Skip to Main content Skip to Navigation
Theses

Estimation et prévision immédiate des précipitations sur un bassin urbain

Abstract : Good quality rainfall estimations and nowcasts are an essential prerequisite for the development of reliable flash flood warning systems, especially for urban catchments, where the socioeconomic consequences of hazardous precipitation events are high. The risks posed by such extreme events are further heightened because of climate change. In this context, this thesis aims to investigate the potential of using a small weather radar in combination with the local rain gauge and national radar networks to improve the quantitative precipitation estimation (QPE) of rainfall, and to deliver reliable nowcasts. This research was carried out in the flood prone urban catchment of Clermont Ferrand-Riom.To improve QPEs with a high spatiotemporal resolution (5 minutes and a 100 m), the performance of geostatistical interpolation techniques has been investigated using rain gauge data as a primary variable and X-band radar data as a secondary variable for the kriging with an external drift and conditional merging techniques. Radar data was used for the inference of climatological variograms for each precipitation type (stratiform, convective or mixed) for all geostatistical interpolation techniques including ordinary kriging. The long-term evaluation of these techniques shows the benefit of using the geostatistical approach to merge rain gauge and radar data, especially to capture the spatial variability of rainfall. Additionally, two methods were examined to combine the X-band LAWR (Local Area Weather Radar) data with the PATNTHERE product (Rainfall sums with a resolution of 5 minutes and 1 km, provided by the national weather service Météo-France). The first method uses the PANTHERE product (using mainly a C-band radar over the area of interest) to correct X-band data from attenuation effects, and the second one consists of applying a quantile-quantile correction to the X-band data using the PANTHERE product to take advantage of its overall better measurement accuracy, both methods have shown satisfactory results in terms of reducing bias of X-band radar data in comparison with rain gauge data.In the second section of this research, a feature-based forecasting method has been applied to two rainfall events in order to investigate the ability of the X-band LAWR of providing reliable nowcasts. The method includes several steps aimed at identifying, tracking, and then interpolating the features of rainfall cells such as area, speed, and average precipitation intensity. The application of this method on two case studies shows that it provided satisfactory results for forecast lead times up to 30 minutes but efficiency degrades for further time frames.In conclusion, the research carried out in this thesis indicates that local X-band LAWR have great potential for rain estimates and forecast and should be considered in the development of flash flood warning systems, especially for urban catchments.
Document type :
Theses
Complete list of metadata

https://tel.archives-ouvertes.fr/tel-03191287
Contributor : Abes Star :  Contact
Submitted on : Wednesday, April 7, 2021 - 9:06:32 AM
Last modification on : Thursday, April 8, 2021 - 3:40:04 AM

File

2020CLFAC055_SECK.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-03191287, version 1

Collections

Citation

Ibrahim Seck. Estimation et prévision immédiate des précipitations sur un bassin urbain. Sciences de la Terre. Université Clermont Auvergne, 2020. Français. ⟨NNT : 2020CLFAC055⟩. ⟨tel-03191287⟩

Share

Metrics

Record views

74

Files downloads

48