Sky/cloud analysis – cloud motion estimation using ground based sky camera

• Main Author: Lee, Chong kai
• Other Authors: Lee Yee Hui
• Format: Final Year Project
• Language: English
• Published: 2017
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/70841
• Institution: Nanyang Technological University

With the increasing interest in using whole sky imager to monitor cloud formation, solar radiation and weather prediction. Nanyang Technological University (NTU) had their own imager which is known as Wide Angle High-Resolution Sky Imager System (WAHRSIS). The upward pointing nature of this camera enable the user to capture images of clouds which can be used for analysing the weather. All data and images were stored in a database where user can retrieve the desired required data. In this project, the main objective was to forecast the cloud motion and using the predicted image to compute a short-term prediction of solar radiation. To forecast cloud movement using images, the concept of optical flow was implemented. Optical flow is the pattern of motion between two consecutive image (20 second or 2 minute apart). When optical flow is obtained, forecasting of cloud motion were done by assuming the movement of cloud is constant. Next, the accuracy of the predicted images was determined by methods such as root mean square error and image segmentation. Accuracy for predicted image with 20s intervals were promising for several frames. The error percentage for predicted image with 2-minute intervals shows promising result for a 2 to 4-minute lead time. A proposed model was introduced to estimate solar radiation. Data was examined and a relationship between luminance of image and solar radiation shows a linearity. Thus, using the model the estimated solar radiation can be computed. The value obtained may not be equitable but the trend follows the data collected from the weather station. Finally, using the forecasted image and estimated solar radiation model, the predicted solar radiation was computed. Similarly, the value may not be the desired result but the general trend follow closely with the weather station data.