Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Grèce → Drought spatiotemporal analysis, modelling and forecasting in Pinios river basin of Thessaly, Greece

University of Thessaly (UTH) 2010

Drought spatiotemporal analysis, modelling and forecasting in Pinios river basin of Thessaly, Greece

Vasiliades, Lampros

Titre : Drought spatiotemporal analysis, modelling and forecasting in Pinios river basin of Thessaly, Greece

Χωροχρονική ανάλυση, προσομοίωση και πρόγνωση ξηρασίας στην υδρολογική λεκάνη Πηνειού ποταμού Θεσσαλίας

Auteur : Vasiliades, Lampros

Etablissement de soutenance : University of Thessaly (UTH)

Grade : Doctor of Philosophy (PhD) 2010

Résumé
In this study an integrated methodological and systematic approach is developed for spatiotemporal forecasting and monitoring of drought in Pinios river basin, Thessaly, Greece. The methodology is based on the integration of artificial intelligence techniques with geostatistical interpolation methods. The distributed raster-based integrated drought monitoring and forecasting system is demonstrated for operational drought management using the meteorological drought index the Standardized Precipitation Index (SPI). The basic elements of the system are determined, and important issues and processes are analytically discussed, such as the representation of the spatial and temporal simulation error, extraction of optimal design solution in machine learning methods for reliable and accurate spatiotemporal drought forecasting. Each one of the aforementioned elements is considered as cores, where integrated algorithms are applied that combine various computational intelligence techniques such as artificial neural networks, genetic algorithms and geostatistical methods. The hybrid spatiotemporal forecasting scheme is developed with the integration of 1) a generalised neural network time series model to capture the temporal characteristics of each spatially separated location, 2) a spatial artificial neural network to discover the hidden spatial and temporal correlation among all locations using the temporal forecasts, and 3) the combination of spatial interpolation techniques with the individual temporal and spatial forecasts for distributed integrated spatiotemporal forecasting of drought. Application of the early warning drought system show that reliable and accurate predictions of drought characteristics (severity, duration and area) are estimated for medium term prediction intervals at larger timescales of SPI and for short term prediction intervals at smaller SPI timescales in Pinios river basin, Thessaly, Greece.

Mots clés : Standardized Precipitation Index (SPI) ; Drought ; Spatiotemporal forecasting of drought ; Spatiotemporal modelling of drought ; Drought early warning system ; Neural networks ; Geostatistical methods ; Meteorological drought indices

Présentation

Page publiée le 15 novembre 2022