What are the available sources for historical weather data?

edit

How do you generate the solar radiation data?

As far as I know, the files from NOAA include several fields for Solar radiation. However, I don´t think all the weather stations record it. When they are not included (filled with 999 or similar), I believe Dragonfly keeps the values from the reference EPW

You might be interested in this discussion - the radiation data is drawn from ERA5 reanalysis data so should be much more accurate than other methods:

https://discourse.ladybug.tools/t/loo...

How do you generate the solar radiation data?

@osuserAT: Could you please tell us the steps and/or assumptions that you made in creating epw file from this source. epw weather file needs more weather parameters than available at this website e.g. solar radiation (different types), snow depth, days since last snow fall etc. I think this will have influence on the results. thanks

@Waseem, the thread on this question is becoming a bit long. I think that there is certainly enough content here to start a new question.

I agree to Neal Kruis (more questions to come !). re. Waseem's question: the steps of how to do it have been described here: https://unmethours.com/question/3083/... (GENERAL HOW-TO for PREPARING A CUSTOM INPUT FILE FOR THE EPW WEATHER CONVERTER). Solar radiation data etc. are not required. I intend to augment my answer there with more details depending on incoming feedback.

INTERPRETATION AND DATA QUALITY ISSUES : AN EXAMPLE by statistical analysis of downloaded real-time weather data I found that sky condition and visibility for full hours (hh:00) were inconsistent with values between hours (hh:50, hh:20) for a European weather station. E.g. at full hours sky was never "Clear" , visibility ranging from 0-80km, whereas in between full hours visibility ranged from 0-10km; verbal classification of sky condition appeared to use different sets of words for (hh:00) and for (hh:20) & (hh:50) resp. with only a common subset - as if from two distinct data sets.

@osuserAT Your post is ambiguous about the "real-time weather data", but since it follows my earlier answer I will assume that you're speaking of WBT historical weather data. In any case, I think you're being misled by the different conventions being used to denote unlimited visibility, since I don't know how anyone could even measure visibility beyond 10 km, especially at an airport where most of the weather stations are located. Therefore, your concern that visibility ranges from 0-80 km in the "real-time data" and 0-10 km in some other data is not an indication of poor data quality.

(sorry, my comment should have followed the original question, therefore nothing re. WBT) My data source states unambiguously that all visibility values are in km, hence there is no need to worry about coding conventions. But if values of 11,12,..,30, 50,60,75km occur at full hours, and never in between full hours, this needs explanation, and a decision on how to treat the data for conversion into epw file.

I frankly think you're making a mountain out of a molehill. Whether visibility max's out at 80 km or 10 km, both basically indicate unlimited visibility. Visibility is also not used in EnergyPlus or any other building simulation program of which I'm aware. It's basically recorded by weather stations due to aviation needs, and happened to be included in the newer weather file formats like TMY2 and TMY3s and then carried over as an element to the EPWs in the spirit of completeness. How are you using Visibility anyway?

it turns out the meteorological data are derived from METAR code at times hh:20, hh:50, and from AAXX code at hh:00, i.e. the data were apparently merged from two different devices or measurement methods. As you point out, visibility is not used for EnergyPlus simulations, but another problem was how to estimate sky cover from verbal sky condition data (with open questions re. interpretation and data quality). Sky Cover and Visibility were included in the DEF file for generating epw converter estimates of solar radiation because the radiation data were also missing.