256 Improving Quantitative Precipitation Estimation by Combining SPOL Radar and Rain Gauge Network Over Sao Paulo Metropolitan Area

Thursday, 31 August 2017
Zurich DEFG (Swissotel Chicago)
Kleber Lopes Rocha Filho, FCTH, Sao Paulo, Brazil; and F. Conde, C. P. Andrioli, and A. S. K. B. Sosnoski

Quantifying precipitation over a given area is one of the main challenges of hydrometeorology. Rain gauge network provide accurate estimative of rainfall locally, but lack spatial representativeness, problem that is increased significantly for high spatial and temporal resolution. Weather radars, in spite of their higher spatial and temporal resolution, present several sources of uncertainties. This article presents an application of conditional merging technique to combine estimations from a SPOL radar with a rain gauge network over São Paulo Metropolitan Area (SPMA). The SPMA has 6000 km² and is located in Southeastern Brazil. It is composed of 39 municipalities in which 20 million people lives. The conditional merging scheme was performed to 6 events of flood occurrences during 2016-2017 summer. It was used the 2x2 km estimations from the São Paulo’s SPOL radar and 197 rain gauges, with time resolution of 10 minutes. Inter-comparisons and cross-validation tests were made with radar estimations and merged fields, for 13 rain gauges that were not used in the merging process. The statistical scores show that merged fields decreased the RMSE from 0.83 mm (radar) to 0.67 mm (merged). The relative bias decrease from -21% (radar) to -3.4% (merged) and correlation increased from 0.64 (radar) to 0.77 (merged). The dispersion between the mean estimations and mean observed data suggest merged estimations perform better corrections for rainfall above 18 mmh-1, where underestimates using radar tends to be more significantly. Results indicate the conditional merging technique is capable to improve QPE in a dense rain gauged area, even for a high spatial and temporal resolution.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner