Improving the membership functions of a fuzzy hydrometeor classifier in an X-Band weather radar system : parameter adjustments and performance validation using ground-based forward scatter sensor observations
Mäkinen, Jere (2023)
2023
Teknis-luonnontieteellinen DI-ohjelma - Master's Programme in Science and Engineering
Tekniikan ja luonnontieteiden tiedekunta - Faculty of Engineering and Natural Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2023-09-19
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202309088057
https://urn.fi/URN:NBN:fi:tuni-202309088057
Tiivistelmä
A weather radar is a remote sensing device that measures rain by transmitting a high-energy electromagnetic signal into the atmosphere and receiving echoes that scatter back to the radar. The weather targets are referred to as hydrometeors which is a generic name for any water or ice particle in the atmosphere. Modern radar technology, especially the so-called dual-polarization technique, enables measuring a large scale of parameters describing hydrometeors' size, shape, and orientation. Based on this information, it is possible to classify the hydrometeors into classes such as rain, wet snow, dry snow, or hail.
This procedure is typically done using algorithms based on fuzzy logic. Fuzzy logic is an extension of classical logic. It is capable of modeling the logical "middle ground" that is not included in classical logic by allowing truth values that are something between true and false. Membership functions are part of fuzzy systems that transform the crisp input values into fuzzy truth values. This work presents the basics of fuzzy logic and how it can be utilized in solving a classification problem.
However, weather radars operate using different frequency bands in their transmitted signal. The frequency bands are denoted with letters S, C, and X, listing from the lowest frequency to the highest. The parameters of the membership functions are dependent on the used frequency band of the radar. The aim of this work is to adjust the parameters of Vaisala's fuzzy hydrometeor classification algorithm for X-band based on the old C-band specific parameters. The adjustments made in this work are based on literature references that describe the polarimetric differences of the different frequency bands and similar adjustment processes that have been carried out before for different algorithms.
The performance of the algorithm after the parameter adjustments is studied by comparing real weather data from an X-band weather radar and a ground-based forward scatter sensor. Analysis is also supported by visual and quantitative comparison of the data with the old and the adjusted parameters. All in all, five different raining events were included in the analysis.
The results of the analysis show that after the adjustment, the number of snow bins incorrectly classified as liquid rain was significantly decreased and the algorithm behavior was more consistent in detecting hail and graupel.
This procedure is typically done using algorithms based on fuzzy logic. Fuzzy logic is an extension of classical logic. It is capable of modeling the logical "middle ground" that is not included in classical logic by allowing truth values that are something between true and false. Membership functions are part of fuzzy systems that transform the crisp input values into fuzzy truth values. This work presents the basics of fuzzy logic and how it can be utilized in solving a classification problem.
However, weather radars operate using different frequency bands in their transmitted signal. The frequency bands are denoted with letters S, C, and X, listing from the lowest frequency to the highest. The parameters of the membership functions are dependent on the used frequency band of the radar. The aim of this work is to adjust the parameters of Vaisala's fuzzy hydrometeor classification algorithm for X-band based on the old C-band specific parameters. The adjustments made in this work are based on literature references that describe the polarimetric differences of the different frequency bands and similar adjustment processes that have been carried out before for different algorithms.
The performance of the algorithm after the parameter adjustments is studied by comparing real weather data from an X-band weather radar and a ground-based forward scatter sensor. Analysis is also supported by visual and quantitative comparison of the data with the old and the adjusted parameters. All in all, five different raining events were included in the analysis.
The results of the analysis show that after the adjustment, the number of snow bins incorrectly classified as liquid rain was significantly decreased and the algorithm behavior was more consistent in detecting hail and graupel.