quadratic equation representing conic sections (e.g., ellipse, parabola, and hyperbola) centered at the radar depending only on divergence and deformation. The presence of a constant background flow translates the conic sections to a different origin away from the radar. It is possible to graphically estimate the characteristics of a linear wind field from the conical sections without performing a VAD analysis. DVAD analysis can deduce quantitative flow characteristics by a least squares fitting and/or a derivative method, and is a natural way to account for nonlinearity. The rVd pattern behaves similar to a type of velocity potential in fluid mechanics where ∇(rVd) is a proxy of the true wind vector and is used to estimate the general flow pattern in the vicinity of the radar.

The ground-based velocity track display (GBVTD) was developed to deduce a three-dimensional primary circulation of landfalling tropical cyclones from single-Doppler radar data. However, the cross-beam component of the mean wind cannot be resolved and is consequently aliased into the retrieved axisymmetric tangential wind . Recently, the development of the hurricane volume velocity processing method (HVVP) enabled the independent estimation of ; however, HVVP is potentially limited by the unknown accuracy of empirical assumptions used to deduce the modified Rankine-combined vortex exponent . By combing the GBVTD with HVVP techniques, this study proposes a modified GBVTD method (MGBVTD) to objectively deduce from the GBVTD technique and provide a more accurate estimation of and via an iterative procedure to reach converged and cross-beam component of solutions. MGBVTD retains the strength of both algorithms but avoids their weaknesses. The results from idealized experiments demonstrate that the MGBVTD-retrieved cross-beam component of is within 2 m s−1 of reality. MGBVTD was applied to Hurricane Bret (1999) whose inner core was captured simultaneously by two Weather Surveillance Radar-1988 Doppler (WSR-88D) instruments. The MGBVTD-retrieved cross-beam component of from single-Doppler radar data is very close to that from dual-Doppler radar synthesis using extended GBVTD (EGBVTD); their difference is less than 2 m s−1. The mean difference in the MGBVTD-retrieved from the two radars is ~2 m s−1, which is significantly smaller than that resolved in GBVTD retrievals (~5 m s−1).

Typhoon Morakot yielded record-breaking precipitation and caused severe disasters in Southern Taiwan in early August 2009. This event revealed the desperate need for extreme rainfall and disaster prediction. The multiscale analysis on circulations demonstrates that Typhoon Morakot not only developed in a monsoon trough, which is a large-scale system in favor of the development of convection systems, it was also involved with complicated features of different scales that included typhoon circulation itself, the circulation of northwestward-propagating 10–30-day oscillation, and the circulation of northward-propagating 40–50-day oscillation. The other remarkable features of Typhoon Morakot are: the asymmetric structure of convection, less damage occurred near the typhoon center than at the fringes, and the record-breaking torrential rain in the mountain areas of south Taiwan induced by the flows on the south fringe of the typhoon. The causes of disastrous precipitation from Typhoon Morakot can be stated as:...

The ground-based velocity track display (GBVTD) technique was developed to estimate the primary circulations of landfalling tropical cyclones (TCs) from single-Doppler radar data. However, GBVTD cannot process aliased Doppler velocities, which are often encountered in intense TCs. This study presents a new gradient velocity track display (GrVTD) algorithm that is essentially immune to the Doppler velocity aliasing. GrVTD applies the concept of gradient velocity–azimuth display (GVAD) to the GBVTD method. A GrVTD-simplex algorithm is also developed to accompany GrVTD as a self-sufficient algorithm suite. The results from idealized experiments demonstrate that the circulation center and winds retrieved from GrVTD with aliased velocity and GBVTD with dealiased velocity are in good agreement, but GrVTD is more sensitive to random observation errors. GrVTD was applied to Hurricane Charley (2004) where the majority of the Doppler velocities of the inner-core region were aliased. The GrVTD-retrieved circulation pattern and magnitude are nearly identical to those retrieved in GBVTD with manually dealiased velocities. Overall, the performance of GrVTD is comparable but is more sensitive to the data distribution than that of the original GBVTD using dealiased velocity. GrVTD can be used as a preprocessor for dealiasing velocity in TCs before the data are used in GBVTD or other algorithms.

