Making NOAA's MMTS wireless


This is a custom wireless MMTS with internal data logger that I built, view from my backyard.

NOTE: yes I know it is not the best placement, this is just a preliminary test.

Those who have followed my work surveying official climate stations and their placement issues also know that I’ve been critical of the MMTS (Max Min Temperature System) which has been deployed at a majority of the COOP and USHCN networks. As of this writing, 55% of USHCN is made up of the original MMTS system, with 16% of the network being the improved “Nimbus” version, which has a display unit with max-min memory to prevent data loss. With 71% of the USHCN network being based on MMTS technology, it represents the major component of surface temperature measurement.

The big problem with MMTS is the fact that it is a cabled sensor, whereas the original Stevenson Screens could be placed anywhere, and often at better locations, the MMTS system cable often prevented proper placement because NWS COOP managers could not easily run the cable under walkways or driveways, since they lacked heavy equipment and time. This despite the fact that the original specification for MMTS cabling allowed for distances up to 1/4 mile. Stated simply; COOP managers don’t have access to trenching machines, and installation work is often done with shovels in a single day. 

So, I decided to solve the problem.

The cabling problem has routinely placed MMTS sensors closer to the residences and offices of COOP observers, which has resulted in most USHCN stations that have MMTS/Nimbus being rated a CRN4 due to building/asphalt/concrete proximity of 10 meters or less.

The USHCN station at Bainbridge, Georgia surveyed by Joel McDade is a perfect example that illustrates how an obstacle like a road prevented placing the MMTS unit at the previous location where the Stevenson Screen used to be. Hence, the measurement environment is now entangled with shade tree, asphalt, air conditioner, and nearby building issues.

Bainbridge, GA USHCN Climate Station of Record

Bainbridge, GA USHCN station, MMTS in foreground, Stevenson Screen in background

The solution is a wireless MMTS, and that idea has been bandied about by NOAA, but never implemented.

From the NWS San Diego website

“Currently, the MMTS requires a cable to connect the sensor with a display. Future plans are for wireless displays. This would eliminate many of the problems associated with cabled systems.”

I’m pleased to say that I have constructed and am now testing a prototype wireless MMTS unit as shown in the photo at the top of this post.

This unit departs from typical design in that is is entirely self contained, running on a small battery, and logs not only temperature, but humidity and dewpoint data also. Connection is via a USB port, and the data can be downloaded into a PC in comma delimited format, ready to graph or to upload to a central data collection point. The unit is a combination of some off the shelf parts, some hardware that I’ve added, and some modifications for the purpose of climate observation. The Gill shield (IR shield) is of my own design, but the datalogger unit could easily be installed inside most any existing NWS MMTS shield, or commercially available Gill shield. I’m working directly with a company that has created the basic battery powered microcontroller design to make it work in this climate monitoring application.

The beauty of this system is that it can be left running for days, weeks, and even months (depending on logging interval) and then the data can be downloaded in the field to a laptop. This allows for placement of these units at locations for use in studies of UHI, and cross checking of existing USHCN locations. The datalogger is programmable in many ways and can be adapted for various monitoring tasks.

Here is what the internal sensor package/datalogger unit looks like:


The  sensor package/datalogger unit simply inserts into the center fitting of the Gill shield, and the entire assembly simply screws onto the pipe fitting on the mounting pole. To get the data, simply unscrew the Gill shield, place it on a surface, plug in a USB extension cable, and download the data. A waterproof USB port could also be installed on the side of the pole to allow field connections without dis-assembly.


The entire sensor package/datalogger and Gill Shield w/pole unit was created for less than $150 in parts. I expect costs will be much lower as production costs are tuned.

Now there’s no reason to compromise placement, or fret over trenching cables.

Here is a list of features and specs:

  • Temperature in Centigrade or Fahrenheit, programmable
  • Range of -35° to +80°C (-31 to +176°F)
  • Dew Point in Centigrade or Fahrenheit, programmable 
  • Humidity in 0-100 %RH
  • Data logging intervals from 10 seconds to 1 hour
  • Internal data memory from days to months, depending on logging interval
  • Battery life up to one year
  • Inexpensive and portable
  • Resolution of 0.5 degree
  • Status LED’s indicate operation and fault conditions
  • Operates on a single 1/2 size AA 3.6V Lithium Battery


Humidity Range 0 to 100%RH
Humidity Repeatability

(short term)

Humidity Accuracy

(overall error)(20-80%RH)

Humidity Response Time 5 seconds
Humidity Long Term Stability 1%RH/Yr
Temperature Measurement Range -35°C to 80°C (-31°F to 176°F)
Temperature Repeatability ±0.2°C (±0.4°F)
Temperature Accuracy

(over entire range)

Typical: ±1°C (±2°F)

Max: ±2.5°C (±5°F)

Temperature Response Time 20 seconds
Dew Point Accuracy

(overall error)

(25°C, 40-100%RH)

±2°C (±4°F)**

I’m working on improving the accuracy of the sensor, and I’m investigating an NIST calibration process.

