What analysis do you plan to do with these locations? If it is strictly solar-related (e.g., PV, Solar hot water, or shading) then you might want to select latitudes that represent the range of human population:

There are also several maps available which show global solar insolation. These can help select locations that might be meaningful for your study as well. A quick Google image search for "solar resource map latitude" gives several options like this:

If you need characteristics other than solar information, you might want to also select locations based on some climate zone definition.

There is no theory, rule of thumb, or body of work that I know of that speaks to the distance at which two weather locations are significantly different. Ultimately, the impact of changing 4 degrees will depend on what you are investigating. For some analyses, like commercial building energy usage, 4 degrees difference might not have any significant impact, but if you are looking at the output of a PV system, then the only way to know if 4 degrees latitude makes a significant difference may be to perform the sensitivity study yourself.

Hi @hema,

Again, your task: "I have to select 3 cities for daylight performance", seems arbitrary. You have a daylight device you'd like to test, apparently for performance across India. It's been stated here that simply changing latitude is not the whole story. Neal was suggesting targeting population centers across the country; Joe and I both suggested that in addition to the change in inclination, other factors affect the climate and in turn the solar availability, and in turn the performance of your device.

The beauty of climate-based simulation is that you can see how local climate affects the performance of one or more changes to a design, and you can do it relatively rapidly. In your case especially, where you wish to deploy the same device on a given building type(s), it's a matter of changing the climate file, maybe some constructions and schedules, and re-running the analysis, per building type. This is easy to do with most of the tools we talk about here every day.

So, again, the best suggestions are to spread your analysis locations around latitudinally, but also look at the climate files (with the visualization tool of your choice) and other tools like the ones Neal already posted, to try and bound the problem as best you can. I would definitely suggest simulating the device in more than three cities, as well.

I'm puzzled by this question. It seems you're assuming the predominant driver in daylighting potential is the latitude, which is true only in the crudest sense, because there are many other factors such as the atmospheric turbidity and the amount of clouds. Instead of using latitude, I would pick a range of solar radiation as shown in the map posted earlier, and if you want only 3 locations to get a range, just pick locations with average, high, and low amounts of radiation :-)

Keep in mind the following when doing your analysis: (1) the illuminances in the weather files are all calculated based on the solar radiation, so take them with a grain of salt, (2) the illuminances shown have to be converted to the amounts entering vertical surfaces, i.e., windows, and can be expected to change greatly, (3) for India, ISHRAE released in early 2014 a new set of typical year weather files for 59 Indian locations with decent solar radiation calibrated against satellite-derived values (I was involved in that work); please contact Vikram Murthy at vikrammur@gmail.com on how to obtain those files.