2016 Asian Water Cycle Symposium March 02, 2016 Prof. Deg-Hyo, Bae(dhbae@sejong.ac.kr) Dept. of Civil & Environ. Engineering, Sejong Univ., Seoul, Korea
Background and Objective Current drought occurrence in Korea (1980 2015) Minor or serious drought occurrence per 2-3 year cycle, but almost every year since 2008 1981 1984 1988 1991 1996 1994 1995 2001 2004 2006 2011 2009 2010 2015 2013 2014 Precipitation 400 300 200 100 10yr Ave Rainfall 2008 Rainfall 2008 년가뭄 1982 1992 2000 2008 2012 0 1 2 3 4 5 6 7 8 9 10 11 12 Main Causes of drought occurrence Imbalance of water budget mainly due to rainfall deficiency and temperature increase Lack of rainfalls from June to August during summer flooding season Low impact of Jangmaand Typhoon during rainy season Impact of El Niño, La Niña phenomena in the worldwide Precipitation 300 250 200 150 100 50 10yr Ave Rainfall 1994 Rainfall 1994 년가뭄 0 1 2 3 4 5 6 7 8 9 10 11 12 Time (monthly) Objectives To introduce our 2014-2015 drought experience in Korea To address our current efforts for drought monitoring and outlook system development
2014-2015 Drought in Korea Drought conditions SPI drought maps during April-November, 2015 Apr. May Jun. Jul. Aug. Sep. Oct. Nov. -2.0-1.5-1.0-0.5 0 0.5 1.0 1.5 2.0
Precipitation condition (Meteorological drought) Annual precipitation in 2015 is recorded only 72% (3 rd lowest year) of normal year (1307.7mm) -Precipitation in May-September is 34-62% of normal year(982.4mm) - Almost all regions has lower precipitation except for southern coast region, especially less than 60% in central regions 2015 cumulative prec. (mm) [Monthly Precipitation] - Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec. Total Monthly Precipitaion (mm) Normal Average Ratio(%) 25.1 25.5 40.5 126.0 56.5 95.5 180.3 111.2 55.1 64.3 127.8 40.2 948.2 89 73 69 153 51 62 62 42 34 134 267 169 72 [Regional Precipitation] - Gyeonggi Gangwon Chungbuk Chungnam Jeonbuk Jeonam Gyeongbuk Gyeongnam All Normal average ratio(%) Preciptation (mm) (`15.1.1~12.31) 709.8 887.4 801.3 809.3 908.5 1238.5 801.0 1235.5 948.2 Normal average ratio(%) 53 65 63 63 70 88 72 85 72 Normalaverage precipitation( mm) 1336.0 1362.3 1277.9 1280.5 1293.6 1401.5 1123.3 1460.6 1307.7 Source : 2015 abnormal climate report (Government Coordination Office, KMA)
Agricultural reservoir condition (Agricultural drought) Average reservoir storage ratio is 61.6% compared to 77.9% in normal year ('15.12.31) - Especially, Jeonbuk and Jeonam are recorded 50.3%, 57.5% respectively [Average Reservoir Storage Ratio] W ater reserve rates(% ) 100 80 60 40 20 0 Jan Feb Mar Apr May Jun Jul Time (monthly) Average Storage Ratio Normal Average Ratio Source : Agricultural Infrastructure Management(http://rims.ekr.or.kr/awminfo/report.aspx) Aug 77.9% 61.6% Sep Oct Nov Dec 365 Water reserve rates(%) Water reserve rates(%) 100 80 60 40 20 0 100 80 60 40 20 0 Jan Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Time (monthly) 75.1% 50.3% Dec [Reservoir Average Storage Ratio in Jeonbuk-Do] Feb Mar Apr May Jun Jul Aug Time (monthly) Sep Oct [Reservoir Average Storage Ratio in Jeonam-DO] 71.4% 57.5% Nov Dec 365 365
Hydrological condition of multipurpose dam(hydrological drought) 9 out of 18 multipurpose dams are caution, alert and serious condition ('15.12.31) -Especially, reservoir storage ratio of Daechungand Boryeongdam are 79% and 48% to the dam capacity - Storagerate(%) (normal year ratio) Han River(3) Nakdong River(2) Geum River(3) Seomjin River(1) Soyanggang Chungju Hoengseong Andong Imha Yongdam Daecheong Boryeong Juam 48.6 (90) 50.9 (99) 31.1 (56) 36.3 (72) 32.1 (80) 26.8 (47) 41.4 (79) 29.3 (48) 35.7 (71) Step caution caution caution caution caution caution alert serious caution Precipitation (mm) 2000 1500 1000 500 0 195 844.4mm Precipitation (mm) 2000 1500 1000 500 0 80 1020.5mm Water Level (m) 180 165 172.7m Water Level (m) 70 60 59.9m 150 50 Jan Feb Mar Apr May Jun Jul Aug Sep Time (year) Daily Precipitation(mm) Water Level Oct Nov Dec 365 Low-water Level Jan Feb Mar Apr May Jun Jul Aug Sep Time (year) Daily Precipitation(mm) Water Level Oct Nov Dec 365 Low-water Level [Soyanggang Dam ] [Boryeong Dam]
