๐๐ฎ๐ซ ๐๐๐ญ๐ก๐จ๐๐จ๐ฅ๐จ๐ ๐ฒ:
Using raw data from over 1,800 MGM (Turkish State Meteorological Service) stations, we spatialized the rainfall onto a ~1 km high-resolution grid to calculate area-weighted sub-catchment averages. We then identified the highest rolling 24-hour rainfall for each catchment and evaluated it against a ๐๐จ๐ -๐๐๐๐ซ๐ฌ๐จ๐ง ๐๐ฒ๐ฉ๐ ๐๐๐ ๐๐ข๐ฌ๐ญ๐ซ๐ข๐๐ฎ๐ญ๐ข๐จ๐ง fitted to each catchmentโs 40-year (1981โ2020) annual maximum series. Sub-catchment return periods are scaled from these catchment-level distributions.
Here is what the data reveals across the animation frames:
1๏ธโฃ ๐๐ฎ๐ฆ๐ฎ๐ฅ๐๐ญ๐ข๐ฏ๐ ๐๐ซ๐จ๐ ๐ซ๐๐ฌ๐ฌ๐ข๐จ๐ง: How rainfall accumulated across sub-catchments over the 50-hour window.
https://www.youtube.com/shorts/oSQg2onWZ8c
2๏ธโฃ ๐๐ก๐ ๐๐ข๐ฉ๐๐ฅ๐ข๐ง๐: Our 4-step analytical workflow: Station Data โก๏ธ Gridded Interpolation โก๏ธ Sub-Catchment Aggregation โก๏ธ Return Period Estimation.
https://www.youtube.com/shorts/nbNy1gG0caM
3๏ธโฃ ๐๐๐ฆ๐ฉ๐จ๐ซ๐๐ฅ ๐๐๐ฌ๐จ๐ฅ๐ฎ๐ญ๐ข๐จ๐ง: Hourly rainfall progression (partly reconstructed from cumulative station totals) vs. the highest recorded rolling 24-hour window.
https://www.youtube.com/shorts/G0lh8LSSlUQ
4๏ธโฃ ๐๐ฉ๐๐ญ๐ข๐๐ฅ ๐๐๐ซ๐ข๐ญ๐ฒ: Final return periods mapped by sub-catchment.
๐๐๐ฒ ๐๐ฒ๐๐ซ๐จ๐ฅ๐จ๐ ๐ข๐๐๐ฅ ๐๐ง๐ฌ๐ข๐ ๐ก๐ญ๐ฌ:
๐ ฤฐzmit Bay โ ฤฐznik: 85.9 mm in 24 hours โก๏ธ ~100-year return period
๐ Lower Sakarya: 85.3 mm in 24 hours โก๏ธ ~66-year return period
๐ Kandฤฑra โ ลile Coast: 104.6 mm in 24 hours โก๏ธ ~45-year return period
๐ Karasu Sub-Catchment: Reached a striking 260 mm area-averaged event total.
๐๐ก๐ "๐๐๐๐ฌ๐ฎ๐ซ๐๐ฆ๐๐ง๐ญ ๐๐ข๐ง๐๐จ๐ฐ" ๐๐ก๐๐ง๐จ๐ฆ๐๐ง๐จ๐ง:
A critical takeaway from this analysis is how sensitive return-period estimates are to the accumulation window. Standard daily meteorological totals are counted from 09:00 to 09:00 (the climatological day), and the peak of this storm fell across that boundary, splitting it between two reporting days. In Lower Sakarya, the daily totals suggest a common ~5-year event, while the highest rolling 24-hour total reaches ~66 years. Fixed and rolling windows are not directly comparable, so consistent windows between the event and the historical series are essential for a reliable assessment of flood risk.
๐๐๐ข๐๐ง๐ญ๐ข๐๐ข๐ ๐๐จ๐ญ๐ข๐๐ & ๐๐ข๐ฆ๐ข๐ญ๐๐ญ๐ข๐จ๐ง ๐จ๐ ๐๐ข๐๐๐ข๐ฅ๐ข๐ญ๐ฒ: The statistical outputs and spatial calculations presented in this analysis are preliminary and intended solely for informational, academic, and exploratory purposes. This study utilizes raw, unvetted near-real-time observational data from automated weather stations, which has not undergone final retrospective quality control or homogenization by official meteorological authorities. Where continuous records were unavailable, part of the hourly series was reconstructed from cumulative station totals. The return periods specified herein are mathematically derived based on a localized 40-year historical baseline using the Log-Pearson Type III distribution; values beyond ~50โ100 years extrapolate this record and carry wide uncertainty, and sub-catchment values are scaled from catchment-level distributions. Alternative statistical distributions, longer historical baselines, or climate change non-stationarity adjustments may yield significantly different return period estimates. Consequently, these findings do not constitute official administrative data, formal hydrological engineering designs, or actionable financial, underwriting, or legal advice. T Rupt Teknoloji A.ล. explicitly disclaims any and all liability for commercial, engineering, or structural decisions made in reliance upon this preliminary assessment.
#FloodRisk #CatastropheModeling #Hydrology #DataAnalytics #MarmaraFlood #SpatialData #TRupt
