Open AccessAssessment of Solar Irradiation Models in A Coruña by Multifractal Analysis
Author(s) -
Rodríguez-Gómez Benigno Antonio,
Meizoso-López María del Carmen,
Mirás-Avalos José Manuel,
García-Tomillo Aitor,
Paz-González Antonio
Publication year - 2013
Publication title -
vadose zone journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.036
H-Index - 81
ISSN - 1539-1663
DOI - 10.2136/vzj2012.0183
Subject(s) - multifractal system , autoregressive integrated moving average , autoregressive model , series (stratigraphy) , radiation , scale (ratio) , time series , mathematics , environmental science , statistics , meteorology , statistical physics , geography , physics , fractal , cartography , geology , mathematical analysis , paleontology , quantum mechanics
The physical processes taking place in the ground are influenced by solar radiation. This climatic variable presents a strong local behavior; therefore local models to estimate irradiation are usually more adequate than others which are more global. The aim of this study was to develop models for global, diffuse and direct radiation in A Coruña (Northwest of Spain), and to apply Multifractal Detrending Fluctuation Analysis (MFDFA) as tool for the assessment of model quality, complementing traditional validation parameters. Autoregressive Integrated Moving Average (ARIMA) methodology was used to obtain daily radiation models. The global irradiation series model explained over 55% of the variance. The model for diffuse radiation showed the lowest prediction errors, and the direct radiation model offered the worst outcomes with the highest errors and the lowest R 2 . MFDFA allowed us to check that the model for global irradiation reproduced the main statistical characteristics of the data series: scale exponent values and points of slope change, and, in general, the multifractal behavior. The diffuse model presented a similar behavior of the series for short‐term large fluctuations, whereas the model for direct irradiation was not capable to reflect the multifractality of the data series. MFDFA can be used as a complement for model assessment, since it offers an analysis of model behavior at different timescales.