Drought causes& 2016 Drought Forecast 2015 Drought Causes Precipitation during flood season (May-September) is only 34-62% of normal year Water ratios of multipurpose and agricultural reservoir are decreased due to less precipitation Failure to secure water storage in water facilities due to 2-year continuous less precipitation (2014, 2015) Precipitation in May-July of 2014 is one-third of normal year level, 4 th lowest record since 1973 River Precipitation (mm) Normal Average Ratio(%) Han Nakdong Geum Seomjin AVE. 78.3 111.5 69.6 108.2 90.8 34.5 47.2 29.2 38.0 38.0 Precipitation (mm) Water Level (m) 2000 1500 1000 500 0 195 180 165 150 844.4mm 172.7m 2010 2011 2012 2013 2014 2015 2016 365 730 1095 1460 1825 2190 Time (year)!?!? Daily Precipitation Water Level Low-water Level Drought prospect in Spring season, 2016 Domestic, industrial and agricultural water use in 2016 Agricultural and Hydrological droughts are expected due to less reservoir water storage
Drought Management Status and Problems Drought management Agencies KMA (Korea Meteorological Administration) Provide various meteorological drought indices through KMA web site Information : Precipitation, Temperature, SPI, PDSI, PN K-Water(Korea Water Resources Corporation) Estimate and provide various hydrological drought indices Information : Dam inflow, GW level, WADI, MSWSI KRE (Korea Rural Community Corporation) Provide agriculture drought information on web Information : Reservoir storage ratio index, No-rainy day index, Normal ratio MPSS (Ministry of Public Safety and Security) Establish the social security framework through the enhancement of institutional drought protection capacity Information : Socioeconomic Drought Index Meteorological Drought Drought Mornitoring & Management Hydrological Drought Agricultural Drought Social-Economic Drought Problems Insufficient understanding of drought task and different drought information between agencies Lack of drought prediction/outlook system compare to drought monitoring system
Real-Time Drought Prediction System Development Core techniques in this system Drought Mornitoring & Outlook Coupled system of Atmosphre-Hydrology Field Operation Create long-term weather forecast information in high quality Establish Mesoscale LSM Coupled system developmnt and accuracy enhancement Drought analysis techniques Technology of land surface model in East-Asia Assessment for hydrometeorological forecast information in East-Asia Establishment of drought forecast/outlook information in East-Asia Establish hydro-meteorological and geographical information DB Develop modules providing disaster information Develop Web-based system providing real-time drought information Long-term Weather Forecast WaterbalanceAnalysis of LSM Production of Drought Information in Korea & East Asia Real-time Drought Early Warning System
Glosea5 (GS5) for generating monthly/seasonal weather prediction East Asia Processing & extraction of main variables Hydrological Prediction Model -Used as input data for LSM PRCP WSPD Data domain information Tave TMAX TMIN Grids 2808 -> 54(lon.)*52(lat.) Longitude 104.5250~149.7750 Latitude 21.275~52.0250 Glosea5 hindcast Information Glosea5 forecast Information Longitude 432 (Interval: 0.83, 92.5km) Longitude 432 (Interval : 0.83, 92.5km) Latitude 325 (Interval 0.55, approx. 49.3km) Latitude 325 (Interval : 0.55, 49.3km) Lead time 240days(1, 4, 7, 10 month), 220days(others) Lead time 230days(1, 4, 7, 10 month), 220days(others) Data form Variables Ensemble Daily Temperature, Maximum temperature, Minimum temperature, Precipitation, U10m-wind, V10m-wind Integrate4 times/ month (1, 9,17, 25 day) (3 emsembles on each starting date) Data form Variables Ensemble Daily Temperature, Maximum temperature, Minimum temperature, Precipitation, U10m-wind, V10m-wind Start model run everyday (2 emsembles on each day)
