Importance of Hydro-Meteorological Data Bank for Use in Coupled Models and Disaster Management Using New Techniques (RS/GIS) in Turkey Prof. Dr. A. Ünal ŞORMAN Middle East Technical University (METU) Department of Civil Engineering 22 – 25 May 2004 Introduction Speech can be divided into 5 main topics: A. Importance of snow and data collection B. Hydrological models and coupling with atmospheric circulation models C. Flood forecasting from early snowmelt/rainfall in 2004 (a case study in Turkey) D. Scaling and meteorological data assimilation E. Future research activities for operational runoff forecast A. Importance of Snow and Data Collection Snow is an important resource of water Determination of SWE is important to forecast the volume of spring melt Ground truth is the main data source in investigating the snow covered areas Reflectance values from the snow surface should be watched during the snow melt period Snow studies between 1964-2002 1. Snow observations Classical methods (snow sticks, snow tubes) # of stations Electrical Res. and Survey Adm. (EİE) 1964 4 1969 6 30 1975 21 92 1997 2002 State Hydraulic Works (DSİ) 107 67 164 State Meteorologic Org. (DMİ) •Measurements are usually collected in/around urban areas. •Snow data recorded as SWE in mm. Recent Studies by DSİ and DMİ In TEFER project, 206 automated meteorological stations are under construction 3 radar stations are to be operated in western regions of Turkey Snow studies between 1964-2002 2. Snow research and modeling in basin scale Basin wide snow studies were initiated by METU, Tübitak-Bilten, EİE, DSİ and DMİ under a protocol sponsored by NATO in 1997. Snow in Eastern Turkey Snowmelt runoff constitutes approximately 60-70% of yearly total volume in Euphrates (Fırat) River, where major dams are located in series (Keban, Karakaya, Atatürk, Birecik and Karkamış). Therefore forecasting the snow potential in advance could result in better management of the country’s water resources. Automated Snow & Meteorological (Snow-Met) Stations Because of high snow potential, Karasu Basin in the Upper Euphrates is selected as a pilot basin for snow studies Karasu Basin Station Locations in Karasu Basin Station (Elevation) Data Logger Wind Speed Çat (2340 m) X X X X Hacımahmut Sakaltutan (2150 m) X Güzelyayla (2065 m) X X X X X X X X X X X X X X Global Radiation X X X X X X X X X X X X X X X X X Real Time Data Transfer X Snow Pillow Lysimeter Soil Temp. Rain Gauge Albedo Net Radiation Thermal Radiation Air Pres. X Snow Depth (1965 m) Relative Hum. (2170 m) Air Temp. Ovacık Wind Direction Station Instrumentation X X X X X X X X X X X X X X X Güzelyayla Snow-Met Station Elev: 2065 m Lat: 40o12`19`` Long: 41o28`18`` Sensors Rain Gauge Snow pillow Snow Lysimeter Güzelyayla Snow-Met Station Sensors Temperature and Relative Humidity Sensor Ultra Sonic Depth Sensor Inmarsat Antenna Wind Speed and Direction Sensor Net Radiometer Solar Radiation Sensor Güzelyayla Snow-Met Station Snow Pillow 3 meter Diameter Hyphalon Snow Pillow Güzelyayla Snow-Met Station Snow Lysimeter Snow Lysimeter measures the • amount • rate • duration of snow melt Snow-Met Station Communication Satellite Snow-met Station Data from snow-met stations are downloaded via satellite or GSM where available. METU Office Snow-Met Station Processed Data 200.000 70 180.000 60 Snow Depth 140.000 Lysimeter 50 120.000 40 100.000 80.000 60.000 Snow Water Equivalent 30 20 40.000 10 20.000 0.000 1-Mar-03 8-Mar-03 15-Mar-03 22-Mar-03 Kar Su Eşdeğeri (mm) 29-Mar-03 5-Apr-03 Date Lizametre Toplam (mm) 12-Apr-03 19-Apr-03 26-Apr-03 Kar Derinliği (cm) Snow data, 2003 water year 0 3-May-03 Derinlik (cm) KSE(mm), Lizamatre (mm) 160.000 Snow Studies Concentrate on Snow cover area monitoring – SCA Snow water equivalent analysis – SWE Snow albedo measurements - Albedo Snow Cover Area (SCA) National Oceanic and Atmospheric Administration (NOAA) Temporal Resolution: 2 or 3 times a day Spatial Resolution: 1.1 km • Supervised Classification • Unsupervised Classification • Threshold (Theta Algorithm) Snow Cover Area (SCA) 13 April 1998 Geocoded NOAA Image Snow Cover Area (SCA) Special Sensor Microwave/Imager (SSM/I) Temporal Resolution: 1 or 2 times a day Spatial Resolution: 30 km Modified Grody/Basist Algorithm, 3 April 1997 Snow Cover Area (SCA) Moderate Resolution Imaging Spectroradiometer (MODIS) Temporal Resolution: 1 or 2 times a day Spatial Resolution: 0.5 km 5 April 2004 Snow Cover Area (SCA) 13 April 1997 Supervised Class. 