Apologies to singer Tony Bennett for the title. Our own “Charles the moderator” writes in with this curious picture from Weather Underground showing downtown San Francisco at about 2:15PM September 8th, 2015.
The disparity of temperature just a few blocks away is quite stunning. In the two temperatures circled in the downtown area yesterday there is a full 14°F (~8°C) difference. Charles adds this note:
These two spots are about 6 blocks apart. No fog and no wind.
The no wind part is backed up by the many circles showing no wind barbs.
One would expect a normal temperature gradient of sorts from the shore to the inland area, due to the influence of the bay, but this gradient is perpendicular to that expected gradient. And there are other examples. At Bush/California Street we have 80°F and a few blocks away at Divisidero and California we have 93°F in the opposite direction of the expected natural gradient. Near Haight-Ashbury, we have an 86°F reading and a 94°F reading just 4 blocks away. In the Richmond District, we have 80°F and 94°F in the Inner Richmond District. Then there’s Mission Bay at 97°F and Potrero Hill at 80°F, and these two stations are approximately the same distance from the water of the bay.
Lest anyone think this is a one-off fluke, I checked this morning at 8AM PDT as I was writing this essay. The screencap is below, note the 81°F and the 67°F near the center, and other examples:
So what is the correct temperature for San Francisco at any given time? I have no idea. Pick a number, any number.
Given that Weather Underground has many weather stations logging data in real time in many cities worldwide, I expect readers can find similar examples elsewhere. It should also underscore that any attempt to use this personal weather station data for science or climatic purposes comes standard with wide error bands.
And despite the usual protestations, siting does matter when it comes to measuring temperature, as this map proves. Cue Mosher rant in 3…2…1
So which temperature should we use to calculate the world ending 2C increase?
No urban temperatures should be included in world temperature data. If we attempt to compare present temperatures with temperatures a couple hundred years ago, or even earlier, we should only compare areas that are similar in landscape to end up with a reliable assumption. Again, no urban temperatures should be included in world temperature data if one is to make any comparison between past and current temperatures.
Absolutely correct. And this is the giant flaw that is studiously ignored.
If we eliminated all that was flawed, there would be nothing left, and Mosh would have a hard time claiming we know enough to act now.
I watch my little Wunderstation (ICUMBRIA1) like the proverbial hawk and this morning’s temp graph demonstrates something but I’m sure quite what.
Situate at the end of a large garden on the SW side of a small (Pop~5,000) town and looking SW over farmland, all of it grassland for cows/sheep. Not a lot of trees and alt~100m. Most definitely not any sort of ‘built-up-area’
An English ‘A road’ about 600 metres away but almost zero night-time traffic. No industry worth talking about, esp not at night, sleepy sleepy sleepy.
All night the wind was drifting N to S, with 5km/h ‘gusts’, otherwise calm.
Every time the wind moved, the temp moved, by about 1degC over 30 mins then drifted back down to the night-time cooling rate.
Even after sunrise, the graph became a (rising) mirror image of the night-time, bumpety bump bump.
If that can happen, at night, in ultra-rural here, how on Earth can that mega-event at daytime Heathrow last month be trusted?
I’m confident the little gusts were disrupting the temperature inversion caused by radiational cooling. At least, if it was clear overnight. What generally happens to me in New Hampshire on clear, windless nights includes:
Dawn: low temp for the day is reached, grass has dew or frost. No wind because the temperature inversion blocks what may be only 10-100 feet above me.
Morning: Temperature climbs quickly, grass dries out, inversion weakens and then get blown apart. The wind now mixes down to the surface, and the sun has to heat a much larger volume of air to raise the temperature.
Afternoon: warming either continues well in to the afternoon or it reaches a plateau if we have a fetch of Canadian air. Radiational cooling peaks along with the temperature. (Duh, but your mother probably never taught you that.)
Evening: after dark, radiational cooling continues. As the ground cools, the air cools, the temperature inversion gets reestablished unless it’s a windy night.
Repeat.
Rick, the IR thermometer still shows my driveway warmer than other surfaces at sunrise (and far warmer the the sky).
Then a question, how do you tell how thick the surface inversion is?