LSM Setup over Korea & East Asia Domain Land Surface Model (LSM) Use hydrologic distributed model for simulating energy and water balance based on atmosphere-vegetation-soil interaction Composed of waterbalance, channel Routing, energy balance modules Widely used for analyzing atmosphere-surface interactions on various grid scale (12.5km) Parameter Input Data Basin Forcing Soil DEM Precipitation Maximum Temperature Minimum Temperature Wind Speed Soil Properties <Construct the Mesoscale LSM > Vegetation Landuse
Local(Korea) & global data are used for LSM input data Meteorological data : Precipitation, Temperature, Wind speed Topographical data : DEM, Land coverage, Soil map Convert local (100)& global (1, 5km) resolution to LSM grid resolution(12.5km, 25km) DEM Landuse Soil Met Establish the input data of LSM DEM Landuse Soil Met
LSM model calibration/verification in East Asia regions 18 out of 34 basins are used for model verification under the assumption of ungauged basins Simulate flows at the selected basins after transferring the model parameters based on regionalization method Statistical results : CC(0.62-0.97), ME (0.30-0.95), VE(<35% ) Basin classification for transfering parameters Calibration basins Verification basins 1 10 12 7 4 5 8 18 21 16 22 19 17 26 Basins CC ME RMSE (mm/mon) Korea 0 Average (8 basins) 0.84 0.74 22.01-1.97 Russia 1 Bolshaya Bira 0.84 0.60 22.50 24.40 China North Korea Japan 4 Hongshui He 0.62 0.30 33.59 12.43 5 Laoguan He 0.97 0.95 9.09 1.37 7 Wu Jiang 0.92 0.79 30.24-14.37 8 Gan Jiang 0.65 0.40 19.68-14.27 10 Liao He 0.83 0.68 11.19-19.31 12 Sangwan 0.66 0.45 44.55-23.04 16 Chikugo 0.94 0.7 66.96-35.57 17 Oyodo 0.91 0.79 77.03-19.35 18 Gono 0.91 0.75 40.7-20.8 19 Yoshino 0.83 0.65 51.78 15.96 21 Yodo 0.86 0.68 26.11 6.42 22 Tenryu 0.89 0.64 40.92-26.39 26 Tokachi 0.82 0.62 26.83-15.62 Average 0.83 0.64 35.80-9.15 VE (%)
Development of drought outlook information system Design drought monitoring and outlook system Drought Monitoring Drought Outlook -3Mon -2Mon -1Mon Current Time +1Mon +2Mon +3Mon Observed weather and Topographical data Weather and hydrological outlook DEM Soil KMA Long-term Weather model Precipitation, Temperature, Average wind velocity Calculation of Hydrologic analysis information Land Surface Model Runoff, Evaportranspiration, Soil moisture Land Surface Model Runoff, Evaportranspiration, Soil moisture Production of Drought Monitoring SPI, PDSI, SRI, SSI, DDI..., Production of Drought Outlook SPI, PDSI, SRI, SSI, DDI..., Evaluation of Drought information
Applicability of drought indices in 2008 and 2009 (Regional analysis) Drought period Drought region Contents of evaluation 2008. 09 ~ 2009. 04 Gangwon-Do Jeolla-Do Gyeongsang-Do SPI performance is insufficient denoting drought demise on Gangwon regions in March PDSI ismissing the demise of the event in May 2009 SRI, SSI perform drought situation properly for both starting and ending of the drought event SPI(3) SPI(3), PDSI SRI(3) SRI(3), SSI(3) Drought regions PDSI SSI(3) 2008_09 2008_11 2008_01 2009_03 2009_05 2008_09 2008_11 2008_01 2009_03 2009_05
ROC analysis for drought indices (SPI, SRI, SSI) The indices are accurate in the order of SRI > SSI > SPI 1 Verifying analysis Drought Event Drought Drought Indices No Drought Drought Hit(H) Missing(M) No Drought False(F) Negative hit(n) H Hit Rate= H+ F, M False Alram Rate= M + N True Positive Rate 0.8 0.6 0.4 0.2 SRI SSI SPI 0 0 0.2 0.4 0.6 0.8 1 False Positive Rate Evaluate the performance of SRI index over the East Asia domain 2007.04 ~ 2007.12 Observation 2007.04 2007.06 2007.08 2007.10 2007.12
