Operational Quantitative Precipitation Estimation Algorithm in Southern Brazil - An Update Blending Dual Polarization Weather Radar Network with Raingauges and Satellite Data

Quantitative estimation of precipitation (QPE) of high resolution, accurate and in real-time, increases the potential of weather radars for many applications, such as flash flood forecasting and hydropower production and distribution management. Using polarimetric variables from dual-polarization weather radars has already shown significant improvements in quantitative precipitation estimation in many countries with diverse weather. In Brazil, in the past ten years, we have seen an increase in dual-polarization weather radar coverage, mostly S-Band and some X-Band, concentrated in the southern parts of the country, an area prone to severe weather with high precipitation and lightning due to mesoscale convective systems. This region's significant economic activity is agriculture and energy production, accounting for more than 33% of the hydro energy generation used in the country. Therefore, the improvement of precipitation estimation is a necessary goal. However, the use of weather radar's QPE depends on calibration, good fit with rain gauges and distrometers, good data filtering, target’s distance from the radar, orography (i.e., relative to the topography), and signal propagation, as well as other factors.  A multi-sensor integration approach of remotely sensed precipitation estimation using weather satellites and weather radar with rain gauges improves the accuracy of hydrological models compared to a model using only rain gauge data. A quantitative precipitation estimation algorithm called SIPREC (System for Integrated PRECipitation) has been used operationally for more than 15 years, combining data from different sources, such as weather radar, rain gauge, and satellite. Precipitation estimates are obtained through an automated precipitation classification scheme based on reflectivity structures. This approach aggregates data from rain gauges by interpolation while maintaining the spatial distribution of the radar or satellite measurement. Statistical results indicate that the method can reduce radar and satellite data errors. This method is an essential advantage in an operational environment since it does not require frequent processing to update the weights as in other known schemes. However, this approach does not solve problems such as uncertainties related to Z-R estimation, spatial variability, and the one-hour temporal resolution. To improve the SIPREC algorithm, we used machine learning classification and regression methods to address the problem of precipitation estimation using dual polarization variables and rain gauge. An enhanced satellite precipitation estimation using GOES-16 data also replaced the previous dataset, and a new quality control algorithm for the network of weather radars was also applied to the dataset. A performance evaluation study shows improvements in precipitation estimation, primarily when used in real-time in an operational environment. This paper presents the results of this evaluation, with applications in severe weather events with high precipitation in the area.