GAUGES RADAR SATELLITE COMBINATION Prof Eng Ezio TODINI

GAUGES – RADAR – SATELLITE COMBINATION Prof. Eng. Ezio TODINI e-mail : todini@geomin. unibo. it

Improvements in Rainfall Estimates are obtained by combining together the different available Rainfall Measurement Sources RAIN GAUGES Reliable but point measures

Improvements in Rainfall Estimates are obtained by combining together the different available Rainfall Measurement Sources RADAR Spatial but less reliable

Improvements in Rainfall Estimates are obtained by combining together the different available Rainfall Measurement Sources METEOSAT Spatial but too coarse resolution

THE MUSIC INTEGRATED SYSTEM PROTOTYPE RAINFALL INPUTS CANIntegrates: BE FROM: The MUSIC Prototype Gauges, Radar, Satellite Hydrologic & Hydraulic models and Meteorological Models Forecasts GIS and Advanced Visual User Interface

ORIGINAL TECHNIQUE TO COMBINE, IN A BAYESIAN SENSE, AREAL PRECIPITATION FIELDS (RADAR) TO POINT MEASUREMENTS OF PRECIPITATION (RAIN-GAUGES) POINT MEASUREMENTS GROUND BASED TELE- METERING RAIN- GAUGE MEASUREMENTS - accurate in a point - spatial significance decays with the distance and with the area SPATIAL MEASUREMENTS RADAR MEASUREMENTS - good spatial representation - poor quantitative estimates - biased measurements Rain-gauge measurements BLOCK KRIGING KRIGED measurements from gauges BLOCK KRIGING estimating the average field over the radar pixels and its Variance from the point rain-gauge measurements SPATIAL MEASUREMENTS over the radar pixels Radar measurements, A PRIORI estimates KALMAN FILTER SPATIAL MEASUREMENTS KALMAN FILTER finding the a posteriori estimates by combining the a priori estimates provided by the radar with the block Kriged measurements provided by the gauges, in a Bayesian framework Combination of radar estimates and gauges measurements, A POSTERIORI estimate Eliminating the BIAS and producing MINIMUM VARIANCE precipitation estimates on pixels 1

RAIN-GAUGES, RADAR AND SATELLITE COMBINATION RAIN-GAUGES Measurements of the rainfall field at different scales. combine measurements at multiple resolution. Meteorological RADAR Meteorological SATELLITE x SAT Scale y RAD Scale UPSCALING x y MODEL: SAT Scale RAD Scale DOWNSCALING The true rainfall at the upper scale can be obtained simply by summing the true rainfall at the lower scale disaggregated estimate at the RADAR scale (from the Bayesian combination) covariance of the estimation errors at the RADAR scale UP-SCALING: aggregated RADAR estimate variance of the estimation errors of the aggregated RADAR estimate

BAYESIAN APPROACH at SATELLITE Scale TRUE BK-RAD SAT BK-RADSAT

BAYEISAN APPROACH at RADAR scale TRUE BK RAD BK-RAD-SAT

RESULTS (1000 time-steps) BIAS VARIANCE RADAR 5. 0449 19. 4154 BLOCK KRIGING -0. 0051 30. 5724 BLOCK KRIGING + RADAR (RADAR scale) -0. 0129 14. 5419 999. 7988 34475. 6368 BK+RADAR AGGREGATED (SATELLITE scale) -1. 1297 27930. 3144 BK+RADAR + SATELLITE (SATELLITE scale) -0. 7325 16312. 8517 BK+RADAR + SATELLITE (RADAR scale) -0. 0050 13. 1567 SATELLITE

BLOCK KRIGING RADAR BK+RADAR 0. 0 mm SATELLITE BK+SATELLITE 2. 2 mm BK+RADAR+SATELLITE 4. 4 mm 6. 5 mm

The present status: - Block-Kriging software package Completed(*) - Raingauge – Radar combination Completed - Raingauge – Satellite combination Completed - Raingauge-Radar-Satellite comb. Completed - Coupling with TOPKAPI (*) Under revision Completed