Who's Adjusting the Climate in Tucumcari: Cows, Canals, or Hansen?

On Sunday I posted about the USHCN climate station of record in Tucumcari, NM highlighting its positive points since it has all the hallmarks of a well sited station with a long and uninterrupted record. But something was odd with the temperature record that didn’t quite make sense at first glance.

I also cross posted my report on Climate Audit since I always value additional input from the community there.

I noted that while this station is in fact well sited, and rates a CRN2, it has some oddities with it’s temperature record around the year 2000, something that looked like a step function to me.

Click for larger graph from NASA GISTEMP

Of course, even though this a truly rural station, 3 miles from the outskirts of town, Hansen and GISS apply adjustments to it anyway, which is part of the flawed “nightlights” algorithm incorrectly flagging this station for adjustment. Even though the adjustment makes the present cooler, it still seems misplaced given the station quality and history. Steve McIntyre said it best:

Here we have a rural station where there doesn’t seem to be any reason to adjust the temperature for population growth/UHI. But in this case, Hansen adjusts Tucumcari as though it were a city. Why is he even adjusting Tucumcari at all? (The “reason” is that its lights value removes it from the rural classification and it goes into the adjustment pool.) While Hansen sometimes seemingly cools rural stations in the past, for the GISS dset2 version here, he warms the past of the station (cools the present).

It’s more that this is a case of another unjustified adjustment by the “adjuster in chief” showing once again that the Hansen adjustments do not do what they are supposed to do – and the best that can be hoped from the Hansen adjustment program by users of this dataset is that the adjustments overall end up being pointless and random, rather than pointless and biased.

The adjustment by GISS looks like this:

Ok adjustments aside, the fact that there is a step at 2000 that remained unexplained until commenters on CA started looking at the data themselves. DaleC provided this graph:

Click for original image

Note the arrow that I place at the year 2000. Notice anything?

The annual average minimum temperature has exceeded 45°F and maintained the rise since 2000. For the first time in the station history going back to 1905, the minimum temperature has gone above that 45°F mark and stayed there. Yes there have been some previous brief excursions above 45°F, but none appear to have lasted more than 2 years. Note that average annual maximum temperatures did not increase during the same period.

What could cause that? We can rule out adjustments, since this is GHCN data before Hansen gets his adjuster mitts on it. We can rule out location change or equipment change, since according to NCDC metadata the station has been in the same place since at least 1946 and possibly longer. It still uses mercury max/min thermometers, so there’s no MMTS next to a building or parking lot to blame.

So what is left? Something around the station in the measurement environment that affects the nighttime readings. I recalled seeing this before. And back in early 2007, I had posted a story about a paper from Dr. John Christy of UAH where he studied a number of stations in the San Joaquin Valley in California because they had exhibited this same symptom:

The culprit? Irrigation. See my story and Christy’s press release

Christy remarks: “Another factor is the dry air, something common to all deserts. Water vapor is a powerful greenhouse gas. Desert air lacks water vapor. The air turns cold at night because it doesn’t retain much warmth from the daytime and it can’t trap what little heat might rise from the ground at night.”

Evaporation from irrigated fields adds water vapor to the air — a process that cools summer days but traps heat rising from the damp soil at night.

“If there is anything I’ve learned in Alabama, it is that humidity can make summer nights very warm,” said Christy, a Fresno, Calif., native who has lived in Alabama since 1987.

Once I mentioned this as a possibility to explain the increase in nighttime temperatures, it didn’t take Steve McIntyre long to find some anecdotal evidence that correlated:

http://cahe.nmsu.edu/news/1997/043097_irrigation_tour.html

A few years ago when cattle prices were high, we saw a tremendous increase in the number of irrigated grass acres in Quay County,” said Jeff Bader, Quay County Extension program director. “One reason was our limited water situation for irrigation, and high cattle prices made it look very attractive.”

When cattle prices dropped again, interest in irrigated pastures declined, but now the cattle market is improving, he said. Producers never really lost concern about irrigated pasture because it fits into the management scheme for water conservation so well in Quay County.

“Irrigated pastures fill a niche in this area because of their ability to produce under varying levels of irrigation,” said Rex Kirksey, superintendent of NMSU’s Agricultural Science Center in Tucumcari. “Pastures remain a viable option in many situations where irrigation water is too limited or unpredictable for corn or alfalfa production.”

What is interesting is that the director of the Ag Science Center, Rex Kirksey, is also the person that took these photos for the station survey. There’s quite a large water project in the area, called unsurprisingly, the Tucumcari Project.

Conchas Dam and Lake

Apparently they have quite a problem with water “disappearing” there as outlined in this report:

“The principal problem has been the loss of over half of the District’s surface water supply in the canal and distribution system that carries Canadian River water from Conchas Reservoir to the irrigated farms in the District.”

