French Guiana is located in South America close to the equator (between 2° and 4° N), a huge asset in the production of electrical power with the use of solar panels. The yearly average solar-radiation is 5.8 kWh/m2day .The main focus of this study is to determine the solar energy potential in French Guiana using Kernel density estimation.
The knowledge of surface solar irradiance may be obtained using physical laws such as radiative transfert functions that link satellite apparent albedo to surface irradiance. However, due to the complexity of physical processes, it is difficult to develop accurate and reliable nonlinear observation law, in particular in a tropical area (our area of study). In this work we used a two dimensional Kernel density estimation with a learning dataset of clearness index data and apparent albedo data to estimate surface solar irradiance.
A time series of images every 30 min from the visible channel of the GOES-13 meteorological geostationary satellites from the year 2010 to 2013 has been selected with a spatial resolution of 1 km x 1 km. Hourly in-situ measurement data from six ground reference stations were provided by the French national meteorological agency.
The predicted solar irradiance values from the Kernel density estimation were given in the form of hourly, monthly and annual maps. The maps obtained highlighted interesting trends: the annual average maps showed that the western side of French Guiana receives the highest irradiance values and is dependent on the position of the InterTropical Convergence Zone (ITCZ).
The mean bias percentage error and mean RMSE values were respectively found to be less than 5% and 10% for the testing stations on a hourly basis. The kernel density estimation show great accuracies for evaluating solar resource potential. These results give a first overview of the solar energy potential in French Guiana.