Rainfall Generation Using Markov Chain Models; Case Study: Central Aegean Sea

• Main Authors: Konstantinos Mammas, Demetris Francis Lekkas
• Format: Article
• Language: English
• Published: MDPI AG 2018-06-01
• Series: Water
• Subjects: Generalized linear model; drought-prone area; summer smooth adjustment
• Online Access: http://www.mdpi.com/2073-4441/10/7/856

Generalized linear models (GLMs) are popular tools for simulating daily rainfall series. However, the application of GLMs in drought-prone areas is challenging, as there is inconsistency in rainfall data during long and irregular periods. The majority of studies include regions where rainfall is well distributed during the year indicating the capabilities of the GLM approach. In many cases, the summer period has been discarded from the analyses, as it affects predictive performance of the model. In this paper, a two-stage (occurrence and amounts) GLM is used to simulate daily rainfall in two Greek islands. Summer (June–August) smooth adjustments have been proposed to model the low probability of rainfall during summer, and consequently, to improve the simulations during autumn. Preliminary results suggest that the fitted models simulate adequate rainfall occurrence and amounts in Milos and Naxos islands, and can be used as input in future hydrological applications.