Operational quantitative precipitation estimation using radar, gauge g and satellite for hydrometeorological applications in Southern Brazil Leonardo Calvetti¹, Cesar Beneti¹, Diogo Stringari¹, i¹ Alex Conselvan de Oliveira¹ and Augusto José Pereira Filho² ¹SIMEPAR Technological Institute - Parana - Brazil ²University of Sao Paulo - Sao Paulo- Brazil Chinese Meteorological Agency Beijing, China 13-17 October,2008 1
Ab t us About SIMEPAR is Parana state Meteorological Service, located in Southern Brazil. Hydrometeorological Monitoring Network 40 automatic weather stations; 42 automatic hydrological stations (river level and raingauge); 20 automatic hydrological stations in Curitiba metropolitan area; 25 automatic weather stations from INMET (National Institute of Meteorology). 200 km And also: S-Band, Doppler weather radar with a coverage of 200km range (blue circle) and 480km range; Lightning Detection Network The Iguaçu Basin (red contour) Area: 67,446 Km²; C iti l streamflow Critical t fl llevel: l 1300 m3s-11 Winter: 300 m3s-1 W t period Wet i d ~ 500-700 500 700 m3s-11 Precipitation Estimates and Hydrometeorological Forecasting System - 4th IPWG Workshop 2
Hydrometeorological Forecasting: I t Integrated t d Precipitation P i it ti System S t Integrated Information: Radar + Satellite + raingauge data The Goal: To improve the precipitation estimates over the sub-basins in the Iguaçu River using radar, satellite and raingauge data integrated, integrated as well as to make use of a global and mesoscale numerical model for performing the quantitative precipitation forecast. The Radar estimates: Marshall Marshall-Palmer Palmer ZR relation applied on 3km-CAPPI products. There are images every 5 minutes. After applying the ZR relation the data are accumulated at each hour and the advection correction is perfomed. The Satellite estimates: The precipitation data of the following techniques are used: CST (Convective-Stratiform (Convective Stratiform Technique) (Adler and Negri, 1988), CMORPH (Joyce et. al., 2004), PERSIANN (Sorooshian et. al., 2000), TRMM (Huffman et. et al., al 1997) and NRLB (Turk et. al., 2003). They are in 0.25 of grid resolution at each 1 or 3 hours. The Surface Stations: Hourly and 15 minutes data are obtained from the hydrometerological stations distributed over Parana state. These informations passes a quality control system. Precipitation Estimates and Hydrometeorological Forecasting System - 4th IPWG Workshop 3
Hydrometeorological l Forecasting: Integrated t Precipitation it ti System Table 1. 2x2 Contingency table Observation Wet Dry Wet a b Dry c d Sa atellite Where: Wet: prec 0.3mm/h Dry : prec < 0.3mm/h BIAS= =(a+b)/(a+c) FAR=b/(a+b) POD=a/(a+c) The validation of the satellite precipitation estimates is done using the data provided by the hydrometeorological surface stations. Based on a (2x2) table of contingency (table 1) some indexes (left) which measure the accuracy of the techniques are calculated. After the validation the estimates are integrated with the radar and raingauge information. CMORPH RADAR Source: SIMEPAR/CPC Wet Dry Observation Wet Dry 164 4 24 94 CM MORPH BIAS=0.9 COR=+0.72 POD=0.88 MAE=-3.5 35 FAR=0.02 aave=2.3mm/h SURFACE STATIONS Precipitation Estimates and Hydrometeorological Forecasting System - 4
Hydrometeorological Forecasting: Numerical Modeling WRF is used for performing the quantitative precipitation p forecasting (QPF). Global model (GFS) is used for boundary conditions (50Km of horizontal resolution) with 10 km horizontal grid, a nested grid of 2.5km and 32 vertical levels. The parameterizations are: cumulus Kain- Fritsch (turned off for nested grid), microphysics of Pardue & Lin, Surface-Soil Model: Noah LSM, Planetary Boundary Layer: YSU (Yonsei University), Long Wave Radiation: RRTM (Rapid Radiative Transfer Model), Short Wave Radiation: Dudhia(MM5). WRF (Weather Research and Forecasting) Model The simulations are performed with 00UTC initial conditions and the forecasting is up to 48 hours. The evaluation of the QPF is done for the 9 sub-basins and for different categories of rainfall amounts as defined d by Calvetti et. al., (2006) (table 2). Table 2.Category of rainfall amount Category No Rainfall Drizzle Weak Rainfall Moderate Rainfall Strong Rainfall Extreme Rainfall Precipitation - mm 0-0,2 0,2 2,5 2,5 10 10 25 25-50 Above 50 Precipitation Estimates and Hydrometeorological Forecasting System - 5