With the unit shown above in my yard, I’ll run a one week test to see how well it compares to my Davis weather station, and I’ll publish the results next week.

Here is a graph of a simple outdoor/indoor test I ran last night and today:


If there is interest, I’ll make these units available to anyone whom wishes to conduct experiments or to install and compare against nearby NOAA stations. if so, please advise me.

0 0 votes
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments
January 12, 2008 9:43 pm

Too cool, Mr. Wizard.

Evan Jones
January 12, 2008 10:16 pm

Hell yeah, Rev!
150 bucks in parts? To be reduced? Wow! A Stevenson Station is a thou a throw. Those ASOS jobs are 5 grand–and if we were using them to take our own temperatures, we’d all think we had the flu.
This fits in with a plan I have been cooking up. (More on this later, but not online. I may want to email you about it at some point fairly soon.)

Stan Needham
January 13, 2008 3:48 pm

There’s a saying (a brand new saying, actually): Them that can, build new, economical, truth seeking gadgets and them that can’t play with climate models. Pretty cool, Anthony.

jerry bono
January 13, 2008 5:31 pm

I’d like one ot two and maybe three. We have some property north of houston in the country and I would like to put one of these there, one in the field and one in the trees.
I might even put one in my attic and see how things are up there. I live in houston and just got a new roof with radiant barrier. I would love to have had this before and after the new roof. There are a lot of claims about radiant barriers and, being an engineer I understand the principle, don’t always tust the claims. I am a perpetual skeptic. I love it.
Jerry bono

jerry bono
January 13, 2008 5:35 pm

As a followup to my previous comment, If we had a detailed temperature profile on a daily basis we couls really come up with a daily temperature based on the curve and integrating that curve to get to get the daily temp. We could still keep Tmax and Tmin but an integrated value based on 15 minute intervals would be great to have. Taking an hourly temp and averaging might be close enough but what’s wrong with higher resolution?
Jerry Bono

January 14, 2008 10:52 am

I was thinking you would go even more wireless. As in Wireless for the whole collection grid of the database, for al of NOAA MMTS stands.
I work in Instrumentation and Controls for the refinery industry. There are several really good wireless temperature units that are designed to be put in place and left for 5 years without calibration or needing the battery changed. And they are truly wireless, some with a several mile range. The Temperature Element is connected to the Transmitter, and the transmitter send out a wireless signal. Then a receiver (so several can be placed with one pickup) takes in several of the inputs, and connects it to a laptop. This in turn puts the signal on the net – and feeds it back to a centeral location for collection, and maybe a portal for people to view it on the internet.
Here is a link to a PDF of some of them:
There are simpler setups, but the basics is: There are Temperature units that are very accurate, robust, secure, and able to be placed without a wire at some very good distances from the pickup. And that data can all be sent online back to the the NOAA with ease. With no daily checking, and without having to have a person even close to the MTSS.
Just some food for thought,

January 14, 2008 11:05 am

Great work ! The logical next steps would be to incorporate short range wireless radio , WiFi . These units have a reliable range of 1/2 mile. I use a driveway alarm made by ‘Dakota Alert’ that sends a motion detection signals 1/4 mile.
For longer ranges why not incorporate a cell phone , set to dial and send data hourly. Solar charging of the cell battery willkeep the phone alive.

Doug Taylor
January 14, 2008 12:34 pm

Before being too carried away about remote wireless, consider that the temperature sensor in you device (it looks like Measurementcomputing model USB 502) is the compensated forward voltage of a semiconductor. This limits the its fundamental accuracy to about .5 K. The unit draws only a small amount of power, therefore it can be battery operated, and remotely placed. Also the unit is inexpensive–$80.00.
I believe that that the temperature sensor for most weather stations is a platinum resistence thermometer(RTD). A well designed RTD unit could have a fundamental accuracy of about .005 K, however the signal conditioning, and the sensor requires significantly more power, which means that it must be hard wired to a power source, either tied to the grid or have a solar/battery combination. This would be expensive.
REPLY: It’s not the measurment computing 502, but is a similar package and comes from the source maufacturer whom I’m working with on this project. Measurment computing does not make their own equipment.
Yes the fundamental accuracy of the solid state device is about 0.5k, but given that the MMTS system is rounded to the nearest K for reporting to NCDC this would seem to be a good fit because again the result would be rounded to the nearest whole number.
While wireless/WiFi might be attractive, but as you correctly point out I don’t want to have an electronics package that needs much power, as again we are back to putting in power cables or solar cells/battery which adds to expense and complication.

