The extended Kalman filter is an extended version of the Kalman filter for a non-linear problem. This study applies this extended Kalman filter to the real-time estimation of the parameters of the dual-pol radar rain rate estimator. The estimated parameters are also compared with those based on the method of least squares. As an application example, this study considers storm events observed by the Beaslesan radar in Korea. The findings derived include, first, the parameters of the radar rain rate estimator obtained by the extended Kalman filter are totally different from those by the method of least squares. In fact, the parameters obtained by the extended Kalman filter are found to be more reasonable, and similar to those reported in previous studies. Second, the estimated rain rates based on the parameters obtained by the extended Kalman filter are found to be more similar to those observed on the ground. In conclusion, the extended Kalman filter can be a reliable method for real-time estimation of the parameters of the dual-pol radar rain rate estimator. The resulting rain rate is also found to be of sufficiently high quality to be applicable for other purposes like various flood warning systems. Acknowledgement: This work was supported by Korea Environment Industry & Technology Institute(KEITI) through Water Management Research Program, funded by Korea Ministry of Environment(MOE)(127559)

This study discusses the uncertainties of precipitation forcing inputs from multiple sources, including satellite or satellite-gauge quantitative precipitation estimates (QPEs), and quantitative precipitation forecasts (QPFs) from the global ensemble forecast system (GEFS) and mesoscale weather model WRF. The impacts of multiple precipitation forcing inputs on streamflow simulations are investigated though distributed hydrological modeling over the upper Huaihe River basin.     First, the distributed Variable Infiltration Capacity (VIC) model is used to investigate the impact of multiple satellite QPEs on the daily streamflow simulations. These QPEs include the TRMM/TMPA 3B42RT product, the NOAA/CPC’s CMORPH gauge merged product, and the PERSIANN CDR product. Also, an error correction method – SREM2D (two-dimensional stochastic satellite rainfall error model) has been applied to three selected satellite QPEs to implement streamflow simulations in the VIC model. Overall, SREM2D provides great potential to facilitate the application of satellite precipitation products in water management and decision making over Chinese river basins.      In addition, the impact of bias-corrected global ensemble precipitation forecasts on the improvement of summer streamflow prediction skill is discussed. Ensemble streamflow forecasting is implemented over the study basin using the VIC model, driven by the bias corrected ensemble precipitation forecasts from the GEFS reforecast data. The encouraging results suggest that the reforecast ensemble dataset exhibits a great value to improve hydrometeorological predictions for operational applications.     To further examine the impact of high-resolution QPFs on hydrological processes, the two-way coupled WRF/WRF-Hydro model is adopted to investigate the hydrological feedback on land surface-atmosphere interactions over the study basin. The differences of spatial precipitation patterns in one-way and two-way modelings are highly associated with the prevailing westerly wind and orographic effect. The hydrological effects result in a wetter near-surface air layer and more unstable atmosphere conditions, which may induce more precipitation during rainy seasons.

Extreme Precipitation Events (EPEs) have increased both in frequency and magnitude in India during the past decade. Various Quantitative Precipitation Forecasts (QPF) are available from the Numerical Weather Prediction Models such as the European Centre for Medium-range Weather Forecasts (ECMWF), Japan Meteorological Agency (JMA), National Centre for Medium Range Weather Forecasting (NCMRWF) and UK Met Office (UKMO). Since several QPFs are available with some characteristic difference, their comparative evaluation is vital to identify the most adequate one which can capture reliable information of precipitation extremes in India. Past studies have assessed the performance of QPFs only on the statistical basis and ignored their reliability to represent the distribution of intrinsic spatial structure of extreme precipitation. In this study, the performance of four different QPFs is evaluated to forecast the occurrence of various EPEs in different parts of India during the past decade. The representation of spatial pattern of EPEs is assessed by creating a network among the grids experiencing extreme precipitation using the theory of complex networks. The performance of QPFs is compared to the network created using the available gridded observation data for evaluating their relative accuracy. For this, first a network of rain grids is identified and created on the basis of their occurrence during the past decade. Three measures of CN such as Degree Centrality (DC), Betweenness Centrality (BC) and Clustering Coefficient (CC) are used to quantify different aspects of the internal spatial structure of extreme rainfall during the monsoon months of June, July, August and September (JJAS). The results from the ongoing work will be presented at the conference.