Bayesian method for enhancement of drought prediction accuracy Real-time Generation System of Drought Forecast Information using Bayesian Method <Bayesian method> <Model resolution>
Prior Distribution: ( ) P θ t Assumed normal distribution : Gaussian distribution is estimated from the mean and variation of monthly recipitation θ 2 t ~ N( µ ot, σot ) θ t µ ot 2 σ ot : observed precipitation for test period : mean precipitation for reference period : variance of precipitation for referenceperiod Likelihood Function : P( xt θt) Use linear regression to extract relationship between observed and predicted values µ = α + βθ V lt t n 1 1 = ( x α βθ ) t t t n 2 t= 1 β = n 1 n 1 n 1 n 1 θ x θ x α = x βθ t t t t t= 1 t= 1 t= 1 n 1 n 1 2 n 1 θt xt t= 1 t= 1 t 2 x t : Simulated precipitation for test period µ lt : Regression estimated precipitation V : Variance of regression t estimated precipitation α : Intercept coefficient β : Gradient coefficient n : Data period <Example of Likelihood function using linear regression>
Posterior Distribution: P( θ x) t t Posterior Distribution is proportional to prior distribution x likelihood function with normal distribution P( θ x) P( x θ ) P( θ ) t t t t t P ( θ µ ) 2 t ot ( θ t) = N( µ ot, σot ) = exp 1/2 2 2 (2 π) σot 2σot 1 2 1 ( xt α βθt) P( xt θt) = N( α + βθt, Vt) = exp 1/2 (2 π) Vt 2Vt 2 µ t σt Mean ( ) and variance( ) of posterior distribution σ 2 t = σ σ σ 2 2 ot lt, 2 2 ot + σlt µ = t µ σ + yσ 2 2 ot lt t ot 2 2 σot + σlt y = ( x α)/ β t t σ = V / β 2 2 lt t < application results of Posterior distribution >
Compare the drought outlook accuracy with/without Bayesian method +1m +2m +3m OBS GloSea5 Bayesian GloSea5 Bayesian GloSea5 Bayesian May Jun. Jul. Aug.
Display real-time drought monitoring & outlook information to KMA Drought information on East Asia & Korea Drought Monitoring Drought Outlook <Main page> <SPI> <PDSI> <SPI> <PDSI> <SRI> <SSI> <SRI> <SSI>
Conclusion and Remarks Introduce the 2014-2015 drought event and issued the problems for drought management in Korea Develop coupled atmosphere-land surface model for drought monitoring and outlook system over Korea and East Asia Suggest institutional cooperation techniques for better understanding and management of future drought disasters in Korea Drought Outlook KMA Drought Monitoring [KMA Homepage]` [Total Drought Information System] Observation data (ASOS, AWS, GTS) Precipitation, Temperature, Relative Humidity, Wind Speed, etc. Long-term forecast data (GS5) Precipitation, Temperature, Relative Humidity, Wind Speed, etc. Meteorological drought information Korea & East Asia domain SPI, PDSI, PN, Cumulative precipitation Hydrological information (LSM) Runoff, Soil-moisture, Evapotranspiration on Korea and East Asia Hydrological drought information Korea & East Asia domain SRI, SSI, Join Drought Index, etc. Mid-term & Numerical forecast data Climate prediction data Satellite, Ocean data [Real-time Drought Early Warning System] K-water KRCC MSWSI WADI [Drought Information System] Hydrological information Dam & Stream flow Water supply information Drought monitoring MSWSI, WADI, SMI, etc. Assessment of Hydrological water shortage Collect all drought information Produce integrated drought information [Integration D/B System] Agricultural information Reservoir storage ratio Drought monitoring RDI, IADI, etc. Assessment of Agricultural water shortage RDI Reservoir [Agricultural Drought Information System] MPSS Establish action plan considering drought characteristics & damage analysis Information releasee Civil Servant People [National Drought Hazard Information System]` Researcher Manager
Thank you