There was a similar situation in the SW part of the Denver area where I live. There was a station that was often 15-20 degrees higher then it’s closest competitor. I complained once to Weather Underground and the station was removed promptly. Although is showed back up several days later.
So, let us consider the krigging method of determining an unknown temperature, let’s say at the corner of Market St and the Central Fwy in the second map — the station shows 67….the surrounding “known” temps are 81, 74, 72, and 70.
Using this marvelous tool for determining temps of missing stations, we would get some temp (this is just a guess, mind you) of around 75 or so (the missing point being closer to 81 than the 72 and other 70s).
Krigging or other homogenization methods can never ever return a result that is higher or lower than the highest or lowest surrounding temperature (or monthly average, which is what I believe is “krigged” in the BEST method).
Look over the maps shown here (ignoring the scale, assume that this is a national or continental map) and find how many, what percentage, of stations show temperatures outside of (above or below) the range of the four nearest surrounding stations.
Ashbury Heights at 74 — nearest neighbors 69, 71, 70, 68
The centreal station marked 81 — nearest neighbors 72, 7, 73, 70
16th and Folsom at 64 — nearest neighbors 70, 78, 72, 74, 67
Lower Hasight, just off Castro st at 75 — nearest neighbors 71, 64, 70, 74, 73
That’s five at just a quick glance, off this little map — all five stations, if they had been missing and their value determined by krigging/homogenization, would not even be close to the actual value, throwing any average of “all stations” off by a large margin.
Just sayin’.
And there is much said about just needing Latitude and Altitude to get really close, except I live where the jet stream flip, flops back and forth over heard, making a 10 or 20 degree temp differences, and it can do this over a day or two.
See the dew point, a good indicator of the source of the surface air
You don’t get anything much more chaotic than the jet stream. How do you krig and homogenize around a dividing line that is constantly moving?
CLIMATE IS NOT WEATHER!!!
err…wait….
WEATHER IS NOT CLIMATE!!!
Wait, that isn’t quite right either….
IT’S A DRY HEAT!!!
….never mind….
Per WMO and IPCC AR5 glossary climate is weather averaged over thirty years.
For what that is worth.
Climate is most accurately defined as a taxonomic ranking of Biomes (biological communities that are shaped by the regional environment) that are easily recognizable at a glance. They are mapped with great imprecision. http://oceanservice.noaa.gov/education/pd/oceans_weather_climate/media/climate_zones.swf
Most people, without a great deal of schooling, can easily tell the difference between Tropical/Temperate/Polar, Wet/Dry, Maritime/Continental, Desert/Mediterranean/Rain-forest. That is because Biomes tell us directly what the average weather is over a lot longer time frame than simply 30 years. Without using thermometers or records. Imagine that!
Climate science: Hijacked by politics.
That would be for “weather” measured by instruments which over the 30 year span were far from (and not downwind of) any significant land use changes. The main problem with our climate records is that far too many weather stations have had significant land use changes very near to them. There are a significant number of weather stations which have survived with little if any land use change, however these are homogenized with the stations which have had significant land use changes. I hypothesize that even small towns (~5,000) have a measurable UHI. Any of these small towns which are either these near jobs or near enough to urban centers to encourage development have had urbanization over the past 50 years.
“I expect readers can find similar examples elsewhere.” Absolutely. I live in a suburban/semi urban area in Michigan. The Weather Underground website provides readings for at least 10 nearby stations. The Official Weather station is at the international airport about 13 miles away. On some days I’ve seen variations as much as 10 degrees within 1 mile….. and none of the stations agree with the official temp …. but the normal variation is about 3 degrees (Fahrenheit). Who ya gonna believe?
None of the above in the case described, certainly not Weather Underground.
I lived in SF for 5+ years, and I see no surprises in this picture.
First of all, the color key at the top shows that these pictures mix temperature reports spread out over the length of a day, as well as over the size of the city. But more to the point, this is a coastal city at the mouth of a large bay, so of course there are going to be massive air flows through it on a twice-daily basis.
I lived near the top of Twin Peaks, the highest point in the city, which is about at the point Market St. runs off the bottom of your map. Go up there in the middle of the night, no matter what part of the year it is (I suggest the parking lot at Portola and Clipper) and you can watch the fog blow in from offshore — easily fast enough to see — until by about sunrise, it has covered everything. Of *course* this carries large temperature changes with it. It couldn’t possibly be otherwise.