Advances in computing power and data assimilation techniques have enabled numerical weather prediction (NWP) models to issue forecasts at increasingly finer temporal and spatial resolution. High-resolution precipitation forecasts are expected to increase the accuracy of forecasts for floods and high flow events, particularly for rapid response catchments. However, all NWP forecasts contain systematic errors that limit their value for hydrological applications, particularly for flood forecasting. In order to reduce such errors and quantify uncertainties in precipitation predictions, we have developed a method called Catchment-scale Hydrologic Pre-processing of Precipitation forecasts (CHyPP).  CHyPP has been extensively tested for pre-processing precipitation forecasts from low-resolution NWP models in a range of Australian and international catchments. In this study we assess the benefit to streamflow forecasts of pre-processing precipitation forecasts from the high-resolution Australian NWP ACCESS-C (~1.5 km) model compared with low-resolutions ACCESS-R (~12 km) and ACCESS-G (~40 km) forecasts. We generate retrospective streamflow forecasts by forcing a hydrological modelling system with CHyPP forecasts from high and low-resolutions NWP models. The hydrological modelling system consists of semi-distributed hydrological models with GR4H rainfall-runoff model, Muskingum channel routing and the hydrological model error ERRIS.  CHyPP is highly effective at reducing forecast errors and produces reliable forecasts from both low and high-resolution NWP models. Despite the increase in NWP model resolution, we find that CHyPP forecasts from high-resolution NWP model produce streamflow forecasts of equivalent quality of those produced from low-resolution NWP models. In this presentation, we discuss the possible reasons for this. Although the high-resolution NWP model does not offer clear benefits over the lower resolution NWP models, the benefits of high-resolution models are likely to be more evident when modelling hydrological processes at finer spatial resolutions and using high-resolution precipitation observations.

Precipitation is one of the key components of the global water cycle, investigated intensively in many scientific fields. The microphysical characterization of rainfall play an important role in the process of precipitation estimation, remote sensing observations, radio communications, and cloud microphysics. However, few studies in the literature have focused on the large-scale assessment of rainfall microphysics characteristics. Since disdrometers measure raindrop size distribution (DSD) at point sites and ground dual-polarization radars that can be used to estimate DSD parameters are available in limited areas, space-based radars and mesoscale numerical weather prediction models provide the possibility to measure the DSD on a large scale. This study therefore proposes a retrieval scheme of large-scale raindrop microphysics characteristics based on Weather Research and Forecasting (WRF) and Global Precipitation Measurement (GPM) Dual-frequency Precipitation Radar (DPR), and assesses DSD, rain rate (R), kinetic energy and reflectivity (Z) by an impact-type disdrometer over Chilbolton, UK. Results show that both retrieval from the WRF and GPM provide an overall good fit to the reality, indicating the feasibility to use measurement or simulation to obtain raindrop microphysical processes on a large scale, especially that GPM mode has a perfect performance, which based on the measured data. This study can help to the improvement of large-scale rainfall microphysics estimation and provide insight as to large scale extension research of raindrop characteristics like rainfall erosivity estimation and accurate precipitation forecast.

When the research area is local scale or regional, and the data is large scale or global, analyzing regional characteristics with large scale data will cause huge errors. Downscaling is a method which increase the spatial resolution of the data and we will have a deeper understanding and accurate analysis of the research area. The study is aim to perform spatial downscaling of wind-related data (wind speed, wind direction,  etc.). The data we use are GCM historical and future simulation data and regional historical observation data. This research applies Diffeomorphic Dimensionality Reduction (Diffeomap), a nonlinear dimensionality reduction algorithm, to establish the relationship between GCM historical data and regional historical observation data. In addition, this also combines the relationship with GCM future simulation data to estimate the regional future data. At the same time, the goal of this study - downscaling - also achieved. The results provide decision-making units as a reference for future risks.

