Evaluation of various QPE algorithms using Korea MOLIT S-band dual-polarization radar observations

Accurate observation of precipitation is a key issue to predict and prevent natural disasters (e.g. flood, landslide, and heavy snow). Especially in Korea, flash flood in the mountainous and urban areas causes severe damages in life and property due to intensive rainfall. Traditionally, the flash flood warnings have been established by ground and single-polarization-radar-based rainfall. However, the methods have been found to expose serious limitations of spatial extrapolation and relatively high uncertainty in rainfall estimation. In the recent past, dual-polarization radar technology has been becoming an alternative method to implement over large areas with high reliability. Technical advances in dual-polarization radars have finally enabled us to observe multiple hydrometeorological variables at various scales for a couple of decades. As becoming able to measure the differential reflectivity, specific differential phase, and cross-correlation coefficient, reliability has been significantly improved compared to the conventional reflectivity-based method (Ryzhkov et al. 2005a). Some of the polarimetric parameters have technical advantages that the specific differential phase is immune to radar miscalibration, attenuation in precipitation, and partial blockage of the radar beam, for example (Zrnic and Ryzhkov 1996). There have been many studies all over the world to present reliable polarimetric Quantitative Precipitation Estimation (QPE) algorithms under various meteorological conditions. Based on the advantages of the specific differential phase, power-law relation between the specific differential phase and precipitation was adopted with regionalized coefficients (Bringi and Chandrasekar 2001; Gorgucci et al. 2001; Matrosov et al. 2002; Park et al. 2005; Ryzhkov et al. 2005a). Given the simplicity, the power-law-based methods have a potential to be used as baseline products. Ryzhkov et al. (2005b) classified precipitation events by scales and types and suggested a synthetic algorithm for the Joint Polarization Experiment (JPOLE). Cifelli et al. (2011) developed a more sophisticated algorithm with different classification criteria. The reflectivity, differential reflectivity, and specific differential phase were employed as thresholds to categorize precipitation events into five groups. Five equations induced by different variables were adopted to make appropriate and accurate estimations. Several government agencies in Korea have just started to adopt the cutting-edge technology and to apply over nationwide areas. As a pilot phase, Ministry of Land, Infrastructure and Transport (MOLIT) has installed and been operating two S band radars at Mt. Sobaek and Mt. Biseul since Nov 2011 and June 2009, respectively (Fig. 1). The radars were designed to cover Nakdong river watershed, which is the second largest watershed located in the southern-east part of Korean Peninsula. Observation range is 100 km and spatial resolution is 125 m. Scan speed for operation is currently 3 rpm and volume scan takes 2.5 minutes of each cycle. In order to set the optimum scan strategy and major thresholds for quality control, domestic researches have been actively conducted. However, assessment of polarimetric QPE algorithms has been barely conducted in Korea. In this study, polarimetric QPE algorithms developed by the institutions securing high level of technology were adopted and validated with ground rainfall observations for various environments. Korea is represented by complex topography with urban and mountainous areas that impacts of spatial scale was also examined. Based on the evaluation results, applicability of the QPE algorithms was fully investigated. This study would contribute to suggest proper application of the QPE algorithms in Korea.