Binned weather data for US (not the equest weather file)

33 posts / 0 new
Last post

Hi All,

Is there a way to get binned weather data for different locations -
preferably free. A quick google search did not come up with anything
concrete.

Thanks
-Rohini

R B's picture
R B
Offline
Joined: 2011-09-30
Reputation: 4

What do you mean by binned weather data? Like what was in the Air Force Engineering
Weather Data manual from 1978? You might check ASHRAE's Weather Data Viewer that can
produce binned temperature data for over 5,000 locations around the world. It's not free,
though, costs $119.

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406

The updated Engineering Weather Data to cover 1973-1996 and
put it on a CD

http://ols.nndc.noaa.gov/plolstore/plsql/olstore.prodspecific?prodnum=5005

It is only $17 now.

Jason

Jason Glazer's picture
Offline
Joined: 2011-09-30
Reputation: 0

Jason,

Thanks for the information. You mean I can now throw out the old copy of the blue AF
manual that I've been keeping for all these years?
I haven't checked out this new revision, but I would like to point out that the blurb on
the Web says that there are 800 worldwide locations
in the manual, whereas the ASHRAE Weather Data Viewer contains over 5,600. On a cost per
station basis, the two are exactly the same
($17/800 == $119/5600 = 2.1 cents per station) :-) :-)

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406

Here's some food for thought (and sorry to hijack this thread):

I was at an AEE meeting yesterday that included a discussion by a local
meteorologist about the increased variability and extremes that are being
predicted in weather models for the future (i.e. climate change).

What is the feeling in the bldg-sim community as to how useful 1973-1996
binned average data will continue to be when it comes to predicting building
operation, loads, and performance in a future that is expected to be more
variable and extreme with regards to weather?

Is anyone preparing a new set of TMY data with more extremes? Are design
day criteria being updated from the historical ASHRAE 1% values? Will 95
drybulb/78 wetbulb (for Boston anyway) no longer be the standard?

Just wondering if our current methods of building simulation (and design)
may leave something to be desired.

Sincerely,
Alec Stevens

Alec Stevens's picture
Offline
Joined: 2011-10-01
Reputation: 1

Or you can easily make your own bin data using TMY3 data files and MS Excel's Histogram function. Works like a charm and the cost is right :) and you can set the bins to match whatever bin size you want. Use some Pivot Table functionality and you can cut those bins down to only occuppied hours etc.

Kevin L. Wyman, P.E.

anjenaire's picture
Offline
Joined: 2011-01-25
Reputation: 3

Apparently there are weather trends for, say, the most recent 10 years which are different than the 30 years looked at by a TMY file. I suspect this could be due to urban heat patterns as well as the general rise in global temperatures. I don't know if these are significantly different than what a TMY or the AFM 88 blue manual might show.
If you want to investigate, I'm pretty sure that Weather Analytics has a data set offering of actual weather data for 10 years for thousands of locations. They offer file formats for various of the energy modeling platforms, but also have a .CSV format that could be "binned".

James V Dirkes II, PE, BEMP, LEED AP

James V Dirkes II, PE's picture
Joined: 2011-10-02
Reputation: 203

Depending on how much custom binning you are looking to do, Rohini, DView
is one option for data visualization for various time scales, etc.

http://www.mistaya.ca/software/dview.htm

*Arpan Bakshi *

Arpan Bakshi3's picture
Offline
Joined: 2012-03-13
Reputation: 0

True they are 30 years worth of data (a tad longer than 10 years) points with the extremes thrown out but you are using binned data. How accurate of a calculation are you really getting anyway? If one were using bin data I would suggest that throughing out the extremes would be warranted.

Kevin L. Wyman, P.E.

anjenaire's picture
Offline
Joined: 2011-01-25
Reputation: 3

Dear Alec and colleagues

There is a niche of research that looks applies the climate change predictions from the meteorologists/climate change experts in building simulation. In the US, the seminal paper in that area has been written by Dru Crawley in the Journal of Building Performance Simulation, and that specific article is open access. DOI:10.1080/19401490802182079. For a wider international perspective, there is the special issue of Building and Environment that I did with David Coley last year (issue 55, 2012).