“The District’s report concluded that seepage losses from the system’s canals become greater, in quantity, each year.”

From that report I obtained a the study area map, and located where the town and the USHCN station is. Unsurprisingly, the USHCN station is situated close to the canals of the water project, and prevailing winds in the area tend to be from south to southwest:

With increased irrigation to pastureland for cattle, and a leaky canal system that loses half it’s volumes and demands ever more water to meet customer deliveries, it seems plausible that the Tucumcari area is becoming more humid, and with the increased humidity, per Christy, increased night time temperatures.

The studies of Roger Pielke Sr. show that land-use changes are an important factor in local and regional climate change. This effects of the changes in agriculture and irrigation on the local measurement of climate in Tucumcari might very well make a good case study.

I had hoped that the automated weather station that sets next to the Stevenson Screen might have humidity data that I could track, but alas it does not.

There is a fairly complete record though of temperature and precipitation at this link from the Western Regional Climate Center.

UPDATE 7/1/08 One of our commenters “AnonyMoose” has brought this well done historical weather and climate report by NMSU superintendent Rex Kirksey to our attention:

http://tucumcarisc.nmsu.edu/documents/rr751.pdf

This merits some further study for the data it contains. I’m committed to other projects today, so if anyone wants to have a go, be my guest and I’ll post it.

0 0 votes

Article Rating

62 Comments

Inline Feedbacks

View all comments

paulc

June 30, 2008 8:59 pm

Considering the amount of effort required to understand the bias in this and the several other sites that have been published, it is difficult to believe that there is any “global average.” Perhaps if errors or anomalies were random, the mass of data would help, but the data on this site usually shows bias, not random error.
Thank you for the good work!!

David Segesta

June 30, 2008 9:22 pm

Why wouldn’t irrigation cause higher temps during the day too? Do they irrigate at night?

David Segesta

June 30, 2008 9:24 pm

OOPS I see Dr. Christy already answered that question. My apologies.

Brendan

June 30, 2008 9:25 pm

I’ve mentioned it before – using nighttime lights is a bad indication of population. For the granualarity, there are much better sources, including the US census data down to the block centroid. There is also a normalized daytime/nighttime population dataset out of Sandia (? – I don’t have the notes with me – it might be Los Alamos) that has been well researched to the 250 m level. There is also the Landscan USA data that, if ever publically released, would give 100 m day/night pop density. But there are lots of data sources that are much better in the US than night lights…

doug w

June 30, 2008 9:41 pm

I seem to recall that nightlights imagery date back to the 1990’s. Of what use would it be 10 or 20 years later?
REPLY: The night satellite imagery was put together in 1995 by the DMSP

Grant

June 30, 2008 9:57 pm

Some flow data for Conchas Canal here, though they stopped measuring monthly discharges in 1992 for some reason: http://waterdata.usgs.gov/nwis/measurements/?site_no=07223300

Steve Keohane

June 30, 2008 10:44 pm

I added the precipitation line in purple1971-2000 to the earlier precipitation graph in green and it appears that precipitation is up 5-10%. The graphs are small, 1 pixel is about 1%. Hope the link works:
http://i32.tinypic.com/206zpub.jpg
REPLY: Precip might be up due to local humidity being up, which helps in building convective thunderstorms, which in turn dump more rain.

Philip_B

June 30, 2008 11:04 pm

As I mentioned in a previous thread, humidity data can be misleading, as humidity with and without clouds produce dramatically different temperatures. Its the combination of humidity without clouds that produces warmer temperatures and this is precisely the effect produced by irrigation. Naturally occuring humidity is generally associated with clouds which lowers temperatures in relatively warm and dry areas like this.

Steve Keohane

June 30, 2008 11:10 pm

That was my guess. I notice the same thermal effects here on the western slope of Colo. that Cristy mentioned re: higher RH% holding in the heat at night. We often have RH% in the single digits, makes for great summer night relief from the diurnal heat.

crosspatch

June 30, 2008 11:13 pm

Increased nighttime lows is a pretty good indication of humidity. I found some archives of hourly humidity measurements on Weather Underground. Here is January 1, 1988 for example.
There is a link at the bottom that will give you a comma delimited file like this. It should be possible to script a process to collect those files and scrape the humidity data from the html into a spreadsheet.
It looks pretty trivial since there isn’t a lot of formatting other than line breaks.

crosspatch

June 30, 2008 11:17 pm

Also, I noticed that the hourly observations are only from 6am to 10pm but it does list the maximum and minimum daily humidity.
REPLY: Hopefully someone can take on the task of extracting this, I’m overloaded as it is.

