National and Kapodistrian University of Athens
Improving satellite precipitation estimates over complex terrain through the use of high-resolution numerical weather predictions
Abstract
dc:descriptionPrecipitation consists a governing parameter in a wide spectrum of human- related aspects. Either its lack or excess can pose major consequences and threaten both human lives and infrastructures. The uncertainty in its accurate monitoring and forecasting has tremendous implications in agriculture, transport, water resource management and green energy production.The main focus of precipitation uncertainty lies upon heavy precipitation events (HPEs) and flash-floods. Both of these events develop over small spatiotemporal scales, an attribute that renders their identification and predictability to particularly challenging tasks. This becomes especially evident over high mountainous terrain and complex terrain areas, where scant availability of in-situ observations is the default. The serious threat these phenomena pose for human lives along with corresponding damages that have repeatedly risen to billion dollar estimates, urge towards a better analysis and more thorough understanding of their nature, that will eventually lead to a more accurate monitoring and forecasting.Given that each individual source of information comes associated with its own list of strengths and limitations, a synergetic approach is often dictated. Satellite rainfall estimates are particularly useful in this context, due to their global coverage and near real-time availability. Observations from rain gauge networks and weather radar scans are usually assigned with the role of ground truth where available. Numerical weather predictions (NWP) have gradually become the epicenter of early warning systems, with recent advances in computational science partly curing the downsides of earlier ages. This improvement is partly attributed to the feasibility of atmospheric simulations at very fine grid scales, that deems important for capturing the initiation and evolution of flash-flood triggering storms. The advances in cloud microphysical schemes can additionally offer a valuable insight into the complex processes of these severe incidents.This thesis proposes a combination of the respective strengths of the individual datasets in order to improve precipitation estimates over mountainous areas. It provides evidence on the relative gain that can be expected from the adoption of very fine NWP model grids in HPE and flash-flood forecasting that stretch to sub-km scales. The relative impact of atmospheric model resolution on the hydrologic predictions is also examined. State-of-the-art NWP simulations are utilized towards the reduction of quantitative errors in satellite precipitation estimates and an insight on the nature of the detection limitations is provided. A novel adjustment technique is presented, where satellite rainfall estimates cater for the spatiotemporal evolution of each storm, whereas their rainfall magnitude is adjusted upon the NWP estimates through probability density function (PDF) matching. The study takes place in three mountainous areas of different characteristics, namely two mid-latitude (Italian Alps and US Rockies) and a sub-tropical (Ethiopian Blue Nile). A state-of-the-art NWP model (RAMS/ICLAMS), a hydrologic model (CREST) and a Lagrangian particle dispersion model (HYPACT) are deployed along with two hi-resolution satellite precipitation products with distinct retrieval characteristics. The evaluation is performed against dense gauge networks and radar rainfall estimates. A microphysical investigation is presented for occurrences of insufficient precipitation detection from the satellite instruments.Results showcase a significant improvement in NWP-adjusted satellite products compared to the raw estimates. The microphysical similarities amongst cases of insufficient satellite detection are identified and thoroughly discussed. The merit of this study lies upon probing the field for future blended rainfall products. The algorithm development of passive microwave satellite retrieval can be significantly benefited from the latter. Finally, the application of the method on a pseudo- operational basis during a historic flash flood, provides an estimate of the operational feasibility and potential benefit from its adoption in early warning and flash flood guidance systems.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2018
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Bartsotas, Nikolaos
- Μπαρτσώτας, Νικόλαος
Subjects
dc:subject × 14Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/43162