As we know, solar radiation data are necessary for a vast range of solar energy researches. But, the most important challenge is the availability of such data. For the locations which have no solar radiation measuring stations there are two solutions: Estimations of radiation data using different techniques especially artificial neural networks and remote sensing. Which one is more reliable. This is while that lots of the papers are allocated to estimation solar radiation using ANN technique.
The advantage of ANN and remote sensing solar radiation data depends on your objective. If your problem is sensitive to input solar radiation, it is advisable to rely on remote sensing data and verify with the available resource. ANN is a mathematical tool which aids you to predict based on the existing data.
https://eosweb.larc.nasa.gov/sse/
is a NASA link, where daily data can be obtained.
It's a quite good, global extrapolation .
Hourly data is much more difficult to obtain and some times very expensive.
ANN is a method to extrapolate or interpolate existing data with holes.
The two approaches can perform a good job, once you have calibrated them. In this sense, try to calibrate at least one of them and analize the associared error.
Good morning.
an additional comment... you need to know how (which dataset is used and its limits) the NN is trained in order to avoid overfitting.
Solar radiation from sun to the Earth may be affected by the distance/angle between sun and the Earth (which results the seasons); however, latitude of the location is also has effects on the radiation absorption. On the other hand, the season and latitude just affected the potential radiation, while geomorphology and topography will have significant effects. Therefore, remote sensing would be a good tool wile, you are supposed to georefere the data by measuring the radiation in some of points.
- Badges
- Science topic