How not to measure temperature, part 44

I’ve seen a couple of weather stations mounted on rooftops before, for example the station in Eureka, CA was on the roof of the post office for many years. This station in downtown Oakland, CA is unique because not only is it in on a rooftop, in the middle of an urban setting, it’s also over grass. Only in California would you find a rooftop with grass growing on it.

The large concrete pad and pea gravel/tar roofing nearby likely creates some bias, but is more likely to be swamped by the surrounding urban heat island (UHI) which is downtown Oakland.

See the picture below, click it for an interactive view, then click again to activate the Java panning applet.


Click for interactive image – Photo courtesy NWS San Franscisco/Monterey

This is COOP station 04-6336, which fortunately, is not part of the climate observing program. Still I thought it interesting and unique enough to bring to your attention as an example of a UHI centric measurement site.

Advertisements

9 thoughts on “How not to measure temperature, part 44

  1. Very cool panning and the map showing the view angle is great. This station gives new meaning to the term “concrete jungle.” What’s that in the CRS – a still, a fire extinguisher, a ????

  2. I’ve recently started out learning about global warming, so forgive me if I ask questions others have already asked. I’m interested in the veracity of temperature data, which brings up the UHI effect. Naturally I came upon your site and am reading it.

    I’ve read your description of Marysville station and your comparison of Marysville station to Orland. Here are my thoughts.

    Marysville’s population in 2000 was 12,268. Hardly a big town. Is Marysville big enough to create an urban heat island effect?

    It’s estimated Marysville’s population shrank between 2000 and 2006 from 12,268 to 11,949, a 2.6 percent decrease. The population of the 95901 zip code decreased by 13 percent between 1990 and 2000. Why would an urban heat island effect increase while the population is decreasing?

    I assume the air conditioners weren’t running during cold weather. Does the station data show an increase in temperatures during cold months when the air conditioners probably weren’t running? If so, that increase can’t be explained by heat from the air conditioners. Similarly, the warming effect of concrete, if any, should be less in wintertime. But there might be more reflection from the cell pole.

    You present the Marysville chart big but present the Orland chart too small for me to see the numbers. When I enlarged the Orland chart using my Preview program, I could see the Orland chart has a full degree C between tics, unlike the Marysville chart which has only half a degree C between tics.

    The temperature at the Orland station rises about one degree C after 1960, from around 16 C to around 17 C. The temperature at the Marysville station rises about 1.5 degrees after 1960, looks like to me, from around 16.5 C to around 18 C. What stands out most to me is that temperatures at both stations increase after 1960.

    I’ve read that one way to determine whether the UHI effect is significant is to compare temperatures during breezy days to temperatures during calm days. On breezy days the UHI effect should be reduced compared to calm days. I’ve read that when that comparison was done it was determined the UHI effect is too small to invalidate data.

    I haven’t read your entire site yet, so I may be premature in asking this question: Have you tried placing temperature sensors at various distances from each kind of offending object — concrete pads, asphalt, air conditioners, parking spots, barbeques, etc. — in order to determine the effect of distance from these things on temperature readings? You assume proximity to concrete patios, etc., boosts temperature readings, but have you demonstrated this?

    I’ve read that the main effect of UHI is due to buildings blocking the radiation of heat into space. It seems to me concrete and asphalt on the ground won’t block that heat loss.

    REPLY Barry, excellent questions. I’m on the road today, so I don’t have time for detailed answers, but will do so this evening or tomorrow morning, in the meantime one quick point. The graphs from Orland and Marysville are from GISS and you can see them both in their original format by clicking the links on http://www.surfacestations.org main page near the photos. I did not modify the format of those graphs and I’m not sure why GISS created them at different scales.

    In the meantime, perhaps our venerable crew of readers can chime in on your questions?

  3. Barry,
    I agree — great questions! In no way am I an expert, but I have been drawn into the AGW debate over the last couple of weeks and find it a great opportunity to use analytical abilities. I read both pro-AGW concerned sites as well as the skeptics, and I would offer a couple of thoughts on your questions. UHI can easilty increase as population dips. First, it is doubtful that the amount of concrete and asphalt decreases as populations dips. Second, people per household is probably decreasing, not the number of households. Likely the number of housing units has gone up, and they all are being warmed or cooled. Third, miles driven per vehicle in general has gone up by more than the 2.6% drop in population you cite. Therefore car exhaust probably has not decreased. Fourth, the number of people commuting to work in the city may have gone up even as the population ot the city has dipped.
    Also, a city of UHI. We humans affect local temperatures in many ways besides greenhouse gases. From buildings & vehicle exhaust to soot & dirt in the snow — they have an impact. I grew up on the farm, and it was always colder on the farm than in the nearby town of 1000 people.
    Those are comments on your first two questions, and now I must run.

