Ανάπτυξη υπολογιστικού συστήματος για τον μονοδιάστατο στοχαστικό επιμερισμό ημερήσιων βροχοπτώσεων σε ωριαίες, σε περιβάλλον R

Abstract

Περίληψη

The study and analysis of hydrological variables require the development and use of special types of stochastic simulation models. These models are a powerful tools for the stochastic simulation and forecast of hydro-meteorological processes. The rain is the most important of hydrological processes. The intermittent character of rainfall time series on fine time scales justify the use of special simulation models. Among the successful model types are the point process models. This type of model has the important feature of representing rainfall in continuous time. According to these models, the rainfall events are simulated through the generation of clustered point or rectangular pulses. The Bartlett - Lewis model has the ability to reproduce important features of the rainfall field from hourly to daily scale and above. A combination of the Bartlett - Lewis rainfall model with proven disaggregation methodology, has proposed by Koutsoyiannis and Onof. This combination improve the ability of the Bartlett - Lewis model to simulate the rainfall on fine time scales. In the framework of my thesis, the model is implemented in a computer program under the name HYETOS-R, on the R environment. The package HYETOS-R is developing in order to provide a complete tool for the simulation of rainfall process on fine time scales. The main purpose of the package is the disaggregation of daily rainfall depth to hourly rainfall depth. The package can work in several modes appropriate for operational use and model testing. Additionally, the user can produce synthetic time series by the Bartlett - Lewis model.