Best regards,
Pieter de Wilde

Pieter De Wilde's picture
Offline
Joined: 2011-10-01
Reputation: 1

Here are my two-cents:

I think the idea of calculating design conditions or developing "typical year" weather data
using a long historical record, e.g., the past 30 years, needs to be re-evaluated in light of
global or regional climate change, as well as such irreversible trends as Urban Heat Islands,

Using longer time periods doesn't bring more accuracy in predicting future climate conditions
(after all, aren't we interested in only how buildings do in the future, and not the past?),
but only anchors us more in the past.

About 15 years ago, a meteorologist in ASHRAE TC 4.2 found that you need only 7 years
of data to overcome the stochastic year-to-year variations in the climate. Therefore, I feel
that it's best to use a shorter time period, more like the last dozen, for these purposes.

That's what I've done for the California Energy Commission in creating the new CZ2010 Title-24
weather files, and what I will be doing to calculate design conditions for the same locations.

There's also been discussion in various places on using GCM (Global Circulation Models) to
calculate future weather and then develop design conditions or "future year" weather files.
I find these methods too hypothetical for my tastes, and I'm sure those skeptics out there
would have a field day if we were to use them, so my position is to keep doing things the
old way with the actual historical data, but shorten the time periods to something more
immediate.

Last year, I did a study of the savings for window awnings, where the client asked me to also
do the simulations using the hottest year over the past decade. The results were quite
noticeable.
The hottest year (judged by cooling degree-days) for 40% of the 50 US locations I studied were
in the last two years (2010 or 2011). The cooling loads in many locations increased by 30-40%
(more in the north, less in the south).

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406

If the interest is in getting historical time series of weather, there's now plenty of detailed data in the public domain, particularly for the US. For example, NREL has produced the new NSRDB that has the historical weather data 1991-2010 for 1454 US locations.
That's right, 29,000 station-years of detailed hourly data in the TMY3 CSV format. That's not so remarkable once it's realized that to create any typical year weather file one has to first produce the historical time series. It's just now that NREL, bless their hearts, has decided to release publicly what had been internal intermediate data for creating the TMY3 files.

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406

I don't understand the concern about needing to do binning of multiple number of years, and why that would affect the accuracy of a bin program any more than saying we should do hourly simulations with a dozen historical years instead of with a single typical year file. The loss in accuracy must be around a percent or so.

I feel that typical year weather files have been unfairly characterized as bland, misses the extremes, etc. That would be true if the typical months were picked based only on their averages. However, the TMY methodology developed by NREL is much better than that. It picks the month with a frequency distribution closest to the long-term FD. So, it doesn't throw out extremes at all; it just tries to replicate the long-term distribution with the best that can be done using a single-year of record.

If you compare the 0.4% condition between the historical record and a typical year weather file, you'll find statistical scatter rather than bias.

In summary, the only conclusion one can make about typical year files is that their frequency distribution is more coarse because they've only got 8760 observations. From the analysis that I've done, I found the biggest effect and concern is with the period of record chosen in their creation. All the other perceived defects, weighting scheme, are quite secondary.

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406

You could use EnergyPus weather files from DoE and use eQ_WthProc to convert them into the required bin format.

http://apps1.eere.energy.gov/buildings/energyplus/cfm/weather_data.cfm

http://doe2.com/index_Wth.html

Jonathan Bowser, MCPD

Jonathan Bowser2's picture
Joined: 2011-09-30
Reputation: 0

What do you mean by binned weather data? Like what was in the Air Force Engineering
Weather Data manual from 1978? You might check ASHRAE's Weather Data Viewer that can
produce binned temperature data for over 5,000 locations around the world. It's not free,
though, costs $119.

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406

I thought Rohini was asking about "binned" weather data, not *.bin weather files used by
DOE-2. BTW, the "bin" here is an abbreviation for the binary format of these files.

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406

www.weatheranalytics.com is great, each 12 month, custom weather file is $75
- you choose the location and actual dates, it comes in correct eQuest
format.