  4. Barry, population is not the only driver of “urban heat island” effect. There are other factors – buildings, roads, etc. Often when you focus on one single factor you do so at the expense of seeing the big picture. So tell me – what OTHER changes happened in 95901 during the period you mentioned? For instance, did a greater percentage of the population switch to SUVs or hybrids? Did more families get second or even third vehicles? Perhaps those aren’t good choices to use as examples (it might seem like I’m arguing the wrong case) but I think you get my point. In case you don’t, to sum up, the urban heat island effect is influenced by a lot more than simple changes in population.

  5. Marysville is a constricted population place. City layout pleasantly convenient for 1850’s. Bound on 2 sides by rivers, the third side is a levee/farm land. Business concerns are or have moved south and west to other municipalities for ‘elbow room’. The daytime population is larger than nighttime. 1960 forward, hospital (significantly larger today) and government operations seem to be the why-people-go. Heavy traffic congestion during daytime. East-West and North-South state highway interchange. Night time I would classify it as an asphalt ‘ghost town’.

  6. Barry:

    The unaccounted-for factor is the microsite violations. That is the gold speck, here. They are not UHI-adjusted (those adjustments are lowballed, anyway) because they do not show up as lights at night.

    That, in a nutshell is the Rev’s great contribution here. He has won my respect, not to say admiration as a result. I never questioned the data itself until I encountered this site. But he went out and actually examined it–what a novel idea! One that had never occurred to the NOAA, at any rate.

    The implications are absolutely staggering.

    Over 5 out of 6 surface stations have microsite violations of 2 degrees C or more warming bias.

    Most of these violations are recently introduced by two factors:

    1.) The MMTS switchover and

    2.) the exurban creep (with astounding incidental violation) that is overtaking the surface stations at a far greater rate than they are overtaking the earth.

    Most of these violations have occurred since 1980, and the effect this has had on the recent historical climate record one can well imagine.

    Consider that the–entire–temperature rise of the 20th century is alleged to be between 0.6C and 0.8C!

  7. Barry

    The Marysville / Orland comparison is interesting.

    Sometimes pictures “are worth a thousand words”. But pictures supported by figures are worth even more.

    The station in Orland, CA is well-positioned and well-maintained., with a grass surface and no buildings or paved areas nearby.

    The site in Marysville, CA, however,”has been encroached upon by growth in a most serious way by micro-site effects”. These include: asphalt paved parking lot, buildings, and air conditioning units directly next to the measuring station.

    The impact of the Urban Heat Island (UHI) effect can be seen by comparing the two temperature records from GISS NASA station data:

    http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=425725910040&data_set=1&num_neighbors=1

    http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=425745000030&data_set=1&num_neighbors=1

    I downloaded the NASA GISS data and plotted them in Excel to make the comparison. If we compare the two official records, we see that Marysville (the poorly positioned and maintained station) shows a higher trend of temperature increase of around 0.23 degrees Celsius per decade (5-year Annual Average over period 1981-2005) as compared to the well positioned and maintained station at Orland.

    This spurious difference between the temperature trends of two fairly close stations is higher than the IPCC’s trend itself for the past 50 years (0.13 degrees C per decade) or its projection for the next two decades (0.2 degrees C per decade). In other words, the error is greater than the trend itself.

    This is just one of many examples cited by Watts which confirm a significant UHI distortion of the US surface temperature record.

    If the work by Watts were the only study showing a significant UHI distortion, we could conclude that it represents a non-representative anomaly.

    There are, however, many studies that all come to the same conclusion of a significant UHI distortion of the global surface temperature record.

    I know that IPCC has reassured us that the UHI distortion is “less than 0.006 degrees C per decade over land”, based on studies by David Parker comparing calm nights with windy nights, which have since been questioned, but I somehow have a real hard time believing this, in view of all the evidence out there to the contrary.

    Regards,

    Max

  8. Pingback: How not to measure temperature, part 81 - roofing the past in Columbia « Watts Up With That?

Comments are closed.