Spatial analysis of solar radiation using computer vision technique and data assimilation

Abstract

Purpose

The presence of broken clouds leads to frequent fluctuations in direct normal incident solar irradiation as well as diffuse radiation from the sky. This brings a lot of challenge for grid integration of solar power plants. The paper aims to discuss this issue.

Design/methodology/approach

A new model is presented to nowcast solar radiation by utilizing hourly global horizontal irradiance (GHI) over a large spatial grid. The spatial distribution of the GHI provides information on the presence of a cloud shadow above a given site. This information is extracted with the help of various data processing techniques. The spatial–temporal data analysis is employed to track the extracted cloud shadow image based on a dynamic model. A Kalman filter is applied for the assimilation of data in the tracking of the extracted shadow over a geographical location.

Findings

The proposed model can provide very good forecasting of solar radiation for various time horizons. However, the variation of shadow features between time steps must be included in the dynamic model to forecast accurate GHI values.

Research limitations/implications

In this paper database used is on hourly basis; it can be further improved for the inter-hour level of ground data for more accuracy.

Practical implications

The outcome of this paper would be useful in the field of solar energy application and for weather monitoring purposes.

Originality/value

The forecasted position of the shadow is utilized to prepare and forecast a GHI map for one hour time horizon. Results show that the model can be utilized to forecast solar radiation with accuracy consistent with the contemporary models.

Citation

Kashyap, Y., Bansal, A. and Sao, A.K. (2017), "Spatial analysis of solar radiation using computer vision technique and data assimilation", Management of Environmental Quality, Vol. 28 No. 1, pp. 107-119. https://doi.org/10.1108/MEQ-05-2015-0083