Vol.125 (Art, Culture, Game, Graphics, Broadcasting and Digital Contents 2016), pp.72-77 http://dx.doi.org/10.14257/astl.2016. A Study on Estimation Technique of Extreme Precipitation Diameter Jin Woo Choi 1, Tae Min Kim 2, Taeg Keun Whangbo 2* 1 Culture Technology Institute, 2 Department of Computer Engineering, Gachon University 1342 Seongnamdaero, Sujeong-gu, Seongnam-si, Gyeonggi-do, 461-701, Korea {cjw49, scc0309, * tkwhangbo}@gachon.ac.kr Abstract. Recently, meteorological information is processed through diversified sensors and researching meteorological pattern by analyzing this information is very important work. At the time of extreme precipitation, cloud features could be modeled by analyzing precipitation pattern. This modeling is implemented by extracting extreme precipitation data by each AWS position and analyzing surrounding precipitation distribution contour diameter relevant to rainfall days and large capacity operation and statistical analysis are required to be automated. In this study, its objective is to research on a technique of being able to estimate max. precipitation diameter by each 5mm unit automatically. Keywords: Extreme precipitation, Contour diameter, AWS 1 Introduction Today, meteorological climate information is being utilized in diversified fields and its accurate and effective use is an important work. By identifying and predicting atmospheric environmental condition, its favorable effect could be exerted to public convenience and overall industry. Therefore, prediction of environmental meteorology that could be also utilized as a simple information by modeling regional, temporal meteorological condition is required to be developed most urgently as an essential condition [5]. Meteorological disaster by local meteorological disturbance such as severe heavy rainfall, seasonal rain front, suddenly changed severe weather brought forth heavy damage of human life and property and such economic damage was increased by over 30% for the last 10 years (1997-2006) [1]. Recently, in case of guerilla type heavy rain being frequently taken place in summertime, it induces landslide and flood by heavy rainfall in short period of time and severe damage by this rainfall has been taken place [5]. Occurrence frequency of extreme precipitation is estimated by collecting meteorological observation data for the past 30 years and analyzing it statistically and general cloud feature is analyzed by estimating information for date and region of such frequency. ISSN: 2287-1233 ASTL Copyright 2016 SERSC
Cloud feature classification is enabled through precipitation distribution analysis at the time of extreme precipitation by each AWS position. In this study, precipitation distribution data of precipitation extreme days is converted to gridded contour data by using Voxler 3D visualized program [2] that may generate 3D image and model such as meteorological data and GIS data without difficulty after estimating precipitation extreme value by each AWS position. Afterwards, feature of extreme precipitation zone is analyzed by developing a technique that may estimate max. diameter of precipitation contour of 5mm unit by each AWS position 2 Extreme precipitation Data used for determining extreme precipitation is rainfall observation data for 1 hour having been observed for 31 years from 1980 to 2010 being observed at 91 positions among ground observation positions being operated by installing it nationwide by Korea Meteorological Administration. Among these, as observation position at 8 points (172, 217, 258, 259, 263, 264, 276, 283) being composed of observation data below 2 years was excluded, total analysis position is observation data of 83 positions [3]. Extreme precipitation estimation method being used in this study is frequency estimation of each extreme precipitation in monthly dataset. Occurrence frequency of extreme precipitation is estimated by 4 rank conditions of 0.1, 0.5, 1, and 5%. In observation data of 1 hour precipitation for 30 years, one