FTP access for Climate Reference Network Data

The CRN station at Mauna Loa, HI – I’m guessing the concrete pad was needed due to the cinders not being a stable anchoring medium.  So this station gets a CRN 5 rating. OTOH, as some readers point out, it is probably cooler than the cinders.

A couple weeks ago I promised that I’d post the ftp address where CRN data can be downloaded when it was made available to me, so here it is. Have at it lads.

Please read the details on the file format and caveats related to the data file timestamps.

NCDC provides ftp access to data from the Climate Reference Network (CRN) via anonymous ftp at the address


(you can open this with most browsers)

here is a sample data file:


That directory contains a separate subdirectory for each year that the ftp data have been available, starting with 2005.  Each such subdirectory contains a collection of ASCII files named according to the following convention:



       FF = 2-digit file frequency, in minutes (currently always 60)

       TT = 2-character file format number (currently always 01)

     YYYY = 4-digit year

       MM = 2-digit month (01=Jan, …, 12=Dec)

       DD = 2-digit day of month

       HH = 2-digit hour of day

       mm = 2-digit minute

Each one of these files contains CRN data that were loaded into the CRN database during the period of time lasting FF minutes and ending at UTC time YYYYMMDDHHmm.  The 2-character sequence TT indicates the file format number.  It is currently always 01.  If at some point in the future for format for CRN ftp data changes, this number will be changed.

Each file contains fixed-column formatted lines, each of which represents one hour’s observation data from one station.  The format of each line is as follows:

    cols 1 — 5 [5 chars] WBANNO

The station WBAN number.

    cols 7 — 12 [6 chars] COOPNO

The station COOP number.

    cols 14 — 21 [8 chars] UTC_DATE

The UTC date of the observation

    cols 23 — 26 [4 chars] UTC_TIME

The UTC time of the observation.

    cols 28 — 33 [6 chars] CRX_VN

The version number of the station datalogger program that was in

effect at the time of the observation.

    cols 35 — 41 [7 chars] LONGITUDE

Station longitude.

    cols 43 — 49 [7 chars] LATITUDE

Station latitude.

    cols 51 — 57 [7 chars] T_CALC

Average temperature, in degrees C, during the last 5 minutes of the hour.  Note: CRN stations have multiple co-located temperature sensors that record independent measurements.  This T_CALC value is a single temperature number that is calculated from the multiple independent measurements.

    cols 59 — 65 [7 chars] P_CALC

Total amount of precipitation, in mm, recorded during the hour.

Note: CRN stations have multiple independent measures of precipitation; this P_CALC value is a single precipitation number that is calculated from the multiple independent measurements.

    cols 67 — 72 [6 chars] SOLARAD

Average solar radiation, in watts/meter^2, for the hour.

    cols 74 — 74 [1 chars] SOLARAD_FLAG

QC flag for solar radiation.  0 means good, 2 means suspect, 3 means erroneous.

    cols 76 — 82 [7 chars] SUR_TEMP

Average surface temperature, in degrees C, for the hour.

    cols 84 — 84 [1 chars] SUR_TEMP_FLAG

QC flag for surface temperature.  0 means good, 2 means suspect, 3 means erroneous.


1. All fields are separated from adjacent fields by at least one


2. Leading zeros are omitted.

3. Missing data are indicated by the value -9999 (or -9999.0).

4. The T_CALC and P_CALC fields do not have QC flags associated with them because they are values that are derived from the raw data (which is not included in these files).  When the raw data is flagged as erroneous or suspect, these derived values are not calculated — they are reported as missing instead.  So the T_CALC and P_CALC fields may be assumed to always be good (unflagged) data, except when they are reported as -9999.0, which indicates a missing value.

5. The files are named according to WHEN THE DATA WERE LOADED INTO THE CRN DATABASE AT NCDC, not according to observation times. Data from CRN stations generally arrives at NCDC and is loaded into the CRN database steadily throughout the hour, but sometimes an observation shows up several hours later, if there was a temporary problem with its transmission.  In some cases, if there is a more long-term problem with data transmissions from a CRN station, data from that station will be transferred manually via PDA, which can happen months after the data were

collected.  In any case, no matter how the data arrives at NCDC, it will show up in an ftp file that is named according to the time at which it was loaded into the CRN database.

This means that in general each .txt file will contain one hour’s worth of data, most of it for the previous hour.  There will occasionally be observations in a file that are from more than one hour in the past, however.  There are also sometimes hours during which no data is loaded at all (for example when the CRN database is down for maintenance for several hours on Sunday nights), which means that the .txt files for this period will be empty.  This is normal.  These periods will be followed by one or more hours in which the .txt files will be larger than average because the system is loading data that that came in over more than a one-hour period.

0 0 votes
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments
Scott Finegan
May 14, 2008 10:24 am

Lets donate artificial turf for that concrete pad.
The concrete is likely to be cooler than the cinders, which could be spread on top of the concrete.
BTW: Is a volcano considered an artificial heat source?

James Acres
May 14, 2008 10:37 am

does this site really need a 5? it’s not like they poured a concrete slab in a meadow. how is a concrete slab a different local environment than a cinder field?
granted, the concrete slab in a pleasant, green meadow needs a five. because that’s a tremendous change from the local environment. but will that same slab in a lava field generate an anomaly of >5C?

May 14, 2008 10:52 am

Wouldn’t be surprised if the concrete provides a negative temp bias vs the volcanic rock. It would be interesting if you took any infrared photos of that site.