Steve Keohane

June 30, 2008 11:28 pm

Phillip_B, here we have little to no cloud cover when the RH% is very low, which I assumed was causal. 20-30% RH feels very humid here. At 6600ft., an overcast deck is often only another 4-6K ft. higher, not quite clearing the peaks. Cloudy winter nights are always warmer by 20-30 degrees or more.

Stef Pugsley

July 1, 2008 12:04 am

So does Hansen just role a dice for each station?
“On a 1-3 I’ll adjust it upwards, on 4-5 I’ll adjust it downwards, on a 6 I’ll make yet another press release about being muzzled.”

Philip_B

July 1, 2008 12:26 am

Steve K, I grew up cool and damp Britain where high humidity plus clouds equalled mild temps, at least in the winter.
The effect I’m describing occurs in warm/hot relatively dry areas like Western Australia. The relative humidity is around 40% to 50% on what I call clear sky humid days when temps go over 40C. Not high humidity by the standards of cool damp places like Britain.
There was a rather dramatic example this summer of cool weather resulting from cloudy humid days. Kalgoorlie about 500Ks from the coast is normally hot and dry in the summer. Typical summer day around 40C. An area of cloud and rain moved in and for 3 days temps didn’t go above 16C. Now this was in the middle of summer.
Anyway the thing about irrigation is it primarily occurs in warm/hot dry climates and seasons. Producing near ground humidity under clear skies. People don’t irrigate when its raining or its cold.
I tried to crunch some numbers to quantify the effect globally but it needs someone with better physics than me.

Atholl Robertson

July 1, 2008 3:00 am

Some basic principles here. Relative humidity is a measure of the water content of the air with respect to the maximum water content the air can hold at the ambient temperature. What really matters is the absolute water content which is given by the dew point. That largely determines the Enthalpy of the air KJ/kg (or BTU/lb)and is much more significant than the air temperature since the heat content of dry air is roughly 1KJ/kgDegC whereas that of the water vapour is more than 2000 times larger. That is why a warm, still day at the coast is uncomfortable compared with a hot day in the desert. The lower the enthalpy of the air, the easier it is to lose heat to it. 40C dry day air probably has comparable enthalpy to the 16C cloudy day. 50% RH at 50C is unbearably uncomfortable. You can cool a lot of hot dry air with not much water but the heat content and the perception of warmth may not change much. The humid air from irrigation will not lose temperature as quickly as dry air. Water vapour is very important in terms of heat content and thermal lag. And of course the water vapour content is related to the so-called greenhouse effect. Moral: Enthalpy is more important than temperature but not as easy to ‘adjust’!

Paul Clark

July 1, 2008 3:13 am

Slightly OT, but has anyone noticed this?
Roy Spencer on Climate Sensitivity:
http://www.weatherquestions.com/Climate-Sensitivity-Holy-Grail.htm
Very interesting stuff around trying to disentangle short-term and long-term relationships between temperature and cloud cover – in both directions.
One of Spencer’s conclusions is that if he’s right about low sensitivity the 1’C of warming we’ve seen (which may be subject to argument in any case) must be largely due to some other factor – and he seems to be suggesting cloud cover changes…
Early in the paper he says that a decrease in low clouds will increase warming. This rings one big alarm bell for me:
Deforestation
From my Permaculture days I understand that tropical forests create their own low clouds by raising humidity through evapotranspiration and providing nucleation points in the form of bacteria. Hence would not massive deforestation reduce low cloud cover and hence increase temperature?
The only fly in this ointment is that most tropical deforestation is in the Southern Hemisphere, which has seen less warming than the NH:
http://www.woodfortrees.org/plot/hadcrut3vnh/mean:60/plot/hadcrut3vsh/mean:60
Also the idea that low clouds decrease warming is the opposite to what you’re talking about, so I may have to think about this some more!
Cheers
Paul

An Inquirer

July 1, 2008 4:45 am

This post is mostly repititive of one that I posted several months ago, but in my visits to Las Vegas over the decades, it seems like the temperatures are hotter there now. The last time I was there, I noticed how green the surrounding land had become — there is a lot of grass and vegetation being grown now in the area, and that grass is being watered daily.

Tom in Florida

July 1, 2008 5:07 am

Crosspatch: “Also, I noticed that the hourly observations are only from 6am to 10pm….”
If this is true then the coolest temperatures of the night are not recorded and omitted from the average. Why?

Daniel Rothenberg

July 1, 2008 5:18 am

Anthony, I don’t know if this was covered in the thread yesterday, but has anyone checked to see if the observation time has changed at the site? You’ve got to be on the right track with the land use investigation, as a “step” in the data almost has to correlate with some sort of systematic change in the data.
REPLY: McIntyre observes that there is a significant TOBS adjustment but that it doesn’t appear relevant. Still, a check is warranted.

