Requirements and Design Specification
This section discusses how the topics in the literature review will be implemented within the project. Following the review of three existing computational solar models it became apparent that there was a requirement for the High Resolution Solar (HRS) Model. As out of the three comparable models the only one with a suitable time resolution, to understand the variability of solar irradiation, is Satel-Light but this model derives data from satellite images and hence understanding cloud cover becomes an issue when the ground is covered by snow and ice and for low sun angles.
Through assessing the existing solar models, different interpolation techniques and the constraints on the data made available by the British Atmospheric Data Centre (BADC) a design specification was able to be produced:
The critical evaluation of each of the interpolation techniques suggest that kriging is the method that should be implemented into the hindcast model based on the following statements:
The model itself will be developed using the high-level programming language and interactive environment Matlab and will be described to be feature-rich as it will have many options and functional capabilities available to the user to understand the variability of solar resource:
Through assessing the existing solar models, different interpolation techniques and the constraints on the data made available by the British Atmospheric Data Centre (BADC) a design specification was able to be produced:
- Input data/spatial resolution: >100 MET stations (increasing the amount of stations adds to the complexity of the interpolation stage but increases the accuracy)
- Time resolution: Hourly data is the highest time resolution data available from the MET office (higher time resolution leads to a greater understanding of global solar radiation variability)
The critical evaluation of each of the interpolation techniques suggest that kriging is the method that should be implemented into the hindcast model based on the following statements:
- Kriging helps to compensate for the effects of data clustering by assigning individual points within the cluster less weight than isolated data points (Bohling, 2005). Data clustering may be an issue with the HRS model as the number of MET stations could lead to clustering
- Kriging requires 50-100 samples to obtain a reliable variogram that correctly describes spatial structure (Webster & Oliver, 1992). As previously stated >100 stations will be incorporated therefore the number of samples exceeds the maxima of the minimum required
The model itself will be developed using the high-level programming language and interactive environment Matlab and will be described to be feature-rich as it will have many options and functional capabilities available to the user to understand the variability of solar resource:
- Generate a CSV file of hindcast data at a requested location in the UK
- Create a daily averaged solar map of the UK over a requested time period
- Produce 24 hourly maps from 00:00 to 23:00 on a requested date
Graham Cairns
University of Edinburgh, 2013
University of Edinburgh, 2013