Jim Fowler, PE, CEA

Jim Fowler, PE, CEM
Jim Fowler's picture
Offline
Joined: 2012-12-14
Reputation: 1

Rohini:

If I understand you correctly, I can think of three possible solutions to solving your (binning) problem:

1) Create your own bin output table from a configured EnergyPlus simulation driven by a particular weather file EPW. Configure your .idf to generate this specific output report: Table:TimeBins and have it report out the weather related variable (dry bulb temp, etc) that you are interested in binning. This report will populate for the entire Run Period of your model, so unless you limit your Run Period to a shorter than annual period, you will only receive bin analysis results for the whole entire year. Use any quick and error free idf to generate this report, as you don't really care about any simulation related energy results anyways, and are only using E+ simulation capabilities to generate this specific bin report for your specific weather related variable. If you are comfortable using the E+ idf editor, this should not take long to create and exercise, and is essentially free.

A snippet of the E+ support text for this type of output table follows:

Output:Table:TimeBins
A TimeBins report shows a grid of the number of hours that specific report variable spend in a range of values. The first table is broken into two parts. The first part is monthly and the second part is for each hour of the day.

2) Essentially automate what EnergyPlus does in (1) by writing or recording your own xls macro which is fed by a structured (EPW or TMY2/3 file) transformation into a CSV file. Create levels of bin resolution (time of year, day of week, hour of day, width of bin, selected variables, etc)) to whatever level you need. If you go this route, expect to invest some real time in this, both for creation, automation and subsequent debugging (i.e. not really free of effort)

3) Purchase and exercise BinmakerPro. (~ $180 from the Gas Research Institute). BinMaker Pro's interface is very flexible in that you can choose individual months, individual days of week, and individual hours with the selected months/days to generate several different type of canned binned analysis against. Downside is that the bin widths are fixed (2 degrees, 5 degrees or 10 degrees). Other downside is that the BinmakerPro easily accepts TMY2 or TMY3 files, all other types of files (EPW, etc) will need be transformed into this format or go through a somewhat tedious custom file format definition process before BinmakerPro will eat them and regurgitate results them.

For what it is worth (and for me, time is money), when I had to decide, I went down route #3, and I have been very happy with the results. It had the best combination of flexibility, ease of use, and repeatability for my purposes.

All the Best,

Chris Balbach, PE, CEM, CMVP, BEMP, BEAP, BESA

Chris Balbach's picture
Offline
Joined: 2011-09-30
Reputation: 1

Hi Rohini,

You can obtain US TMY3 weather files in .CSV format from the following website...

http://rredc.nrel.gov/solar/old_data/nsrdb/1991-2005/tmy3/by_state_and_city.html

I often open these in excel and create my own bins using the excel 'histogram' function (in the excel data analysis tool pack).

LYLE KECK LEED AP BD+C

Lyle Keck's picture
Offline
Joined: 2011-09-30
Reputation: 1

You are correct, sir. I was mistaken. :)

Jonathan Bowser, MCPD

Jonathan Bowser2's picture
Joined: 2011-09-30
Reputation: 0

In the old days I had luck with using one of the Regional Climate Centers. I used the Northeast (NRCC) based at Cornell. For a fee they would develop bin data for any of the cities they had data for, either single years or combined groups of years i.e. 1993 to 2003.

http://www.nrcc.cornell.edu/page_services.html

Yes, they are agriculturally oriented but they have a lot of data. Data is data.

[cid:image001.jpg at 01CBEEF3.08329F30]

Michael W. Oliver, P.E.

Mike Oliver's picture
Offline
Joined: 2011-09-30
Reputation: 1

NREL has most locations around the world in csv file format. I have been using these files for years. Also, this is the data that a lot of other energy modeling programs use for weather simulation data. Hopefully this is what is needed. Here is the link to download the files.

http://rredc.nrel.gov/solar/old_data/nsrdb/1991-2005/tmy3/by_state_and_city.html#Top

Sincerely,

Brandon Damas, P.E., LEED ? AP

Brandon Damas's picture
Offline
Joined: 2013-01-10
Reputation: 1

This is the first I've heard of NREL having (weather data ?) "around the world in csv file
format". Outside of a few selected countries or regions (SWERA, China, India), I was not
aware that NREL was working on weather files for non-US locations. Furthermore, their
international focus has been mainly on estimating solar radiation, not on developing
weather files per se. If you know of such data at NREL, please share more information on
what they are or where the files are available.