Actually, coastal stratus is fairly rare during the winter / rainy season. Not enough upwelling and very little onshore push. During the dry season, yes, you are correct.
I was in town all day the day of these readings. Dead calm, very un SF like.
We’re already getting into Fall mode. Several other indicators of that. For example the nearly dry cold front about 10 days ago. Fronts like that usher in Fall. Who knows, we may lose the fog early this year, overall. Maybe little to no fog after the end of the month.
I had the opportunity to spend a few months in San Mateo at the end of the last millennium while moving a data center from downtown San Francisco across the bay to Hayward, and the experience was an eye-opener for the definition of microclimates. The city sits on the north end of a mountainous peninsula sandwiched between the cold ocean currents of the Pacific Ocean and a warmer San Francisco Bay, with the hot desert interior of California just a few miles away. It is possible to draw a line through San Francisco only 30 miles long where the western end is shivering in the cold 55°F Pacific waters of the California current and the eastern end is baking in the 100°F desert east of San Leandro. In between, the altitude changes from sea level, climbing to between 500 and 1500ft (depending on exactly where you draw the line), back down to sea level, up to 1500ft, and back down to 500ft in the hot interior with the temperatures fluctuating wildly along this line from generally cold to generally hot. The building we were vacating in South San Francisco was so cold there was no air-conditioning, except in the data center. The building we were moving into at Hayward just 14 miles away was always too hot, despite having a kiloton of air conditioning.
We used to pick our lunch venues in San Francisco according to how we were dressed … too cold, move east a block or two; too hot move west a block or two. It was crazy. Fortunately enough, the Ghirardelli Chocolate Factory in Ghirardelli Square situated on Beach St near the Hyde St cable car turntable seems to always be at that perfect temperature where the chocolate is just below its melting point, surely one of the great happy accidents of microclimate geography.
But don’t take my word for it… the source of all knowledge has an article on microclimates that mentions San Francisco … https://en.wikipedia.org/wiki/Microclimate
So yes, nobody can ever agree what temperature it is in San Francisco, because two people standing on opposite sides of the same street can measure temperatures that differ by 5°F. If you don’t believe me, then I suggest you walk around the town – just make sure that one of your stops is Ghirardelli’s.
Where I grew up in Ohio, there are no microclimates like we have in the West. California temperatures can vary by 30 F within 15 miles or less. It is currently 74 F in Morro Bay and 100 F only 17 miles to the northeast in Atascadero. 10 miles south of Atascadero in Santa Margarita, it is 105 F (all from Weather Underground).
But we can have 3 or 4 seasons weather in a day or so without going anywhere (due to the movement of the jet stream).
The NWS reported the temperature extremes for the U.S. yesterday were both in California. Boca Reservoir at 26F and Death Valley at 116F. So if you drive a little farther than 15 miles, you can experience a 90F temperature swing from low to high.
I’m sure BEST will be able to sort out San Francisco with their famous scalpel …
Re: Ric @1058
I remember Santa Clara Valley when It resembled the writing of Jack London.
Love New Hampshire too. Went camping there when stationed at the navy base in Newport, RI. Another good place to sail except the winters are brutal. Sailed in shirt sleeves in February in SF Bay. Also needed my winter coat in July.
I lived a lot of places as an adult but swore I would not go back back to California with children because of the schools, drugs, and gangs. When you have just spent a brutal inter working near Detroit, and you that job offer from Cleveland is looking good; then you get a call from a nuke plant in California, it look like a good choice. I thought the schools might be better in a small town. Half the teachers are draft dodgers or quitters.
Because of politics, the plant I worked closed. Subsequent to California, we have had school age children in Washington State and Virginia. Great school and a nice house only requires one income. Good schools, low taxes, and no crime. It is a pleasure to watch county government work well by serving the community not some screwball social justice agenda.