RESULTS 1.600 0 1.500 5 1.400 10 1.300 15 1.200 20 1.100 1.000 900 800 700 9/5/2007 10/5/2007 11/5/2007 12/5/2007 13/5/2007 14/5/2007 15/5/2007 16/5/2007 17/5/2007 7 18/5/2007 19/5/2007 20/5/2007 Precipitação WRF Precipitação Vazão WRF Vazão Figure 1. Streamflow forecast by TopModel using QPF from WRF predictions (red line), streamflow observed (blue line), 24h forecast of rain by WRF (orange bars) and average precipitation observed by raingauge network on União da Vitória watershed (blue bars). The QPF was performed for the 9 sub-basins. On average, WRF skill score is around 75%. WRF has improved the performance in 10% when compared to GFS. The larger spatial resolution might facilitate the better simulation of the atmospheric environment as the microphysics and the convective processes. References ADLER RF, AJ NEGRI. 1988. A Satellite Infrared Technique to Estimate Tropical Convective and Stratiform Rainfall. Journal of Applied Meteorology: Vol. 27, No. 1, pp. 30 51. CALVETTI L, C BENETI, AC OLIVEIRA, AJ PEREIRA FILHO, D STRINGARI. 2008. Quantitative Precipitation Forecasting applied in a hydrologic model for streamflow predictions in Iguaçu River, southern Brazil. International Symposium on Weather Radar and Hydrology, 2008, Grenoble. Proceedings of the International Symposium on Weather Radar and Hydrology. 25 30 35 40 45 Table 3. Categorical Skill and False Alarm for WRF and GFS QPF in 9 sub-basins in the Iguaçu River for 24 hours. The indexes were obtained from comparison with raingauges average in each sub-basin during September of 2005 and April of 2006. B1 B2 B3 B4 B5 B6 B7 B8 B9 Skill Score GFS 0,54 0,46 0,49 0,63 0,65 0,59 056 0,56 0,6 0,67 WRF 0,66 0,7 0,77 0,75 0,8 0,77 078 0,78 0,73 0,78 False Alarn GFS 0,24 0,23 0,22 0,19 0,18 0,20 021 0,21 0,20 0,18 WRF 0,21 0,19 0,16 0,15 0,12 0,15 015 0,15 0,16 0,13 (a) (c) Figure 2. Monthly precipitation estimates for January/2007. It is presented in (a) the interpolated data from surface stations, in (b) the CST satellite technique, in (c) the CMORPH satellite technique and in (d) the integrated information, including the radar data. Figure 2 presents the satellite precipitation estimates, surface station data and the integrated product for January/2007 over Parana state. The precipitation amount was above the average and caused severe impacts on the streamflow level. In the integrated product the radar and satellite informations improved the spatial distribution of the rainfall while the best quantitative accuracy is garanteed by the raingauge data. HUFFMAN GJ, RF ADLER, M MORRISSEY, DT BOLVIN, S CURTIS, R JOYCE, B MCGAVOCK, J SUSSKIND. 2001. Global precipitation at one-degree daily resolution from multi-satellite observations. J. Hydrometeor., 2 (1), 36-50. JOYCE R, JE JANOWIAK, PA ARKIN, P XIE. 2004. CMORPH: a method that produces global precipitation p estimates from passive microwave and infrared data at high spatial and temporal resolution. J. Hydrometeor., 5, 487-503. Precipitation Estimates and Hydrometeorological Forecasting System - (b) (d) SOROOSHIAN S, K HSU, X GAO, HV GUPTA, B IMAN, D BRAITHWAITE. 2000. Evaluation of PERSIANN System satellite-based estimates of tropical rainfall. Bull. Amer. Meteor. Soc., 81, 2035-2046. TURK FJ, E EBERT, H-J OH, B-J SHON, V LEVIZZANI, EA SMITH, R FERRARO. 2003: Validation of an operational global precipitation analysis at short times scales. Proc. 12 th Conf. Satellite Meteor. and Ocean., Amer. Meteorl. Soc., USA. 6