Persistencia de la precipitación en la Comunidad Autónoma del País Vasco a partir de modelos markovianos. Regionalización mediante sistemas de información geográficos

  1. Fernández de Arróyabe Hernáez, Pablo
Supervised by:
  1. Juan Carlos García Codrón Director
  2. Javier Martín Vide Director

Defence university: Universitat de Barcelona

Fecha de defensa: 22 December 2005

Committee:
  1. Eugenio Ruiz Urrestarazu Chair
  2. Josep M. Raso Nadal Secretary
  3. Domingo Fernando Rasilla Álvarez Committee member
  4. Agustín Ezcurra Talegón Committee member
  5. Jesús Asín Lafuente Committee member

Type: Thesis

Teseo: 129576 DIALNET lock_openTDX editor

Abstract

RAINFALL PERSISTENCE ANALYSIS AT THE BASQUE AUTONOMOUS COMMUNITY BASED ON MARKOVIAN MODELS AND GEOGRAPHIC INFORMATION SYSTEMS The main aim of this work is related to the idea that persistence is an important characteristic of the daily rainfall in the Basque Country. The study area is located in the North of Spain, just in the border between the Eurosiberian and the Mediterranean biogeographical regions. In order to demonstrate the mentioned hypothesis, daily data have been collected at 41 weather stations over the Basque autonomous territory for the period 1,965-1,994. These stations are homogeneously distributed on space. Geographic information system technologies (GIS) have played an important role in this thesis. They have been used to develop large series of climatic maps and advanced geostatistical methods have also been applied. Firstly, it is defined the climatic context of the region using several precipitation numeric indicators. Spatial and temporal distribution of rainfall are also analysed for the study period. Secondly is defined the rainfall structure on each particular location. The absolute frequency of wet chains of n days is identified in each weather station for the study period and its empirical probability under three different rainfall cut points (1, 5 and 10 mm/day) Moreover, wet chains are analysed internally considering the way in which different sequences of wet and dry days are linked. Conditioned probabilities have been estimated and empirical persistence curves have been graphically represented for each weather station datasets. Thirdly, five markovian orders are used to calculate the theoretical probabilities of having a wet chain of "n" days under the three cut points defined before. The markovian models estimate the theoretical probability of having a rainy day after having one, two, three or "n" rainy days before. Empirical and theoretical probabilities values are compared and the statistical significance of theoretical models is produce based on the X2 statistical indicator. Finally persistence is regionalised according to the markovian models that better fit the physical reality of the rainfall in the study area. "