We continue to develop resources related to the COVID-19 pandemic. See COVID-19 initiatives on Appropedia for more information.
Modeling methods with spatial climate data
|This page was developed by the Queen's University Applied Sustainability Research Group.|
The World Radiation Data Centre http://wrdc-mgo.nrel.gov/ keeps records of radiation of hundreds of ground weather stations throughout the world. It seems that these data were updated until 1993, however it is not very easy to access them. In terms of Canada alone there is a nice presentation on the current warehouses of these data http://me.queensu.ca/courses/MECH430/Assets/Files/Recommended%20Reading/Weather%20Data%20From%20Environment%20Canada.pdf.Again to access these data requires contacting people and we would much prefer some right at hand resource. RETscreen comes in handy. (By the way, is it possible to transfer a file from here directly onto an Excel spreadsheet?)
Current approaches and tools
What one also notices is that the locations of the weather stations are hardly close to the sites we conduct study (otherwise life is easy) so the way to go is to fit the data into a function, on which we can approximate the values of the site in question. When it comes to estimating potential PV yield the following approaches are more or less popular:
- thin plate smoothing splines, developed, elaborated and promoted by Hutchinson et al (more below)
- r.sun package within GRASS GIS, which was built in Europe and therefore very much inclined to European data (more below)
- Solar Analyst an extension of ArcGIS (more below)
- built in interpolation functions of GIS softwares (I am unsure whether GRASS/ QGIS/ SAGA support all that ArcGIS does, which includes Inverse Distance Weighting, Nearest Neighbor Analysis and Kriging) (more below)
It is really a matter of cost benefit analysis (time/ money/ data availability and acquired skills of the researcher).
Interpolation within a GIS software
Solar Analyst an extension of ArcGIS
PV GIS with r.sun
The model is explained in detail in http://skagit.meas.ncsu.edu/~jaroslav/trento/Hofierka_Jaroslav.pdf or on the PV GIS project website http://re.jrc.ec.europa.eu/pvgis/index.htm. A background reading from Duffie and Beckman makes a much easier transition into the Muneer model for diffuse radiation on inclined surfaces, which the implementation of r.sun was based on. Aside from tricky notation interpretation (e.g. the book and the paper use very different terms for the same angle) it is helpful to remember that if angle A and angle B sum up to 90 degree, then cosine A = sin B and vice versa. Similarly, for the zenith angle (thetaZ) and solar altitude angle (h0), sin h0 = cos thetaZ. The right-hand side is used throughout the book whereas the left hand side is prevalent all over the paper.
Thin plate smoothing splines with ANUSPLIN
A significant amount of work has been done on interpolating radiation data (and many other climate variables) by folks at the Canadian Forest Service (part of Natural Resources Canada) using ANUSPLIN. Check out their solar radiation maps.
Their group has many publications on application of ANUSPLIN in Canada and the USA, including an analysis of solar radiation across Canada for PV applications published in Solar Energy: McKenney, D.W., Pelland, S., Poissant, Y. et al. (2008) Spatial insolation models for photovoltaic energy in Canada. Solar Energy doi:10.1016/j.solener.2008.04.008.
Note that HUTCHINSON's name is spelled thus.
Before proceeding to any paper it is recommended that you read Chapter 9: Digital Terrain Modeling, Concepts and Techniques in Geographic Information System, 2nd edition, Lo & Yeung (2007) and the first two chapters in Solar Engineering of Thermal processes, 3rd edition, Duffie & Beckman. In fact the latter text is included in the reference section of many of the papers listed below.