I do know that the National Climatic Data Center (NCDC) has put a lot of weather data from
around the world online as the ISH (Integrated Surface Hourly) data base
http://www.ncdc.noaa.gov/oa/climate/isd/index.php

In fact, the new NSRDB (National Solar Radiation Data Base) 1991-2010 that I mentioned
earlier is now accessible both at a NREL site
http://rredc.nrel.gov/solar/old_data/nsrdb/1991-2010/SolarAnywhere/
as well as via ftp at a NCDC web site:

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406

This is the first I've heard of NREL having (weather data ?) "around the world in csv file
format". Outside of a few selected countries or regions (SWERA, China, India), I was not
aware that NREL was working on weather files for non-US locations. Furthermore, their
international focus has been mainly on estimating solar radiation, not on developing
weather files per se. If you know of such data at NREL, please share more information on
what they are or where the files are available.

I do know that the National Climatic Data Center (NCDC) has put a lot of weather data from
around the world online as the ISH (Integrated Surface Hourly) data base
http://www.ncdc.noaa.gov/oa/climate/isd/index.php

In fact, the new NSRDB (National Solar Radiation Data Base) 1991-2010 that I mentioned
earlier is now accessible both at a NREL site
http://rredc.nrel.gov/solar/old_data/nsrdb/1991-2010/SolarAnywhere/
as well as via ftp to a NCDC web site:
ftp://ftp.ncdc.noaa.gov/pub/data/nsrdb-solar/

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406

I agree with Joe's assessment here on the length of time for a TMY file.
Since we construct TMY files on demand, we are actually thinking about
making the number of years a user selectable variable..."Build from the last
___ years of data". Would that be useful?

Chuck Khuen

Chuck Khuen's picture
Offline
Joined: 2011-10-02
Reputation: 0

We have some good discussion at the NY IBPSA meeting last night about TMY
vs. AMY weather data, and looking at both on projects even though AMY data
is not common in U.S. standards / rating systems.

Any thoughts here? Do others run concurrent simulations using AMY data?

*Arpan Bakshi *

Arpan Bakshi3's picture
Offline
Joined: 2012-03-13
Reputation: 0

Since ASHRAE is already adjusting design day information (see Dru Crawley's post earlier today), I think the best use for AMY data in our industry will be an energy model of an existing building for which you want to match actual vs predicted energy use. Using the actual weather eliminates a significant unknown and allows you to focus on the other uncertainties as you calibrate the model.

James V Dirkes II, PE, BEMP, LEED AP

James V Dirkes II, PE's picture
Joined: 2011-10-02
Reputation: 203

I'm personally a big proponent for modeling location-specific AMY weather whenever there is calibration to historical data to consider, but that certainly doesn't apply to every model or standard. I don't consider AMY/TMY to be an either-or proposition as much as it makes sense to use either as appropriate to the analysis at-hand.

I have myself used both kinds of weather files with the same model, to different ends. If you are trying to make long-term ROI decisions, averaged weather data is absolutely preferable to AMY as you don't want to over- or under-account for the weather extremes that occur over any given period in isolation. If you are calibrating an existing building to past utility bills and measurements, that calibration can be performed with much more certainty using AMY.

As to using 30+ years of averaged data vs. a smaller averaged set of more recent data for TMY, taking any position on the matter seems like "timing the market" on some level to me... I'm very curious to hear the opinions of the weather scientists and researchers more informed than myself. I wonder (out loud) whether we should consider the anticipated lifespan of the building and how long the owner will be paying the utility bills in making such a decision on a per-project basis? Perhaps a school district occupying the same building for many decades is better served making decisions with a more conservative set of averaged weather data (not necessarily "shorter," but with amplified extremes)?

So far as LEED/standards like the 90.1 performance rating method are concerned, I think TMY remains preferable to AMY to avoid "cherry picking" a year which would benefit the demonstrated performance. I don't think the duration of years involved in the TMY data should at least have some minimum (if that's not spelled out somewhere already), if not an explicitly required number of years, for the same reasons.

NICK CATON, P.E.

Nick-Caton's picture
Offline
Joined: 2011-09-30
Reputation: 805

I would like to clarify that ASHRAE TC 4.2 has not adjusted the design
day information in the HOF. What Dru meant I think is that global
climate change and urban heat island effects are being seen the design
day information as the tables are being periodically updated every 3-4
years. There has been no change in the design day calculation procedure
that's related to capturing more recent results or projecting trends
into the future. The closest that TC 4.2 have come is to list the trend
for each station over recent years.