13 April 1997 Snow Covered Area Snow Water Equivalent (SWE) Snow Water Equivalent is the actual amount of water stored in the basin which will turn into runoff once snow melt occurs. Snow Water Equivalent (SWE) Snow pillows are used to measure continuous SWE at a point SWE data are randomly checked by snow tube measurements done by state organizations near snow-met stations Hacimahmud Hm_DSI Guzelyayla Gy_DSI Ovacik Ova_DSI Cat Cat_DSI 400 375 350 325 300 275 SWE (mm) 250 Station SWE, 2003 Water Year 225 200 175 150 125 100 75 50 25 0 01-11-02 15-11-02 29-11-02 13-12-02 27-12-02 10-01-03 24-01-03 07-02-03 21-02-03 07-03-03 21-03-03 04-04-03 18-04-03 02-05-03 Snow Albedo Albedo is a very critical parameter in snow as it determines the amount of absorbed solar energy (major energy for snowmelt) for melting process to take place, “Energy Budget”. Dry fresh snow albedo ~ 0.80-0.90 Wet dirty snow albedo ~ 0.20-0.30 Snow albedo is a function of snow grain size, depth, age, impurities… Albedometer present at Güzelyayla and Ovacık Snow-met stations Snow Albedo Daily Average Albedo, Winter of 2003-2004 Guzelyayla Ovacik 1.000 0.900 0.800 0.700 0.600 0.500 0.400 0.300 0.200 0.100 0.000 01-Nov-03 15-Nov-03 29-Nov-03 13-Dec-03 27-Dec-03 10-Jan-04 24-Jan-04 07-Feb-04 21-Feb-04 06-Mar-04 Daily average snow albedo, 2004 water year 20-Mar-04 Snow Albedo MODIS Albedo Daily and 16-day albedo values from MODIS Aqua/Terra satellite are analyzed Snow albedo variation is significant especially during snow ablation stage. Therefore, temporal variation as well as spatial variation is important Snow albedo is used in energy balance models and modified temperature index models in hydrologic modeling B1. Hydrological Models SRM (Snowmelt Runoff Model) Switzerland-USA, Temperature Index Model HBV (Hydrologiska By-rans avdeling for Vattenbalans) Sweden-Norway, Temperature Index Model SNOBAL (Snow Balance) USA, Point Two Layer Energy Balance Hydrologic Models (SRM) Qn+1 = [cSn . an (Tn + Tn) Sn + cRn . Pn] (A.10000/86400) (1-kn+1) + Qn kn+1 Snow melt Flow Recession Rainfall Parameters Variables • Snow Covered Area (S) • Temperature • Precipitation (T) (P) • Snow runoff coef. (cSn) • Rain runoff coef. (cRn) • Degree day factor (a) • Temp. lapse rate (γ) • Critic temperature (Tcrit) • Rainy area • Recession coefficient • Time lag (RCA) (k) Hydrologic Models (HBV) Model Structure Snow routine Critical Temp, Degree day, Rain/Snow correction coeff. Soil Moisture Field Capacity, Pot. Evap. Upper Zone Quick recession coeff. Lower Zone Slow recession coeff., Percolation Hydrologic Models (SNOBAL) Q = Rnet + H + LE + G + M Q: Rnet: H: LE: G: M: net energy change in snowpack (W/m2) net radiation (W/m2) sensible heat flux (W/m2) latent heat flux (W/m2) ground heat (W/m2) advection (W/m2) Near Real Time Forecasts NOAA (optic) SSM/I (passive mw) MODIS (optic) METU Web site cd/ftp Modem-Satellite Phone Hydrologic models DSI Runoff Stations GSM ftp ECMWF MM5 DMI ECMWF GRIB format Grided Binary Boundary Conditions (40x40km) Remote Sensing NOAA/AVHRR MODIS GIS High spatial elevation model MM5 (9x9km) [1.2GB] Format Conversion NCAR Non hydro static Atm. Model Forecasted Grid Data Model Variables (Temp., Prec.) Model Parameters Snow Covered Area Basin Characteristics Grid Distributed SCA P/ T Hydrological Models Forecasted Runoff Integration of Real Time Atmospheric and Hydrological Models for Runoff Forecasts in Turkey Results & Conclusions from hydrological model studies Formation of a common digital data banks Format conventions and parameter selections Enabling research oriented data sharing Installation of new hydro meteorological stations and quality increment by optimization Use of RS and GIS in basin model studies. Related software, hardware and satellite selection. Results & Conclusions from hydrological model studies Simulation and forecast studies by Lumped/Distributed (full/semi) models in {daily, monthly and yearly basis} Providing the cooperation between universities and governmental organizations Selection of projects having national priorities B2. Atmospheric – Hydrological Model Coupling Atmosphere – Circulation Atmosfer – Sirkülasyon Models Modelleri (Forecast or Analysis) GCM ve RCM ve RCM • GCM Global ECMWF – ETA • Global, ECMWF – ETA (40 km – 40 km) (40 km – 40 km) (6 saat) (6 hr)MM5 •Bölgesel • (9 Regional, MM5 km – 9 km) (9 km – 9 km) (1 saat) (1 hr) Hydrological Models Energy and Mass Balance Operational Models Models (Lumped and distributed) (Lumped and distributed) Elements of Hydrologic Cycle State and Diagnostic Parameters (Snow