dataset by each month is composed (app. 720ea). In this dataset, temporal/spatial information relevant to 0.1, 0.5, 1, 5% is estimated respectively. Through mutual comparison of extreme precipitation occurrence frequency of 0.1, 0.5, 1, 5% being estimated in monthly dataset for 30 years, one extreme precipitation with most severe rainfall is determined (Table 1). Table 1. Precipitation extreme data in certain position of AWS Position Date 0.1% Date 0.5% Date 1% Date 5% 90 199808150300 21.7 199808080700 12.3 200408181800 9 198008261800 2.1 100 200308250500 30.5 199108232000 15.3 198408151900 10.7 199608261900 2.9 101 199107252100 30 200407162100 16.5 199507160500 11.7 199907280600 2.8 105 198508101700 30.9 200908111700 12.5 200508250100 8.5 200808030100 2 108 199808080300 39.4 198008131700 18.9 201007021100 12.5 200307191000 2.5 112 200807201900 32 200708141200 16 198007132200 10.1 198107060400 1.9 114 198007140900 30.5 200607021000 17 200307220700 12 200207051000 2.5 115 200709220200 19.5 199107071700 8.9 200308231200 6 200907091600 1 119 200007221400 32 199707150800 17.2 199607280200 11.3 199407011400 2 Copyright 2016 SERSC73 73
127 199707011100 28 200107210700 14.2 199107151700 10 200307180600 2 129 200608170200 28.5 200708040800 17.5 200008270100 10.8 200507170600 1.5 232 199508090200 31.5 200708050900 16.5 198908291900 10.5 199507191500 1.5 235 200008200500 37.5 200208060600 17 198508021200 10.5 198807262100 1.1 236 199808111100 32 200207221400 16 199707061300 11 198307291700 1.9 238 200007230600 27.5 199707051400 15 198407050900 10.1 199907020900 2 243 198707261200 29 200307110600 16.5 199007160300 10 200507010600 1.5 244 198408310300 31.5 199608210300 15.5 200007021900 10 198907262300 2 245 200307042400 28.5 200907160800 15 200707010800 10 199307120400 1.5 In monthly dataset of 0.1% extreme precipitation, date and position value corresponding to 0.74, 0.72, 0.696, 0.672 is determined by extreme precipitation and in extreme precipitation of 0.5%, a value corresponding to 3.72, 3.6, 3.48, 3.36, in 1%, 7.44, 7.2, 6.96, 6.72 and in that of 5%, 37.2, 36, 34.8, 33.6 are determined as extreme precipitation. 3 Estimation of contour data by each extreme precipitation position Precipitation data of AWS position relevant to extreme rainfall days by each AWS position is converted to input data form of Voxler and composed as shown on Table 2. Table 2. 0.1% Extreme Precipitation distribution data at No. 90 position of AWS Longitude Latitude Precipitation(mm) 128.5647 38.25085 21.7 127.3042 38.14788 17.2 127.0607 37.90186 22.5 128.7183 37.67713 10.3 127.7357 37.90256 7.7 128.891 37.75147 8.5 127.9466 37.33756 3 130.8986 37.48129 2.3 127.4407 36.63924 0.8 127.3721 36.372 3.3 127.9946 36.22023 0.3 128.7073 36.57293 0.1 129.3796 36.03259 0.2 126.7614 36.0053 4.6 128.619 35.88515 0.1 127.155 35.8215 0 129.3203 35.56014 0.1 74 Copyright 2016 SERSC
In order to express input data of Table 2 as contour, it is required to be converted to formulated grid data. If applying inverse distance weighted method (IDW) by using each AWS position and precipitation, gridded rainfall data could be generated. Gridded rainfall data could be visualized in a form of contour as shown on Fig. 1 through Voxler. Fig. 1. Visualization of AWS rainfall data contour 4 Estimation technique of rainfall contour max. diameter In order to estimate diameter of contour by each 5mm unit in specific area based on AWS position of rainfall contour as shown on Fig. 2, position value of max. diameter (both end point) is required. In order to identify position of both ends, position value of each image pixel is required to be estimated. Each pixel of image being extracted from Voxler does not include position (longitude, latitude) value. In order to assign position value to each pixel, reference outline is required to be expressed as shown on Fig. 2 through a function of drawing polygon of Voxler. Copyright 2016 SERSC75 75
