J. L. Bosch1*, L. F. Zarzalejo2, F. J. Batlles1 and G. Lopez3

1 Universidad de Almeria, Departmento de Fisica Aplicada, Ctra. Sacramento s/n, 04120-Almeria, Espana
2 CIEMAT, Departamento de Energia, Madrid, Espana
3 EPS-Universidad de Huelva, Departamento de Ingenieria Electrica y Termica, Huelva, Espana

Corresponding Author, jlbosch@ual. es

Abstract

The correlation of solar irradiation data in flat and homogeneous areas is relatively high and classic interpolation methods are very suitable for its estimation. However, in complex topography zones, a simple data interpolation is not adequate. On the other hand, spatial variability of solar irradiation is also affected by site latitude and cloud cover distribution. In this work, a methodology has been implemented consisting in daily solar irradiation estimation for all sky conditions, by means of Meteosat satellite images and additional information from a Digital Terrain Model (DTM) of the studied area. Solar irradiation is calculated following the HELIOSAT-2 methodology; and a method is presented to obtain the horizon of the studied points using the DTM. The effect of the snow covers is also studied. Model performance has been evaluated against data measured in 14 radiometric stations located in a mountainous area, offering good results, with a Root Mean Square Error (RMSE) around 11%. Finally, a daily solar irradiation map has been generated for the complex topography site.

Keywords: Daily Irradiation Mapping, DTM, Meteosat, Complex Terrain

1. Introduction

Incoming solar radiation, through its influence over the energy and water balances of the earth surface, affects processes like air and soil heating, photosynthesis, wind or snow thawing. Therefore, its knowledge is important in diverse fields and necessary for several applications. For most of these applications, global radiation measures are needed over wide regions, for long time periods and with a high spatial resolution.

At local scales the topography is the most important factor in the distribution of the solar radiation on the surface. In plane and homogeneous areas, classic interpolation methods can estimate the solar radiation accurately. However, in zones with a high topographical variability the spatial correlation is difficult to detect, and for distances between 300 and 1000 m is very small or disappear [1]. The use of interpolation in this kind of terrains can lead to large errors and more complex models that include topographical information are needed [2]. In recent years, digital models of terrain (DMT) have been utilized to develop radiation models incorporating topography and consequently, the spatial variability
of terrain mentioned before. In addition, spatial variability of solar radiation is affected by latitude and cloud cover distribution. This issue has been studied recently with the aid of satellite images. The geostationary satellites (METEOSAT) permanently occupy the same zone over the earth surface, acquiring several images per day, for this reason they are suitable for estimating the solar irradiation and evaluating the energy potential of a wide area. The Centre Energetique et Precedes (CEP), the School of Mines of Paris, in cooperation with other European Centers of Investigation, developed a statistical model to estimate the solar irradiation in the terrestrial surface from images of METEOSAT. The mentioned model is known as HELIOSAT [3]. The basic idea of the HELIOSAT is the interrelation between the cloud cover and the global incident irradiation on a point on the earth’s surface. This model was one of earliest used for the evaluation of the global irradiation from images of satellite. It was developed using measurements of French stations and its goal was the estimation of average monthly values of global irradiation. Afterwards different modifications were introduced [4] until a new version named HELIOSAT-2 was implemented.

In this work, a methodology is presented and tested, in which the estimation of solar irradiation is performed by using a modified HELIOSAT-2 [5], together with the information contained in a DTM. Instead of a single irradiation value for a pixel, the horizon of each inner point is used to estimate around 1000 irradiation values for every pixel. Computational cost of the horizon calculation can be a problem when dealing with large areas; that issue has been addressed by developing an algorithm that reduces drastically the time utilized in this process, without loosing much information about the actual horizon. Additionally, the happening of snow covers can lead to a subestimation of the model, because those pixels can be considered covered by very bright clouds instead of snow, this problem has been also addressed in this work with satisfactory results.

The main goal is to perform an irradiation map of daily values from the satellite images, fitting the spatial resolution of pixels (~ 3.5 km) to the resolution of the DTM (100 m). Ground measurements registered at 14 stations located in a complex topography area have been used for validation purposes, observing an error reduction after the consideration of the horizon and snow effects. It is also interesting to note that this procedure can be applied under all kind of sky conditions.