Revision history [back]

I'm writing a python script to generate some TMY style files from some custom historical data (if anyone knows of a python version that is open and already exists, please let me know!), but some of the calculations don't make sense to me. From the various sources, such as here pg11, in theory the FS method of finding the difference in the various variable of each year from the whole data set makes sense in theory, but practically I'm a bit confused. In summary it's the average difference in CDF for each month or a particular variable, let's use max temperature as an example. So in simple terms I interpret that as finding the absolute difference in the max temperature for each day (for each year file), from the max temperature from the whole data set (for the month in question) and then averaging them - cool. So now for each year, I've got this value. Next we do the same thing for all 9 relevant variables, and then apply the weighting factors to each, sum them up, and then pick the best, more or less. Assuming that's a correct interpretation (and please let me know if I've missed any key points), there seems to be an issue of units here. Each of those statistics is still in the units of the variable. Then we apply the unit-less weighting factors to rank important and sum them. To me this seems like in inadvertently has a second weighting based on typical absolute values of numbers (temperatures are much smaller numbers than solar radiation, for example). Am I missing a step?

I'm writing a python script to generate some TMY style files from some custom historical data (if anyone knows of a python version that is open and already exists, please let me know!), but some of the calculations don't make sense to me. From the various sources, such as here pg11, in theory the FS method of finding the difference in the various variable of each year from the whole data set makes sense in theory, but practically I'm a bit confused. In summary it's the average difference in CDF for each month or of a particular variable, let's use max temperature as an example. So in simple terms I interpret that as finding the absolute difference in the max temperature for each day (for each year file), file, for the month in question) from the max temperature from the whole data set (for the month in question) and then averaging them - cool. them. So now for each year, year/file, I've got this value. Next we do the same thing for all 9 relevant variables, and then apply the weighting factors to each, sum them up, and then pick the best, best month, more or less. Assuming that's a correct interpretation (and please let me know if I've missed any key points), there seems to be an issue of units here. Each of those statistics is still in the units of the variable. Then we apply the unit-less weighting factors to rank important and sum them. To me this seems like in inadvertently has a second weighting based on typical absolute values of numbers (temperatures are much smaller numbers than solar radiation, for example). Am I missing a step?

I'm writing a python script to generate some TMY style files from some custom historical data (if anyone knows of a python version that is open and already exists, please let me know!), but some of the calculations don't make sense to me. From the various sources, such as here pg11, the FS method of finding the difference in the various variable of each year from the whole data set makes sense in theory, but practically I'm a bit confused. In summary it's the average difference in CDF for each month of a particular variable, let's use max temperature as an example. So in simple terms I interpret that as finding the absolute difference in the max temperature for each day (for each year file, for the month in question) from the max temperature from the whole data set (for the month in question) and then averaging them. So now for each year/file, I've got this value. Next we do the same thing for all 9 relevant variables, and then apply the weighting factors to each, sum them up, and then pick the best month, more or less. Assuming that's a correct interpretation (and please let me know if I've missed any key points), there seems to be an issue of units here. Each of those statistics is still in the units of the variable. Then we apply the unit-less weighting factors to rank important and sum them. To me this seems like in inadvertently has a second weighting based on typical absolute values of numbers (temperatures are much smaller numbers than solar radiation, for example). Am I missing a step?

I'm writing a python script to generate some TMY style files from some custom historical data (if anyone knows of a python version that is open and already exists, please let me know!), but some of the calculations don't make sense to me. From the various sources, such as here pg11, the FS method of finding the difference in the various variable of each year from the whole data set makes sense in theory, but practically I'm a bit confused. In summary it's the average difference in CDF for each month of a particular variable, let's use max temperature as an example. So in simple terms I interpret that as finding the absolute difference in the max temperature for each day (for each year file, for the month in question) from the max temperature from the whole data set (for the month in question) and then averaging them. So now for each year/file, I've got this value. Next we do the same thing for all 9 relevant variables, and then apply the weighting factors to each, sum them up, and then pick the best month, more or less. Assuming that's a correct interpretation (and please let me know if I've missed any key points), there seems to be an issue of units here. Each of those statistics is still in the units of the variable. Then we apply the unit-less weighting factors to rank important and sum them. To me this seems like in inadvertently has a second weighting based on typical absolute values of numbers (temperatures are much smaller numbers than solar radiation, for example). Am I missing a step?