May 14, 2008 11:37 am

Yup, the concrete pad requires a CRN rating of 5. But the light-colored concrete might be cooler than the surrounding cinders, so in this case the rules don’t seem quite right. Although normal rules might not apply to a thermometer on a volcano.

May 14, 2008 1:25 pm

Just think if we could locate more recording sites on Volcanoes we could prove global warming. (which hasn’t been questioned)(the question is how much and why) and blame it on the CO2 being coughed out from the volcano never mind all the other gasses and aerosols.
CRN 5 maybe, it might be a cooling greater than 5C of course all bets are off after the next major eruption.

Tom in Florida
May 14, 2008 2:15 pm

I read the article that Roads linked to. The reported inserted a statement describing carbon dioxide, as “the chief greenhouse gas” . What else did he get wrong in his story.

May 14, 2008 4:17 pm

CRN 5 simply means that there are too many problems with the environment around the sensor to have any real confidence in the temperature measurements.
Both the cinders and the slab add to this uncertainty, even if they have different signs.
Two uncertainties does not make things any more certain.

Mike Bryant
May 15, 2008 4:55 am

James Acres,
I think a better question is, why put a weather station in a lava field?

May 15, 2008 5:10 am

I think that’s because CO2 was voted to that office by the Greens. It was probably a disputed election that the water vapor advocates claim was stolen by all the CRN 5 stations false reporting of results. 🙂

Fred Middleton
May 15, 2008 5:26 am

Is lava rock-cinder stone full of air pockets?
REPLY: Generally yes

Gary Plyler
May 15, 2008 8:12 am

As I see it, we don’t know in advance whether the exposed concrete pad will elevate or depress the measured temperature. That being said, why wasn’t the site pre-evaluated as needing a burried foundation with the equipment support posts being a few yard (meter) longer extending up through restored cinders?
I am sick and tired of our government being willing to spend billions (with a b) every year on research (and some alarmists willing to jeapordize trillions of our economy based on the CRN data) while saving a pittance by not engineering this site. The government truely lives and breaths the one size fits all (or in this case one design fits all) philosophy.

Andrew Blackburn
May 15, 2008 11:29 am

I haven’t been to the top of Mauna Loa, but I’ve been to Mauna Kea. Mauna Loa is an active volcano; there’s nothing you can do to make it a reliable temperature measurement site for climate records! 🙂 But that doesn’t mean that the temperature up there doesn’t matter…

May 16, 2008 7:56 pm

Can someone tell me where I can buy a Stevenson screen and/or how to contact Anthony Watts regarding this?
REPLY: they can be purchased from Novalynx in Auburn, CA
I can be reached at: info [-at-] surfacestations dot org

May 20, 2008 7:56 am

Just a question, this appeared on the net,
Note that not all cities show a warming relative to their rural surroundings. For example, Hansen et al. (JGR, 2001) adjusted trends in urban stations around the world to match rural stations in their regions, in an effort to homogenise the temperature record. Of these adjustments, 42% warmed the urban trends: which is to say that in 42% of cases, the cities were getting cooler relative to their surroundings rather than warmer. One reason is that urban areas are heterogeneous, and weather stations are often sited in “cool islands” – parks, for example – within urban areas.
Does anyone know where these COOL urban weather stations are situated.

May 23, 2008 4:58 pm

[…] couple days, not only with my real work, but also trying to decode the CRN data. About 2 weeks ago, Anthony Watts posted the link to the FTP site where anyone can download the temperature data from the new network. There are many […]

August 11, 2008 4:44 pm

The Specific Heat of the concrete pad at Mauna Loa is less than that of the surrounding cinder/lava fields, exagerration of the rating of the station is
rather untenable unless one has a presupposed position on the matter.
Mike Helfert

August 11, 2008 7:52 pm

Mike Helfert (16:44:02) :

exaggeration [sp] of the rating of the station is rather untenable unless one has a presupposed position on the matter.

Please check
Please note this excerpt:
Classification for Temperature and Humidity
Class 1: Flat and horizontal ground surrounded by a clear surface with a slope below 1/3 (<19 degrees). Grass/low vegetation ground cover 3 degrees.
Class 2: Same as Class 1 with the following differences. Surrounding Vegetation 5 degrees.
Class 3 (error 1 C): Same as Class 2, except no artificial heating sources within 10m.
Class 4 (error >/= 2 C): Artificial heating sources /= 5 C): Temperature sensor located next to/above an artificial heating source, such a building, roof top, parking lot, or concrete surface.
It appears ambiguous to me whether this means that any “concrete surface” is an “artificial heating source”. If so, the station is class 5 by definition. Clearly the ranking system was not written with volcanoes or forested regions in mind.
I’d expect the specific heat of most rocks would be about the same (s.h. is the number of calories to raise the temperature of one gram of the material one Kelvin). I’d also expect the heat conductivity of the concrete to by much higher than volcanic cinders, though I’m most familiar with cinders from Arizona and Oregon. If Hawaiian cinders have few/no gas bubbles, then their conductivity ought to be on a par with typical concretes.

August 11, 2008 10:30 pm

Sigh – I just realized I forgot to worry about all the less-than signs in the .pdf text, and those were swallowed by the HTML processor at WordPress.
Reading the FTP site is best, here’s an attempt to get the Class 5 description readable in full:

Class 5 (error >/= 5 C): Temperature sensor located next to/above an artificial heating source, such a building, roof top, parking lot, or concrete surface.

Verified by MonsterInsights