I think this emphasises one of the flaws in the alarmist position. Who cares about a change in temperature of 1,2 or even 4 degrees. The people who live where the temperature is recorded as 14 degrees higher I suspect haven,t been fried. I defy anyone to accurately know the difference between the temperature now and 50 years ago without reference to a thermometer. The 2 degree variance plucked by warmists is less than the adjustment people make by raising or lowering their air conditioning of their homes yet we don’t have any repercussions. When you have variances of the magnitude highlighted in this article you realise ( or a person not indoctrinated by AGW theory) that just the error bars of local temperatures are probably greater than the amount of warming that has occurred in recent history. When you can’t vouch for accuracy of data how can global weather bureaus vouch for the accuracy of adjustments. It’s a total farce.
Thats just the san fransisco marine/fog influence. Just drive across the golden gate and 4 miles north at the same elevation and it could be 90f while its 50f at the wharf in the city.
Not this day, Dead calm, no on shore flow.
Anthony, as an air quality meteorologist for over 40 years, I totally agree with your statement “siting does matter when it comes to measuring temperature”. The sensor shielding, aspiration, and placement at the site also matter greatly, as well as sensor quality control and quality assurance. And the siting for one purpose may not be suitable for another.
Here’s a photo I took of an air quality and weather station placed at the top of the tallest building in Houston back in 1986. The shielded and motor-aspirated temperature sensor can be seen about 2 meters above the top of the building.
https://www.flickr.com/photos/oz4caster/8060698271/
I enjoying checking weather measurements on the WunderMap, but most of these measurements are not very representative of larger scales, especially for wind, and many are essentially micro-scale monitors that do not represent larger areas very well. Usually there is little or no information about the monitor siting, although if the location is accurate you can zoom into the aerial imagery view to get a rough idea. With your detailed analysis of USHCN monitor siting I’m sure you are acutely aware of this problem.
This map raises some interesting questions about estimating “global temperature” and associated anomalies. Representativeness of selected inputs is critical. I personally like the CFSR approach. I can’t imagine using only the GHCN data as input to drive real-time weather models for surface temperature. Not that the CFSR data are perfect either, but there is much better coverage. It’s interesting that the CFSR data show more of a downward trend in recent years compared to GHCN derived NCEI estimates.
There are many locations that present a challenge for the meaning of “surface temperature” measurements, and and urban street canyon environments are one of them. Vast jungles are another and probably cover much more surface area than urban environments. How do you represent the surface temperature in the middle of a jungle or in the middle of skyscrapers or at the top of the numerous unmonitored high mountain peaks around the globe? Again, I like the approach of using the data we use to initial the global weather forecast models to provide our best general representation that we have for surface temperature.
As you were an air quality meteorologist for over 40 years, I would like to know what the precision was of the instruments that you used, especially thermometers? I used calibrated glass thermometers in the oil industry for over 25 years, and even the expensive ones had an accuracy of +/- 0.5C…
Mark, I worked in the public and private sector and the temperature measurements associated with air quality measurements were obtained using shielded, motor-aspirated electronic temperature sensors logged on data loggers as five-minute averages and processed into hourly averages. The sensors had an accuracy of about +/- 0.5C. The public temperature sensors were usually mounted on towers with the wind sensors at about 10 meters above ground level. The private group monitored temperature at about 2 meters as in the photo I provided. Both groups provided annual independent audits and data validators reviewed the data for quality assurance.
Another thing to consider is the temperature sensor/computer interface.
I was responsible for the decay heat removal system at a PWR nuke plant that had been shut for an extended period of time. It was common practice in the navy and BWRs to cool the plant done, turn off the pumps, and allow natural circulation to mix the water while the plant slowly heated up. At some point ambient loses would be greater than decay heat so cooling pumps would not have to be restarted.
I could program the plant computer to record incore thermocouples. I noticed there a were random temperature jumps of about 10 degrees. The computer recorded data set every two minutes and the temperature jump would be gone. Of course the temperature did not change.
Error is normal when measuring things. I had tanks containing boric acid that we sampled periodically. There is no mechanism to change the concentration (by design) but the chemistry guru tells me to expect out-of-spec samples. It is a statistical thing. Take another sample and it is very unlikely statically for the random reading to repeat twice in a row.
I when we do stuff at a nuke plant, we check to see if we get the expected results. I was running a test and got the expected results. The NRC observing the test goes running down the hall going OMG, OMG. Why was I not concerned because the test results did not match the expectations of a Washington DC bureaucrat? My agenda is based on science and years. While there are things we can not take credit for in modes, real life is not a model.