Austin

July 1, 2008 5:22 am

Humidity only affects temperature when energy flows into changing the state of the water – not the temperature of the air.
When air cools, the temperature will drop until the humidity reaches 100% and then heat will flow out of the water vapor and into the air as the water vapor condenses. It takes a lot of heat flow out of the air to force the water vapor to condense so the end result is that the nightime low reaches a floor and stays there. The same process will occur in reverse as the air heats up – the water will vaporize keeping the air temp down until it all has vaporized.
Data in 15 minute divisions should show lags in temperature as compared to the pre-irrigation days.

Perry

July 1, 2008 5:53 am

Anthony,
My apologies for posting this item here, but I could not find your email address.
http://blog.snowdonia-active.com/
Very briefly, the blog details the construction of the new restaurant on Snowdon’s summit. On March 18th the construction teams restarted after the winter layoff, but 15th April they were frozen out, much to their surprise and consternation. There are good photographs and the route to the summit is also a great day out. http://www.snowdonrailway.co.uk/news.php
I hope you don’t mind, but it’s another example of the reality of a colder planet.
Regards,
Perry
REPLY: reality bites ‘eh?

Ric Werme

Editor

July 1, 2008 5:57 am

There’s more to elevated low temps than water vapor and the Greenhouse Effect. Dew and frost formation releases a lot of latent heat (condensing water releases 540 calories per gram condensed, freezing water releases 80 calories per gram, and in between cooling water releases 1 calorie per degree Celcius).
During humid summer days in New Hampshire we can use the dew point as a starting guess for the night’s low temp, assuming clear sky and light wind. Overnight air temp follows one exponential curve down to the dew point, then a shallower one from there to dawn. On dry winter nights, the longer night and lower water content lets us sail past the “frost point” which is a degree or so warmer than the dew point and end up well below the dew point at dawn.
I should make .gif files, but my home computer should serve them up okay.
Our first 2008 heat wave has a good example of summertime cooling. From dew start to dawn we only dropped 5-8 degrees F. I had more trouble finding a good winter stretch with low dew points and low wind, but this is good and shows about a 10 F drop after frost starts. The dew point change is more striking, 5 degrees in summer, 15 in winter.
Every 20 F drop in dew point requires halving the water vapor content of air, so there’s only about 1/8 the water vapor to work with in my winter graph vs. the summer.
Dew point is a critical measurement of local weather conditions. It’s a pity that USHCN records pretty much don’t include it.
BTW, the summer graph shows an interesting effect. The 93 F temperature cap on June 8 shows vertical mixing. In the morning the sun heats the ground, that heats the air. Eventually the low level air becomes neutrally buoyant and blobs mix up and down. Now the air temperature can rise only if the entire air column warms up, and that takes a lot more heating. The air exchange also allows winds aloft to mix down to the ground, so when the wind picks up I know we may be close to the high temperature for the day. Another rule of thumb I like is to take the air temperature at Mt Washington (6288′) and apply the dry adiabatic adjustment of 1 F per 200′, i.e. about 30 F down to my altitude and use that as what the high temperature for the day might be.

Gary Gulrud

July 1, 2008 6:09 am

Every adjustment begets another; if the adjustment wasn’t correct it is insufficient or excessive with respect to expectation.
Does any one suppose that the data will ever fit someone’s model or that fiddling with the model will be easiest at every turn?
“Of the heart’s deceipt there is no end; who can understand it?”

Ric Werme

Editor

July 1, 2008 6:22 am

Random thoughts:
1) I hate smilies. It’s not funny that Mt Washington is 6288′ above sea level. 🙂
2) Perhaps correlation studies between HadCrut/GISS and UAH/RSS should concentrate on high temperatures, not average. Since the satellite data is for “low tropospheric” and ground data is prone to inversions, the high temperatures at ground correlate much much better than low temperatures. Also, low ground temperatures are amazing prone to micro climate effects. Someone here used to report his low temp both at 6′ and and inch or so above his garden. On clear mornings the difference was striking.

Stephen McIntyre

July 1, 2008 6:43 am

What could cause that? We can rule out adjustments, since this is GHCN data before Hansen gets his adjuster mitts on it.

Anthony, as you know (but isn’t clear from the above), GHCN (and USHCN) both adjust data e.g. the USHCN time-of-observation adjustment, USHCN SHAP and FILNET adjustments, GHCN has its own adjustment. I think that your point is that none of these adjustments appear to be relevant to the issue at hand, though Tucumcari has a noticeable TOBS adjustment (which doesn’t appear to be material here. )
REPLY: True, and thanks for making this point. When I said “we can rule out adjustments” I was referring to the results on the GISS graph, and yes the GHCN tweaks aren’t relevant to the issue at hand.