Solid-state radars MAKING modern radar software Meteopress reveals how advanced software is making solid-state radar more reliable, accurate and easy to deploy, while new AI applications are enhancing numerical weather prediction forecasts Michal Najman, CEO, Meteopress R TOP: RhoHV dual polarization data displayed in Meteopress SMRT software adar in the 21 st century must feature software-based solutions. Analog devices have now become tools for hobbyists rather than professionals. In modern equipment, software is the brain of the operation and needs to function as well as the radar hardware itself. But just like in the human brain, it must play multiple roles at the same time. handled as a standard computer program, which is easy to modify and develop. The company’s engineers have implemented several features that make its radars stand out. One of these is partial pulse correlation. Because the signal processing is done in software, the pulse compression scheme can be adjusted to see targets in the feared solid-state radar blind range. Therefore, Meteopress radars, despite being solid-state, do not have the blind range problem and don’t require ‘fill pulses’ to solve the blind range. Another feature is wi-fi packet detection and filtering. Thanks to seeing wide raw radio data, Meteopress uses a special technique to remove wi-fi interference in C-band radars and other sources of interference. This technique includes base stations in the S-band and neighboring radars in the X-band. Finally, another feature that makes the Meteopress radars stand out is the infinite Doppler. By envisioning a sophisticated signal transmission scheme, Meteopress radar can extract velocities far surpassing the classic range-Doppler dilemma. Thanks to raw data being in a simple and accessible format, the Meteopress radar is also a great tool for researchers wishing to implement new groundbreaking algorithms themselves. Modern software architecture and languages Software-based radar ABOVE: Live reflectivity radar data in SMRT software At the beginning of the radar chain, there is signal generation. Unlike the old-fashioned tube-based radars, such as magnetrons, the signal is generated in a standard PC as a piece of code. The signal is called a chirp and if it were played live on speakers, it really would sound like a bird chirping. This is when artificial intelligence comes in for the first time. The Meteopress AI team has fine-tuned the signal using advanced machine learning techniques and optimized it for the given piece of equipment and deployment location. The shape of the chirp affects the sensitivity, resolution and sidelobes. The company tailors different chirps to different situations, such as very precise and sensitive clean-air scanning or the high-resolution scanning of convective events. This is how Meteopress has overcome the supposed drawbacks of solid-state radars, such as range sidelobes and low sensitivity. Instead of using the traditional field-programmable gate arrays (FPGAs), Meteopress radars use a data acquisition card (A/D converter), which sends raw radio data to the computer. The processing of this data is then All Meteopress’s software combines only open software and in-house technologies, making the radar system fully modifiable to meet any requirement. The modern technological stack, comprising Python and C++ for data processing, a JavaScript UI interface and a WebGL visualization layer, unlocks possibilities for rich data analysis; open programming interfaces offer unlimited possibilities for customers and external developers. The software deployment options are extremely versatile, ranging from an installable desktop application and a remote web-based service to a public viewer or closed, highly secure systems. An interactive map-based environment or true live sweep data connected via real-time web sockets enable the observation of data measurements with only a fraction of a second delay. 58 • www.meteorologicaltechnologyinternational.com • April 2024