Search

Agrometeorological stations have horizontal solar irradiation data available, but the design and simulation of photovoltaic (PV) systems require data about the solar panel (inclined and/or oriented). Greenhouses for agricultural production, outside the large protected production areas, are usually off-grid; thus, the solar irradiation variable on the panel plane is critical for an optimal PV design. Modeling of solar radiation components (beam, diffuse, and ground-reflected) is carried out by calculating the extraterrestrial solar radiation, solar height, angle of incidence, and diffuse solar radiation. In this study, the modeling was done using Simulink-MATLAB blocks to facilitate its application, using the day of the year, the time of day, and the hourly horizontal global solar irradiation as input variables. The rest of the parameters (i.e., inclination, orientation, solar constant, albedo, latitude, and longitude) were fixed in each block. The results obtained using anisotropic models of diffuse solar irradiation of the sky in the region of Castile and León (Spain) showed improvements over the results obtained with isotropic models. This work enables the precise estimation of solar irradiation on a solar panel flexibly, for particular places, and with the best models for each of the components of solar radiation. [ABSTRACT FROM AUTHOR]

Image processing entered the era of artificial intelligence, and machine learning algorithms emerged as attractive alternatives for time series data processing. Satellite image time series processing enables crop phenology monitoring, such as the calculation of start and end of season. Among the promising algorithms, Gaussian process regression (GPR) proved to be a competitive time series gap-filling algorithm with the advantage of, as developed within a Bayesian framework, providing associated uncertainty estimates. Nevertheless, the processing of time series images becomes computationally inefficient in its standard per-pixel usage, mainly for GPR training rather than the fitting step. To mitigate this computational burden, we propose to substitute the per-pixel optimization step with the creation of a cropland-based precalculations for the GPR hyperparameters θ. To demonstrate our approach hardly affects the accuracy in fitting, we used Sentinel-2 LAI time series over an agricultural region in Castile and Leon, North-West Spain. The performance of image reconstructions were compared against the standard per-pixel GPR time series processing. Results showed that accuracies were on the same order (RMSE 0.1767 vs. 0.1564 [ m 2 / m 2 ] , 12% RMSE degradation) whereas processing time accelerated about 90 times. We further evaluated the alternative option of using the same hyperparameters for all the pixels within the complete scene. It led to similar overall accuracies over crop areas and computational performance. Crop phenology indicators were also calculated for the three different approaches and compared. Results showed analogous crop temporal patterns, with differences in start and end of growing season of no more than five days. To the benefit of crop monitoring applications, all the gap-filling and phenology indicators retrieval techniques have been implemented into the freely downloadable GUI toolbox DATimeS. [ABSTRACT FROM AUTHOR]

This article evaluates horizontal daily global solar irradiation predictive modelling using artificial neural networks (ANNs) for its application in agricultural sciences and technologies. An eight year data series (i.e., training networks period between 2004–2010, with 2011 as the validation year) was measured at an agrometeorological station located in Castile and León, Spain, owned by the irrigation advisory system SIAR. ANN models were designed and evaluated with different neuron numbers in the input and hidden layers. The only neuron used in the outlet layer was the global solar irradiation simulated the day after. Evaluated values of the input data were the horizontal daily global irradiation of the current day [H(t)] and two days before [H(t−1), H(t−2)], the day of the year [J(t)], and the daily clearness index [Kt(t)]. Validated results showed that best adjustment models are the ANN 7 model (RMSE = 3.76 MJ/(m2·d), with two inputs ([H(t), Kt(t)]) and four neurons in the hidden layer) and the ANN 4 model (RMSE = 3.75 MJ/(m2·d), with two inputs ([H(t), J(t)]) and two neurons in the hidden layer). Thus, the studied ANN models had better results compared to classic methods (CENSOLAR typical year, weighted moving mean, linear regression, Fourier and Markov analysis) and are practically easier as they need less input variables. [ABSTRACT FROM AUTHOR]