Bob Cormack
January 14, 2008 4:26 pm

There is a company in Colorado (I live there) called Inovonics ( that builds a large set of wireless sensors, transmitters, repeaters, etc. They have battery lifetimes as long as 8 years (lithium), and are quite small. These are usually used in creating wireless security networks, remotely reading water, gas, and electric meters, etc., but there is no reason why they couldn’t be used to create a very wide web of temperature/humidity sensors in quite remote places. There are ways to aggregate the data locally and send it over quite large distances. The battery lifetimes usually assume queries at least every 15 minutes, so might last longer for coarser data collection.
There are vast areas of Colorado (and probably many other places) where there is no systematic environmental data collection — it would be interesting to gather high frequency data at the many Bristlecone pine groves at timberline here that are so popular with dendroclimatalogists, for example.
Of course, given that NOAA shut down their ground-level UV montoring stations when it looked like they might defuse the ozone-hole scare, I don’t expect much government interest in getting widespread, accurate temperature data.

January 14, 2008 5:21 pm

per Josh:
“I was thinking you would go even more wireless. As in Wireless for the whole collection grid of the database, for al of NOAA MMTS stands.
I work in Instrumentation and Controls for the refinery industry. There are several really good wireless temperature units that are designed to be put in place and left for 5 years without calibration or needing the battery changed. And they are truly wireless, some with a several mile range. The Temperature Element is connected to the Transmitter, and the transmitter send out a wireless signal. Then a receiver (so several can be placed with one pickup) takes in several of the inputs, and connects it to a laptop. This in turn puts the signal on the net – and feeds it back to a centeral location for collection, and maybe a portal for people to view it on the internet. “

I was actually thinking about the same thing. I noticed that it was a USB plug, and I thought to myself, I have a USB Wi-Fi plug that I use on my desktop, cost me all of $15. Thinking, if there is anyway to incorporate the little 3/4″ X 1″ chip that’s inside this wifi dongle, into your USB receiver. Maybe just catch the solder points to similiar points inside the sensor, and depending on what kind of power the sensor uses, it could maybe be upgraded. All in all I think a difference of half a centimeter in size for the whole sensor package. But I would really have to see the sensor.
I could just be dreaming really. My favorite show when I was a kid was MacGuyver.

John Lederer
January 15, 2008 3:01 pm

So, let’s assume that one could build and place these units for $150 each thanks to Andrew’s fine work.
Would it them be possible to blanket a good sized area with them, laid out on a spatial grid (not urban emphasized), with careful siting, and validate all sorts of assumptions that go into the GISSTEMP record?

Evan Jones
January 17, 2008 10:48 pm

Hey, Rev.
That 0.5C margin of error. How evenly distributed is that? What’s the skew?
REPLY: Don’t know yet, will do more analysis this weekend.

April 12, 2008 11:35 am

Would very much like to try one of these out if possible. I currently own 4 Davis weather monitorII stations and a vp2 station. and will be getting a nimbus unit for coop from NWS next month.

September 14, 2008 4:11 pm

Anyone who wants to record temp/humidity in their own back yard [in order to compare their numbers with the NOAA’s ‘adjusted’ numbers – or just for the fun of it], can buy Anthony’s datalogger for under $60 here: click

November 18, 2008 10:29 am

Where did you get the gill plates at?
REPLY: Here is a link:

November 19, 2008 6:34 am

But rather build my own, already have a davis.

February 6, 2009 9:37 am

I am Geologist and a scientific staff of University in Iran. There is an individual point in northern Lut desert, SE of Iran, which said that probably is the Hotest point in the world. An informal measuring show 61C at shadow. Which is more than Dead valley in Calif.
I want to carry a project for taking the formal measurement of tempreture in Lut desert after finding the best point based on landsat’s Thermal photos. This desert is the most abandoned points of the Earth and there is no inhabitantfor the hundered miles. Thus we need a reliable, accurate, and programmable remoute sensor MMTS. We have no way than put the sensor in the best poit for about 1 or 2 months (comming July Agust) with a helicopter. The Sensor must programable for measuring tempreture with loging interval 1 hour and save the logs in its internal memory.
If we succeed, we could record this point as the Hot Polar of the Earth in the world. Now we are looking for a suitable MMTS for our project. It would be accurate and durable for using in the unusual conditions of this desert. We would appreciated if you are interest in cooperating with our project.
Please reply via my e-mail address.

%d bloggers like this:
Verified by MonsterInsights