In mathematics, topology concerns about the invariant features under continuous deformation (i.e. stretching, bending, or twisting), for example, compactness and connectedness, which is different from the geometrical properties like length and curvature. For data analysis, the properties of topology make it more robust to noise and sufficient for complex datasets. As a result, our study applied the topological data analysis in the large-scale climate system, the Pacific. The Pacific Ocean covers almost one-third of the earth and more than 30,000 islands within. The climate system here is complex and chaotic, such as trade wind, South Pacific Convergence Zone (SPCZ), and El Niño Southern Oscillation (ENSO). In our study, we quantified the climate characteristic by topological data analysis strategies including Betti numbers, Euler integral, and persistence homology, providing new information for the Pacific hydrology.

Despite having a subtropical climate, typhoon-induced precipitation in Taiwan rarely occurs between November and May. Therefore, water scarcity caused by the uneven seasonal rainfall distribution results in reducing or suspending water supply for irrigation purposes, especially in Taipei, Taoyuan and Hsinchu. Keeping sufficient water levels in the reservoirs and decreasing the likely impact of the drought in these areas have become a major challenge for the authorities.This study conducts system dynamics and bifurcation analysis of water resources systems in Taiwan under uncertain scenarios. The main focus is the equilibrium and stability of the systems. Furthermore, bifurcation and chaotic behavior of uncertainty are simulated. 

Taiwan is an island that presents a south-north-long type. Most of the rivers in Taiwan flow laterally and the slopes of the river bed are steep.  The rainfall is usually concentrated in the early-summer rainy and typhoon season per year. The precipitation during a typhoon or torrential rain period increases the flow rate of rivers in the catchment area. Sediments carried by flood flow deposits in the reservoir, which may affect the hydrological and ecological environments downstream of the reservoir. Therefore, it is important to concern reservoir desiltation management strategies which require the measurement and analysis of high sediment concentration in the reservoir region. In this research, time-domain reflectometry (TDR) and ultrasonic monitoring system have been designed and manufactured for measuring sediment concentration during a typhoon and torrential rain periods. The stratified sediment concentrations at the water intake in front of Zengwen dam (at EL.177.5, 185, 190, 200, and 210 m.) were observed by the equipment during Typhoon Soulik in 2013. The sediment concentrations sampled in the field (at EL.177.5, 185, and 210 m.) were also obtained by the oven drying method in the laboratory. Besides, the concentration of the bottom outlet (at EL.177.5 m.) is calculated by a two-dimensional hydraulic and sediment transport model. The measured data by the sediment monitoring system are compared with the results by the oven drying method.  The numerically simulated results of sediment concentrations are similar to those observed by the sediment monitoring system. The monitoring technology can be applied to reservoirs in Taiwan for sediment concentration measurement, and the results of measured data and developed numerical model in the field can be both used for improving reservoir desilting operations during typhoon floods.

Limited by the natural environments of heavy rain and strong wind, it may be impossible to use unmanned aerial vehicles (UAV) to photograph the water surface of reservoirs during typhoon floods. Alternatively, in this study, the images recorded by the CCTV built at the reservoir banks are adopted as the data source for the flow patterns analysis at the reservoir. There are four locations where the image data was taken. They are images of upstream and downstream of No. 1 barrier taken by a network camera and images of upstream and downstream of No. 2 barriers captured by a dome camera. In the analysis, the time period with stronger surface ripples on images is selected for flow-field analysis using the Particle Image Velocimetry (PIV) and Particle Track Velocimetry (PTV) method. Then, a discussion of the feasibility and applicability of flow analysis using CCTV images is given. In this study, the images on June 14 of 2019, and May 22 and 23 of 2020 corresponding to the heavy rain period are used for analysis. The preliminary result shows that the measured flow field could be consistent with the observation if the surface tracking particles are sufficient. However, when surface ripples are lack, the analysis of the flow field could be difficult and result in unsatisfactory results. Therefore, in this study, it is still difficult and limited to use CCTV images to analyze the flow field at the reservoir. A further study is needed.