The use of remote sensing technologies from the ground and space together with past data and the use of machine learning are necessary for research to predict extreme weather events
Flash floods are rapid flooding events and with high intensity which are mainly caused by heavy rains. Since flash floods have a short response time of several hours, they are problematic to predict and can cause heavy damage to agricultural areas and sometimes even loss of life.
A team of researchers from Ariel University and M&P Mizrah, tested whether it is possible to predict floods and extreme weather events up to 24 hours before they occur, by learning the changes in the amount of water vapor in the air. The results are encouraging.
בין The causes of flash floods, such as soil saturation or surface coverage, the distribution in space and time of the precipitation is the most significant based on the analysis of the results from hydrological models.
The rainfall regime in the arid and semi-arid regions of the eastern Mediterranean region is highly variable, and consists mostly of short, high-intensity events.
Therefore, in order to predict flood events, the location and timing of the heavy rainfall events that can be monitored using remote sensing platforms (such as weather radar) must be considered first. Another option is to measure the amount of water vapor in the atmosphere In order to identify significant amounts of moisture that are a necessary condition for heavy rainfall events. One of the reliable methods for estimating the amount of water vapor in the air is by monitoring signals from global navigation systems. By monitoring these signals received on the ground, it is possible to derive the amount of delay that the signal has during its passage through the troposphere and convert it with great accuracy to the amount of water vapor in the air using the pressure and temperature data on the ground. Using the navigation satellite networks, can continuously provide an almost real-time estimate of the water vapor columns in the air above the location of the receivers placed on the ground and thus produce a tremendous spatial coverage.
in the last 30 years, the amount of water vapor in the air derived from the global global navigation system (GNSS) is thoroughly tested against many other remote sensing platforms, direct measurements and reanalysis products with accuracy in the estimated values (1-3 mm). Furthermore, the water vapor maps can also be integrated into the modern weather forecast models, thereby reducing the errors in water vapor estimation by more than 30% compared to direct measurements (radiosonde).
The scientific use of remote sensing technologies from the ground and space is necessary For research on extreme weather events. The ability to predict when and where an extreme weather event such as a flood will occur in a certain area still remains challenging in the research field of monitoring, forecasting and preparedness for extreme weather events. While most of the forces and basic mechanics related to these events can be assimilated into physical and numerical models, the lack of suitable data from real-time measurements makes it difficult to build accurate predictions of extreme events. Due to this, more and more scientists are currently investigating the option of using multivariate analysis methods from the fields of data mining and machine learning in order to develop an ability to assess the occurrence of future extreme events based on patterns in past events.
"In this article, explains Dr. Yuval Reuvani From Ariel University and Mizrah R&D, we were able to predict, given 24 hours of water vapor data in the air, whether a flash flood will occur or not, with an accuracy of over 90%. We presented the real potential of using the values of water vapor in the air calculated from 9 ground GPS stations in the arid part of the eastern Mediterranean region, in order to predict flash floods. The different models tested were tested for a large number of grade indices, and even managed to achieve better results when additional measurements such as the pressure in the ground were added to the learning process. An in-depth analysis of the importance of the various measurements shows that the most important characteristic is the change in the water vapor values in the air between 6 and XNUMX hours before the occurrence of a flash flood. These promising results lay the groundwork for a real-time prediction and warning system that can save lives and prevent heavy damage to property."
The research was conducted under the leadership of Dr. Yuval Reuvani A researcher and lecturer from the physics department at Ariel University and a senior researcher at Mizrah R&D, and the postdoctoral student Dr. Shlomi Ziskin Ziv from the physics department at Ariel University and a senior researcher at Mizrah R&D. Also, the research was funded by the Ministry of Science.
More of the topic in Hayadan:
