| Summary: | Photosynthetically active radiation (<i>PAR</i>) is a useful variable to estimate the growth of biomass or microalgae. However, it is not always feasible to access <i>PAR</i> measurements; in this work, two sets of nine hourly <i>PAR</i> models were developed. These models were estimated for mainland Spain from satellite data, using multilinear regressions and artificial neural networks. The variables utilized were combinations of global horizontal irradiance, clearness index, solar zenith angle cosine, relative humidity, and air temperature. The study territory was divided into regions with similar features regarding <i>PAR</i> through clustering of the <i>PAR</i> clearness index (<i>k<sub>PAR</sub></i>). This methodology allowed <i>PAR</i> modeling for the two main climatic regions in mainland Spain (Oceanic and Mediterranean). MODIS 3 h data were employed to train the models, and <i>PAR</i> data registered in seven stations across Spain were used for validation. Usual validation indices assess the extent to which the models reproduce the observed data. However, none of those indices considers the exceedance probabilities, which allow the assessment of the viability of projects based on the data to be modeled. In this work, a new validation index based on these probabilities is presented. Hence, its use, along with the other indices, provides a double and thus more complete validation.
|
|---|