Fig. 2. Rainfall contour image including reference outline In order to obtain longitude, latitude difference by each pixel, number of longitudinal, traverse outline pixel is calculated and max. longitude, latitude difference of outline is also calculated. Through ratio between calculated outline pixel number and longitude, latitude difference value, longitude, latitude difference value per 1 pixel could be obtained. When longitude, latitude value of each pixel is assigned in above method, max. diameter of precipitation by 5mm unit in specific area by each AWS position could be extracted. Table 2. Illustration of rainfall contour diameter information around AWS No. 112 position (0.1%, 32mm, 200807201900) Precipitation diameter(km) Latitude1 Longitude1 Latitude2 Longitude2 5mm 122.707 127.126006 37.689394 126.071472 37.143939 10mm 122.134 127.156353 37.386364 126.071472 37.098485 15mm 84.294 126.458387 37.727273 127.148766 37.204545 20mm 76.686 126.473561 37.712121 127.103247 37.242424 25mm 36.696 126.488734 37.689394 126.807370 37.545455 30mm 17.274 126.572186 37.712121 126.655639 37.492424 35mm 13.907 126.587359 37.696970 126.663225 37.530303 40mm 9.976 126.594946 37.674242 126.648052 37.553030 Extreme precipitation of Rank 0.1% in AWS No. 112 position is 32mm and observation time is 19 hours, July 20, 2008. Diameter information of rainfall contour 76 Copyright 2016 SERSC
by 5mm around AWS No. 112 position is as shown on Table 2. Latitude, longitude1 & 2 are both ends of max. diameter and max. diameter (km) could be obtained by calculating the distance of two points. 5 Conclusion In order to analyze cloud features during extreme precipitation, feature survey like precipitation distribution around extreme position by each AWS position is required. By calculating extreme precipitation frequency of 83 AWS positions of which precipitation was measured for 30 years, precipitation date and quantity of 0.1%, 0.5%, 1%, 5% were estimated. Gridded distribution data of surrounding AWS precipitation on relevant rainfall date was made by using Voxler that is 3D visualizing tool and through such data, contour image was generated. As a lot of time is required and error size is likely to be magnified if manually estimating max. diameter of precipitation by 5mm unit of contour image being generated like this, we developed a technique of being able to calculate precipitation diameter automatically. By adding reference outline information to precipitation contour image being generated through Voxler, position value was assigned to all the pixels of precipitation contour image and through each pixel value having position information, max. diameter of precipitation zone by 5mm unit in specific area defined by user by each AWS position could be estimated. Through this study, it is expected that general cloud features could be analyzed by modeling regional and daily precipitation intensity and its zone size. Acknowledgments. This research was supported by Basic Science Research Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Education(NRF-2013R1A1A2057851) References 1. Statistics Korea, www.kostat.go.kr 2. Voxler Related Information, http://www.goldensoftware.com/voxler 3. Choi, Y., Kim, M.G., Kim, Y.J., Park, C.Y.: Characteristics and Changes of Extreme Precipitation Events in the Republic of Korea, 1954~2010: Their Magnitude, Frequency, and Percent to Total Precipitation: vol.6, pp. 45--58. Journal of Climate Research (2011) 4. Nam, J.Y., Lee, Y.H., Ha, J.C., Jung, G.Y. : Generation of precipitation re-analysis data through development of synthesized precipitation production technique in Korean peninsula: pp. 136--137. Conference Proceedings of Korean Meteorological Society (2012) 5. AGENCY FOR DEFENSE DEVELOPMENT : Meteorological data processing and analysis services, AGENCY FOR DEFENSE DEVELOPMENT Report (2014) 6. Tae, H.U., Kim, H.I., Park, K.D.: Development of a Virtual Reference Station-based Correction Generation Technique Using Enhanced Inverse Distance Weighting: vol.4 no.2 pp. 79--85. Journal of Positioning, Navigation, and Timing (2015) Copyright 2016 SERSC77 77