I’ve had a Wunderground station for almost 8 years. Located in Southern Vermont, the elevation is around 1700 ft. Other stations in the area are located in the surrounding valleys. My station is the default for reporting by Zip Code lookup, as there are no others closer. However, it will not show up if it is a calm morning and the valleys are cooler by more than 3 F than my station is reporting.
I talked to Wunderground about this behavior and they told me they essentially turn off reporting for a station if it is outside some magical boundary compared to it’s peers. They have no concept of elevation or station location (in some of the other spots, the sun doesn’t rise till after 9AM.)
Note that this data censure only happens if you ask for it by location, like Zip or City or Airport. The station is always available on-line, it just is not considered in the overall location average. In my case, being the only station for a Zip, it is ironically too apparent.
As an avid paraglider I can tell you that the tempature differences are quit common in nature as well. Depending on the Suns angle, the materials being heated, the wind pushing thermals together in certain areas. You may note that, within an hour or so, that gradiant might disappear altogether and another one pops up some place else.
Just to make sure, and I apologize if someone else has already asked, but are all the readings reported at the same time? Is there any possibility that the lower temperature reading lags by a few hours? I’ve seen that happen before….
Maybe I missed it, but what did the Weather Underground report for the 2:15PM September 8th, 2015 official temperature of San Francisco? Looks like they could have used 77 or 97 degrees – which did they use?
The temperatures are “all over the map” between 77 and 97 for that afternoon.
The NWS used 91F for Sept. 8th.
I take that back. The 91F was for Sept. 9th. They reported 92F for Sept. 8th.
Thanks Richard. At least they didn’t use 97 F…
So what is the temperature variation between the top of the buildings and the ground? Can the shaded side of the top of the building provide cooler denser air which sinks along the face of the building into the warmer system close to the ground? An often used phenomena when sailing to use the katabatic winds off of hills near the shore line that provide more favorable winds. At the farm we know that cold air flows down off hills and forms pockets that are significantly colder and may harm fruit trees.
In the 1940’s my Grandfather would fire up the fans in the Citrus Orchard to keep the cold air pockets from forming.
I always heard it was a hot time down in Hashbury…
From what I can tell about reading Hadcrut, Berkeley Earth et. al. methodologies, they claim to be removing stations randomly and seeing if the final result changes. I’m not completely sure about this, the papers are a bit vague unless you are the one who wrote it (a common problem with all sorts of topics).
However, the law of large numbers still applies. Yes, you will get bullcrap data for San Fransisco, because there’s not enough stations, but over the planet there are enough stations. The accuracy still increases by some constant (error bars for a single station) divided by the sqrt(N).
My Monte Carlo analysis on auto-correlated surfaces show that constant is 2.5x that of the assumption of Gaussian uncorrelated noise in those papers. However, the e/sqrt(n) still applies. So given the fact there’s thousands of stations across the planet, they can still get more accuracy than you see in San Francisco.
Don’t confuse the forest for the trees.
Peter
Well, whatever it is its hot. Working over in Alameda today right on the water where normally, it is quite shangra-la like. Not today. Never fear though global panickers, low 70’s by Saturday.
Best way to get Mosher not to appear?
Challenge him to appear.
Very clever on your part.
If I could answer for him he would say the UHI effect is already built in to compensate for city warming.
And that it has no noticeable effect anyway.
Several times I’ve observed “bubbles” of warm air moving along the surface and not immediately mixing with colder air. I felt this happen with a moderately strong wind pushing warm air over a cold lake and its a strange phenomenon to feel the air temperature change 10F-20F every 2 seconds. No gradual temperature gradient whatsoever, instantaneous changes. A similar thing can happens with water when warm river water enters a cold lake, sometimes there can be no gradual temperature gradient from warm water to cold, the warm water stays can stay intact and the temperature change from warm to cold can be instantaneous. Its strange to think of warm air/water and cold air/water behaving like oil and water and staying separate but it does happen. I would think when air is stagnant its also possible for “bubbles” of warm air to form and stay relatively intact with little to no mixing with surrounding cooler air, just an instantaneous temperature change when you move out of the warm air “bubble”.