This study investigates a long-duration mesoscale system with extremely heavy rainfall over southwest Taiwan during the Terrain-influenced Monsoon Rainfall Experiment (TiMREX). This mesoscale convective system develops offshore and stays quasi-stationary over the upstream ocean and southwest coast of Taiwan. New convection keeps developing upstream offshore but decays or dies after moving into the island, dropping the heaviest rain over the upstream ocean and coastal regions. Warm, moist, unstable conditions and a low-level jet (LLJ) are found only over the upstream ocean, while the island of Taiwan is under the control of a weak cold pool. The LLJ is lifted upward at the boundary between the cold pool and LLJ. Most convective clusters supporting the long-lived rainy mesoscale system are initiated and develop along that boundary. The initiation and maintenance is thought to be a “back-building–quasi-stationary” process. The cold pool forms from previous persistent precipitation with a temperature depression of 2°–4°C in the lowest 500 m, while the high terrain in Taiwan is thought to trap the cold pool from spreading or moving. As a result, the orography of Taiwan is “extended” to the upstream ocean and plays an indirect effect on the long-duration mesoscale system.

In this study, a fuzzy logic algorithm is developed to provide objective guidance for the prediction of afternoon thunderstorms in northern Taiwan using preconvective predictors during the warm season (May–October) from 2005 to 2008. The predictors are derived from surface stations and sounding measurements. The study is limited to 277 days when synoptic forcing was weak and thermal instability produced by the solar heating is primarily responsible for thunderstorm initiation. The fuzzy algorithm contains 29 predictors and associated weights. The weights are based on the maximum of the critical success index (CSI) to forecast afternoon thunderstorms. The most important predictors illustrate that under relatively warm and moist synoptic conditions, sea-breeze transport of moisture into the Taipei Basin along with weak winds inland provide favorable conditions for the occurrence of afternoon convective storms. In addition, persistence of yesterday’s convective storm activity contributed to improving today’s forecast. Skill score comparison between the fuzzy algorithm and forecasters from the Taiwan Central Weather Bureau showed that for forecasting afternoon thunderstorms, the fuzzy logic algorithm outperformed the operational forecasters. This was the case for both the calibration and independent datasets. There was a tendency for the forecasters to overforecast the number of afternoon thunderstorm days. The fuzzy logic algorithm is able to integrate the preconvective predictors and provide probability guidance for the prediction of afternoon thunderstorms under weak synoptic-scale conditions, and could be implemented in real-time operations as a forecaster aid.

This study examines a subtropical oceanic mesoscale convective vortex (MCV) that occurred from 1800 UTC 4 June to 1200 UTC 6 June 2008 during intensive observing period (IOP) 6 of the Southwest Monsoon Experiment (SoWMEX) and the Terrain-influenced Monsoon Rainfall Experiment (TiMREX). A dissipating mesoscale convective system reorganized within a nearly barotropic vorticity strip, which formed as a southwesterly low-level jet developed to the south of subsiding easterly flow over the southern Taiwan Strait. A cyclonic circulation was revealed on the northern edge of the mesoscale rainband with a horizontal scale of 200 km. An inner subvortex, on a scale of 25–30 km with maximum shear vorticity of 3 × 10−3 s−1, was embedded in the stronger convection. The vortex-relative southerly flow helped create local potential instability favorable for downshear convection enhancement. Strong low-level convergence suggests that stretching occurred within the MCV. Higher θe air, associated with significant potential and conditional instability, and high reflectivity signatures near the vortex center suggest that deep moist convection was responsible for the vortex stretching. Dry rear inflow penetrated into the MCV and suppressed convection in the upshear direction. A mesolow was also roughly observed within the larger vortex. The presence of intense vertical wind shear in the higher troposphere limited the vortex vertical extent to about 6 km.

During the Terrain-Influenced Monsoon Rainfall Experiment (TiMREX), which coincided with Taiwan’s Southwesterly Monsoon Experiment—2008 (SoWMEX-08), the upper-air sounding network over the Taiwan region was enhanced by increasing the radiosonde (“sonde”) frequency at its operational sites and by adding several additional sites (three that were land based and two that were ship based) and aircraft dropsondes. During the special observing period of TiMREX (from 15 May to 25 June 2008), 2330 radiosonde observations were successfully taken from the enhanced network. Part of the challenge of processing the data from the 13 upsonde sites is that four different sonde types (Vaisala RS80, Vaisala RS92, Meisei, and Graw) were used. Post–field phase analyses of the sonde data revealed a significant dry bias in many of the sondes—in particular, in the data from the Vaisala RS80 sondes that were used at four sites. In addition, contamination of the sonde data by the ship’s structure resulted in poor-quality low-level thermodynamic data at a key oceanic site. This article examines the methods used to quality control the sonde data and, when possible, to correct them. Particular attention is given to the correction of the humidity field and its impact on various convective measures. Comparison of the corrected sonde humidity data with independent estimates shows good agreement, suggesting that the corrections were effective in removing many of the sonde humidity errors. Examining various measures of convection shows that use of the humidity-corrected sondes gives a much different perspective on the characteristics of convection during TiMREX. For example, at the RS80 sites, use of the corrected humidity data increases the mean CAPE by ∼500 J kg−1, decreases mean convective inhibition (CIN) by 80 J kg−1, and increases the midlevel convective mass flux by greater than 70%. Ultimately, these corrections will provide more accurate moisture fields for diagnostic analyses and modeling studies.

Three years’ worth of radar reflectivity data from four radars in an area of complex terrain (Taiwan) from 2005 to 2007 were analyzed and a reflectivity climatology was developed. The climatology was applied in the construction of new hybrid scans to minimize the impacts of ground clutter and beam blockages. The reflectivity climatology showed significant seasonal variations and captured distributions of ground/sea clutters, beam blockages, and anomalous propagations in addition to precipitation systems in the radar domains. By comparing the reflectivity climatology with gauge observations, it was found that 15 (20) dBZ was a good approximation for rain/no-rain segregation during cool (warm) seasons. Comparisons between the standard (i.e., based on terrain and scan strategies only with the assumption of standard propagations) and nonstandard (i.e., standard plus the clutter and blockage mitigation using the reflectivity climatology) hybrid scans showed that the former did not accurately reflect the clutter and blockage distributions in the real atmosphere. The application of the reflectivity climatology was shown to significantly reduce the impacts of clutter and blockages and provided improved radar quantitative precipitation estimates (QPEs) in the complex terrain.

A tropical cyclone (TC) eye tracking (TCET) algorithm is presented in this study to objectively identify and track the eye and center of a tropical cyclone using radar reflectivity data. Twelve typhoon cases were studied for evaluating the TCET algorithm. Results show that the TCET can track TC centers for several hours. The longest tracking time is about 35 h. Eye locations estimated from different radars showed consistency with a mean distance bias of about 3.5 km and a standard deviation of about 1.5 km. The TCET analysis shows decreasing eye radius as TCs approach land, especially within 50 km of the coastline. The TCET algorithm is computationally efficient and can be automated by using the TC center in the previous volume or the estimated center from satellite images as an initial guess. The TCET may not accurately find the TC center when a TC is weak or does not have an enclosed eyewall or when it does have highly noncircular eyes. However, the algorithm is still suitable for operational implementation and provides high spatial and temporal resolution information for TC centers and eye radii, especially for intense TCs.

The primary circulation of atmospheric vortices, such as tropical cyclones and tornadoes, can be estimated from single-Doppler radar observations using the ground-based velocity track display (GBVTD) algorithm. The GBVTD algorithm has limitations in the following four areas: 1) distortion in the retrieved asymmetric wind fields, 2) a limited analysis domain, 3) the inability to resolve the cross-beam component of the mean wind, and 4) the inability to separate the asymmetric tangential and radial winds. This paper presents the generalized velocity track display (GVTD) algorithm, which eliminates the first two limitations inherent in the GBVTD technique and demonstrates the possibility of subjectively estimating the mean wind vector when its signature is visible beyond the influence of the vortex circulation. In this new paradigm, the GVTD algorithm fits the atmospheric vortex circulation to a new variable VdD/RT in a linear azimuth angle (θ′), rather than the Doppler velocity Vd in a nonlinear angle (ψ), which is used in GBVTD. Key vortex kinematic structures (e.g., mean wind, axisymmetric tangential wind, etc.) in the VdD/RT space simplify the interpretation of the radar signature and eliminate the geometric distortion inherent in the Vd display. This is a significant improvement in diagnosing vortex structures in both operations and research. The advantages of using VdD/RT are illustrated using analytical atmospheric vortices, and the properties are compared with GBVTD. The characteristics of the VdD/RT display of Typhoon Gladys (1994) can be approximated by a constant mean wind plus an axisymmetric vortex.

Summary Instantaneous quantitative precipitation estimation (QPE) and quantitative precipitation forecasting (QPF) by utilizing the meteorological radar data are potentially of great benefit for operational hydrology in river basins. The most commonly used technique of radar-based rainfall estimation is a best fitted power-law function between reflectivity ( Z ) and rain intensity ( R ). An emerging tool in QPE/QPF using radar data is the artificial neural network (ANN) that is capable of learning complex nonlinear relationships. In this study, we introduce a dynamic ANN approach to construct instantaneous QPE and one-hour-ahead QPF by using a three-dimensional radar data structure, which takes into account the terminal velocity and the horizontal advection. Radar measurements and three rain gauges for six typhoon events in the Keelung River, Taiwan, were used for calibrating and evaluating the Z – R relation and dynamic ANN models. The results of current rainfall estimation and one-hour-ahead rainfall forecasting all indicate that the dynamic ANN can produce much more accurate and stable performance than the Z – R power-law models. This study demonstrates that the dynamic ANN can be applied successfully in instantaneous QPE and QPF by using remote sensing data.