water equivalent, depth, snow surface temperature, Elements of net energy, melt speed, Stream flow, etc.) Model Input Flow Grid •Atmospheric Weather Prediction Geophysical Maps (Digital elevation Model, Land use, soil type, vegetal cover) (Analysis or Forecast) • NOAA / AVHRR Images (1100 m resolution) (Snow covered area, cloud, land) Physical Downscaling Point •MODIS Images •Meteorologic observations •Hydrometric flow observations (500 m Resolution) (Snow covered area, albedo) Quality Check Hydrological Model Model Integration and Outputs Atmospheric Model (Forecast/Analysis) Forecast / Analysis data Air temperature Precipitation (rain/snow) Wind Humidity Air Pressure Cloud Integration Hydrological Model (Operational / Research) State and Diagnostic Data Snow water equivalent Snow depth Snow covered area Snow temperature Melt rate Flow Energy flux Physical Downscaling of Thermodynamic Variables Thermodynamic Variables (Pressure, Temperature, Humudity) DEM Elevation greater than Model elevation? Yes No Extrapolate temperature and virtual Temperature to DEM elevation; Compute pressure via hydrostatic relation Interpolate pressure, temperature and virtual temperature to DEM elevation Derive relative humudity from temperature and pressure ECMWF DEM Turkey MM5 DEM Turkey Terrain MM5 Land Use Map ECMWF Temperature (3 May 2004) MM5 Temperature (3 May 2004) Read Interpolate Plot (RIP) Air Temperature (3 May 2004) Read Interpolate Plot (RIP) Precipitation (3 May 2004) C. Analysis of the early 2004 flood event An unexpected snowmelt event has occurred during late February and early March of 2004 in the eastern and southern parts of Turkey An analysis of the flood event is simulated using +1 day weather forecast data in a hydrological model to forecast runoff in Upper Karasu Basin (Kırkgöze Basin), where real time ground data (snow, meteorological, stream flow) are collected Hydrological Runoff Forecasting HBV Model (Temperature Index Model) Input data into HBV model from global weather forecasts (ECMWF) Daily total precipitation Daily average air temperature Forecast simulations during the period of 28 February - 7 March 2004 in Kırkgöze Basin Global Weather Forecasts - ECMWF Daily Total Precipitation (mm) of 5 May 2004 Air Temperature (oC) of 5 May 2004 12:00 Hydrological Model (HBV) Runoff Forecast Observed and Calculated Runoff for : Kırkgöze (R2=0.64) Observed 3 m /s Calculated 18 16 14 12 10 8 6 4 2 0 1-Feb-04 8-Feb-04 15-Feb-04 22-Feb-04 29-Feb-04 Observed and calculated runoff hydrographs at Kırkgöze Basin outlet, DSİ 21-01 7-Mar-04 Hydrological Model Forecast Results R2, Nash efficiency criterion, is used in HBV model to show the goodness of fit of the observed and calculated values (from - to +1.0, the higher the value the better the model fit). 2 2 Q Q Q Q O O S O R2 2 QO Q O where QO =observed runoff, QO =average runoff, QS =calculated runoff Normal values during HBV model calibrations are within the range 0.5-0.9. For this analysis, R2 is 0.64. D. Data Assimilation and Downscaling 1. Data collection, analysis and storage 2. Quality control 3. Physical downscaling of numerical weather prediction (ECMWF and/or MM5) model outputs 4. Real time forecasting of stream flow with hydrological models 5. Comparison of model outputs with observations 6. Data assimilation and renew Tools Satellites Aircrafts Balloons Meteorological Weather Stations Products Snow covered area Users Snow depth DSI, DMI Snow water equivalent EIE, KHGM Snow surface temperature Ministry of Env. and Forest Precipitation Soil moisture Downscaling 40 km 1 km Temperature and Precipitation Biases E. Future research activities for operational runoff forecast Develop/validate hydrological models and coupled model sub-components. Improve precipitation (snow/rain) and runoff processes related to spring snow accumulation/melt Conduct experiments to understand the effects of terrain data (DEM, land use, soil moisture, vegetation) Evaluate the effects of coupled model resolution on seasonal and diurnal land-surface atmosphere interactions in complex terrain regions. Develop techniques for assimilating new Remote Sensing products for MODIS / LANDSAT Develop and understand cold season precipitation including snow and frozen-ground Investigate the effects of climate change senarios for mid and long term Assess and improve runoff models in coupled form to validate streamflow estimates to be used by managers / decision makers Decrease the effects of flood and drought with water resources planning strategies THANK YOU