Weather Radar Network Benefit Model for Flash Flood Casualty Reduction
John Y. N. Chi and James M. Kurdzo
Summary
Statistical data shows that a better radar coverage improves flash flood warning performance, and thus reduce casualty rate. Given this relationship, a weather radar network benefit model was constructed in line with flash flood casualty reduction. The monetized geospatial model was applied to the contiguous United States (CONUS) weather radar and resulted a flash flood benefit of $316 million per year. The remaining benefit pool yielded $13 million per year and including better radar coverage and flash flood warning improvements, the maximum benefit pool is $69 million per year. These indicates the model's effectivity with regards to flash flood warning decision process.
Problem and Statement of Purpose
Weather radars are useful especially in meteorology because they provide observational data for better weather forecasts. They play an important role for depicting weather for real-time decision making. A benefit model for tornadoes was previously published so in this paper, they focused on heavy-rain induced flash floods. It was assumed that a better radar coverage will also improve warning performance and a better warning performance in turn, will reduce casualty rate. The purpose of this model is to monetize its benefit in relation to its effect on casualty reduction. In the given chain from weather radar coverage to casualty rate, how the flash flood warning decisions are made must be examined, and how the flash flood warning affects casualty rate with the monetization follows.
Review of Related Literature
A benefit model for tornadoes was previously published and it was the study that influenced this benefit model for flash floods. The tornado benefit model was also monetized and it was primarily developed for weather radar network configurations. Results show that two key radar coverage parameters such as the fraction of vertical space observed and cross-radial horizontal resolution led to better tornado warning performance, which are also used for the flash flood benefit model. Similarly, better tornado warning performance reduced casualty rate just like what is hypothesized in the flash flood study. The model was also ran in the contiguous United States (CONUS) weather radar network for which it provided a benefit of $535 million per year, with a remaining benefit pool of $676 million per year.
Hypothesis
The hypothesis for this study is that better weather radar coverage improves flash flood warning performance, which in turn reduces casualties. This can because tested through the benefit model and the results are monetized.
Data and Analysis
In the given chain from weather radar coverage to casualty rate, how the flash flood warning decisions are made were first explained. In the US, flash flood warning decisions mainly rely on the concept of flash flood guidance (FFG). The FFG outputs the rainfall accumulation needed in a certain span of time to cause flash flood in a given area. Weather forecast offices (WFO) use different types of FFG model, but regardless of type, what the forecasters basically look for is the accumulated quantitative precipitation (QPE) to exceed the FFG threshold in a given catchment basin when issuing a flash flood warning. Flash floods occur within 6 hours of the causative event, so when the cause of flash flood is heavy rain, in order for WFO to issue a timely warning, forecasters mostly utilize multi-sensor precipitation estimator (MPE) products for comparison with the FFG thresholds. MPE provides radar, rain gauge, and geostationary satellite data. The dominant MPE contributor is radar QPE while satellite QPE is used to fill the gaps in radar coverage. The warning process therefore primarily depends on the FFG and MPE data. The FFG threshold errors are dependent on FFG type which are specific in each catchment basin. For MPE errors, sources include choice of algorithm, radar calibration and rain gauge density. In this study, they assumed that flash flood warning performance will directly depend on radar coverage even without considering other error sources to simplify the analysis. To yield statistically meaningful results, they relied on large datasets and statistically insignificant variables were excluded. The geographic scope was limited to the contiguous United States (CONUS) where most of the needed data were available.
Data as many as possible were needed but they were still careful to avoid biased results. The primary data source was the US Flash Flood Observation Database by the Flooded Locations and Simulated Hydrographs (FLASH). The analysis period started from 1 October 2007 up to 31 December 2018 and only reports with "heavy rain" causes were included. The storm warning data for the said period was obtained from IOWA Environmental Mesonet NWS Watch/Warning Archive. A warning was considered a hit if any part of the flood polygon was inside the warning polygon; otherwise, it was labeled false alarm.
Another paper that I reviewed/critiqued. Hope you learn something. Thank you. :)
Note: I noticed this was too long and I myself, gets tired before even reading it so I decided to cut it and put it in another narrative as part 2. Btw, yeah I sometimes re-read my entries here that's why. :)