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406
Julien Marrec's picture
Offline
Joined: 2013-01-11
Reputation: 0

The topic of using actual year versus typical year weather data (please
note my rant about nomenclature at the end of my post below, with
apologies to Steve Cornick :-) ) was discussed last October, so maybe we
need to create a FAQ on this topic ? which reminds me that ASHRAE TC
4.2 has a FAQ about various types of weather data and their appropriate
use ( see http://tc42.ashraetcs.org/faq.html#6 ). To me, the
distinction is obvious - if you're evaluating actual building
performance, use the actual year weather file; if you're evaluating
hypothetical building performance (as for standards compliance or LEED
rating), use a typical year weather file.

This topic is separate from the one about the best method to create a
typical year or calculate design conditions. This is as much an
engineering question as a meteorological one, so I'm afraid the insight
you will get from climatologists will be limited. When I worked with
meteorologists in several countries, esp. in the US and China, I was
struck that the concept of a "design day" or a "typical year" was
something outside their experience.

I agree that using an actual year weather file to evaluate a building's
performance is "cherry picking" the data, but is anyone really doing
that? Wasn't that why typical year weather files were developed in the
first place more than 30 years ago? My interest in seeing the time
period for developing typical years or calculating design conditions
reduced from 30 years is because I think the results would then respond
better to ongoing trends instead of lagging by a decade or more. It
would be an easy hypothesis to test and I'm thinking of doing that for a
selected number of US stations, i.e., create typical years of different
time periods and see which time period best matched the results for the
following X number of years. As for the minimum acceptable number of
years to create a typical year, that's been implicitly decided by ASHRAE
TC 4.2 and NREL as either 7 or 8 years.

The idea of a more conservative set of averaged weather data seems
problematic to me. It's hard to define what's conservative because it
would depend on the building as well as the climate, and any effort to
massage real data would open up a whole can of worms. You would be much
better off to simply run the building over the past so many historical
years, and pick the year with the highest energy consumption to do your
projection.

Joe

my rant on weather nomenclature (as promised earlier): I prefer saying
"typical year" when talking about the general concept, and use TMY only
when referring specifically to the TMYX data sets or methodology
developed by NREL. I find using TMY as a general term confusing and not
fair to all the other typical year data sets, such as TRY, WYEC, IWEC,
CZXXRV, DRY, etc. I also find the term AMY odd, as if real weather data
needed to be branded (rant over).

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406

This is the first I've heard of NREL having (weather data ?) "around the world in csv file
format". Outside of a few selected countries or regions (SWERA, China, India), I was not
aware that NREL was working on weather files for non-US locations. Furthermore, their
international focus has been mainly on estimating solar radiation, not on developing
weather files per se. If you know of such data at NREL, please share more information on
what they are or where the files are available.

I do know that the National Climatic Data Center (NCDC) has put a lot of weather data from
around the world online as the ISH (Integrated Surface Hourly) data base
http://www.ncdc.noaa.gov/oa/climate/isd/index.php

In fact, the new NSRDB (National Solar Radiation Data Base) 1991-2010 that I mentioned
earlier is now accessible both at a NREL site
http://rredc.nrel.gov/solar/old_data/nsrdb/1991-2010/SolarAnywhere/
as well as via ftp at a NCDC web site:

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406

This is the first I've heard of NREL having (weather data ?) "around the world in csv file
format". Outside of a few selected countries or regions (SWERA, China, India), I was not
aware that NREL was working on weather files for non-US locations. Furthermore, their
international focus has been mainly on estimating solar radiation, not on developing
weather files per se. If you know of such data at NREL, please share more information on
what they are or where the files are available.

I do know that the National Climatic Data Center (NCDC) has put a lot of weather data from
around the world online as the ISH (Integrated Surface Hourly) data base
http://www.ncdc.noaa.gov/oa/climate/isd/index.php

In fact, the new NSRDB (National Solar Radiation Data Base) 1991-2010 that I mentioned
earlier is now accessible both at a NREL site
http://rredc.nrel.gov/solar/old_data/nsrdb/1991-2010/SolarAnywhere/
as well as via ftp to a NCDC web site:
ftp://ftp.ncdc.noaa.gov/pub/data/nsrdb-solar/

Joe Huang

Joe Huang's picture
Offline
Joined: 2011-09-30
Reputation: 406