Reply on RC1

Published: March 21, 2024

Abstract. Fast urban pluvial flood models are necessary for a range of applications, such as near real-time nowcasting or processing large rainfall ensembles uncertainty analysis. Data-driven can help overcome the long computational time traditional simulation models, and state-of-the-art have shown promising accuracy. Yet lack generalizability data-driven to both terrain events still limits their application. These usually adopt patch-based framework multiple bottlenecks, data availability memory constraints. However, this approach does not incorporate contextual information surrounding small image patch (typically 256 m x m). We propose new deep-learning model that maintains high-resolution local incorporates larger context increase visual field with aim enhancing models. trained tested in city Zurich (Switzerland), at spatial resolution 1 m, 1-hour 5 min temporal resolution. demonstrate our faithfully represent depths wide events, peak intensities ranging from 42.5 mm h^-1 161.4 h^-1. Then, we assessed model’s distinct settings, namely Luzern (Switzerland) Singapore. The accurately identifies locations water accumulation, which constitutes an improvement compared other Using transfer learning, was successfully retrained cities, requiring only single event adapt terrains while preserving adaptability across diverse conditions. Our results indicate by incorporating into patches, proposed effectively generates maps, demonstrating applicability varied events.

Language: Английский

---

Mapping the Flood Vulnerability of Residential Structures: Cases from The Netherlands, Puerto Rico, and the United States

Geosciences, Journal Year: 2024, Volume and Issue: 14(4), P. 109 - 109

Published: April 19, 2024

Floods are consistently ranked as the most financially devastating natural disasters worldwide. Recent flood events in Netherlands, Caribbean, and US have drawn attention to risks resulting from pluvial fluvial sources. Despite shared experiences with flooding, these regions employ distinct approaches management strategies due differences governance scale—offering a three-site case study comparison. A key, yet often lacking, factor for risk damage assessments at parcel level is building elevation compared elevation. First-floor elevations (FFEs) critical element vulnerability of flooding. US-based insurance policies require FFEs; however, data availability limitations exist. Drone-based FFEs were measured all locations assess vulnerabilities structures. Flood profiles revealed 64% buildings vulnerable form inundation, 40% belonging “moderate” or “major” inundation means (IEMs) −0.55 m, 0.19 0.71 m within US, Puerto Rico sites, respectively. Spatial statistics more responsible site while topography was Netherlands sites. Additional findings reveal next highest floor (NHFEs) future sea rise (SLR) elevations. The provide support developing novel multi-layered reduction that include We discuss work recommendations how different sites could benefit significantly strengthening FFE requirements.

Language: Английский

---

Comment on hess-2024-63

Published: May 11, 2024

Abstract. Fast urban pluvial flood models are necessary for a range of applications, such as near real-time nowcasting or processing large rainfall ensembles uncertainty analysis. Data-driven can help overcome the long computational time traditional simulation models, and state-of-the-art have shown promising accuracy. Yet lack generalizability data-driven to both terrain events still limits their application. These usually adopt patch-based framework multiple bottlenecks, data availability memory constraints. However, this approach does not incorporate contextual information surrounding small image patch (typically 256 m x m). We propose new deep-learning model that maintains high-resolution local incorporates larger context increase visual field with aim enhancing models. trained tested in city Zurich (Switzerland), at spatial resolution 1 m, 1-hour 5 min temporal resolution. demonstrate our faithfully represent depths wide events, peak intensities ranging from 42.5 mm h^-1 161.4 h^-1. Then, we assessed model’s distinct settings, namely Luzern (Switzerland) Singapore. The accurately identifies locations water accumulation, which constitutes an improvement compared other Using transfer learning, was successfully retrained cities, requiring only single event adapt terrains while preserving adaptability across diverse conditions. Our results indicate by incorporating into patches, proposed effectively generates maps, demonstrating applicability varied events.

Language: Английский

---

Reply on RC2

Published: May 14, 2024

Abstract. Fast urban pluvial flood models are necessary for a range of applications, such as near real-time nowcasting or processing large rainfall ensembles uncertainty analysis. Data-driven can help overcome the long computational time traditional simulation models, and state-of-the-art have shown promising accuracy. Yet lack generalizability data-driven to both terrain events still limits their application. These usually adopt patch-based framework multiple bottlenecks, data availability memory constraints. However, this approach does not incorporate contextual information surrounding small image patch (typically 256 m x m). We propose new deep-learning model that maintains high-resolution local incorporates larger context increase visual field with aim enhancing models. trained tested in city Zurich (Switzerland), at spatial resolution 1 m, 1-hour 5 min temporal resolution. demonstrate our faithfully represent depths wide events, peak intensities ranging from 42.5 mm h^-1 161.4 h^-1. Then, we assessed model’s distinct settings, namely Luzern (Switzerland) Singapore. The accurately identifies locations water accumulation, which constitutes an improvement compared other Using transfer learning, was successfully retrained cities, requiring only single event adapt terrains while preserving adaptability across diverse conditions. Our results indicate by incorporating into patches, proposed effectively generates maps, demonstrating applicability varied events.

Language: Английский

---

Sustainable Management in River Valleys, Promoting Water Retention—The Opinion of Residents of South-Eastern Poland

Sustainability, Journal Year: 2024, Volume and Issue: 16(11), P. 4648 - 4648

Published: May 30, 2024

Sustainable development is implemented not only at the global level, but primarily in local environments. Shaping space of river valleys becomes particularly important face climate change and growing water deficit. The article therefore addresses issue social perception management context change. aim was to answer questions: what awareness change, sustainable solutions are socially accepted? research carried out south-eastern part Poland, Podkarpackie Lublin voivodeships. diagnostic survey method, an original form, CAWI technique were used. study group analyzed global, negative megatrends, challenges related retention task identify respondents’ new methods valleys. Due fact that studied area largely agricultural, differences items sought, depending on place residence. It assumed inhabitants rural areas have greater contact with nature, which may their perception, looked for region Differences perceptions phenomena also searched for, respondent’s sex. calculations show residence (urban–rural) regions (Podkarpackie–Lublin voivodeships) do differentiate most examined items. However, sex affects megatrends results indicate lack about natural forms retention. Respondents expected implementation outdated technical flood protection. Expectations focused mainly embankments large dam reservoirs. There strong belief among respondents regarding impacts economic life. A knowledge deficit identified relation favor

Language: Английский

---