It is with joy and some trepidation to report that White Box Technologies (WBT) is updating all its historical weather files with satellite-derived solar radiation. Joy because this overcomes what has been the most significant question mark with weather files; trepidation because of the amount of work needed to carry out and maintain this effort. To show that this is more than marketing hype, I need to give a rather long explanation about this development.
The bane of weather data over the past three decades has been the solar radiation (global horizontal and direct normal) which are not measured parameters, but derived using various solar and sky models. All the familiar "typical year" sets, i.e., TMY, WYEC, IWEC, etc., let alone the historical weather files, have modeled solar radiation. Although a lot of work has gone into such models (see M. Iqbal,"An Introduction to Solar Radiation", Academic Press, 1983), there remain an almost intractable problem of the lack of good measured solar to tune any of these models. For example, in the ASHRAE IWEC2 weather files, my team was able to find one or two years' measured data for less than 50 locations, from which were derived 28 sets of regression coefficients then used for all 3,012 IWEC2 locations.
For the past decade and a half, researchers around the world have been working to derive solar radiation from weather satellite imagery, driven largely by the needs of the solar power industry for the siting of solar power plants and getting "bankable" solar estimates for their arrays. Our little building energy simulation sector can of course benefit by hanging on the coattails of the solar power industry, but the downside has been to be totally priced out, since the commercial cost for one year's solar data for one location (grid cell) typically runs around $1,000.
A welcome development over the last five years is that various government offices or affiliated consortia are now beginning to also providing access to satellite-derived solar radiation at minimal or more acceptable costs under various conditions. Over the past three years, WBT has obtained access to such data and permission for its use in WBT weather files. WBT is now either replacing the solar radiation on its historical weather files, or using satellite-derived radiation to develop custom solar coefficients for each location to extend the satellite-derived solar to time periods outside the available time window. With the exception of polar locations above or below 60/66 degrees, island nations in the Pacific and Indian Ocean, and a few unfortunate "blind spots", the entire land mass is being covered with at least 10 years up to 18 years of hourly solar records.
Starting in 2018, WBT historical weather files in the following areas will all have satellite-derived solar radiation for the following time periods: Europe, Africa, South America east of 66 West, i.e., Brazil and Uruguay (2004 to date), Australia (1999 to date), and East Asia (2007 to date, access pending). WBT historical weather files in the following areas will have satellite-derived solar radiation for the indicated time periods - North America and Central/South American down to 20 South (1998-2015), South Asia (2000-2014), with modeled solar radiation from 2016 on that has been individually tuned to the past satellite-derived solar.
Another benefit to the satellite-derived solar is to increase the number of available weather stations, which in many places has been limited by the lack of cloud cover data needed to model the solar radiation. For reasons that are not immediately identifiable, several English-speaking Commonwealth countries has seen a marked drop in the number of available stations due to the decreases in the reporting of cloud cover (see plot, ZAF = South Africa). For example, the number of stations in the UK has dropped by almost 2/3s between 2001 and 2017 (174 to 64), but with satellite-derived solar, it will go back up to 190, while in Australia and South Africa the comparable numbers are from 175 to well over 500, and from 37 to 184, respectively. Although I won't know until the updating is all done, but my working estimate is that for 2016 the number of files will increase from the current 8,230 to 11,016 (2266 USA, 748 Canada, and 8002 International; for some other representative countries, France goes up from 151 to 179 files, Switzerland 38 to 142 (!), Germany from 25 to 77, Egypt from 25 to 33, etc.)
