Godowitch et al 1985 on Urban Boundary Layers

Today I want to talk about a terrific 1985 article Godowitch et al,, 1985. Evolution of the Nocturnal Inversion Layer at an urban and nonurban location, J Clim Appl Met 791 ff available online here, which helps put some of the UHI discussions in a more complete perspective.

Instead of just considering UHI from a snail point of view (ie. transects from city center to country along the surface) it considers the entire vertical structure over the course of a night, which turns out to be richly textured and to offer much insight to what’s going on at surface.

This is all obviously old hat to Pielke Sr who seems to have cut his teeth on boundary layers, but is helpful to people like us coming into the middle of the debate. (It might have been useful to Parker as well, who also seems to have come into the middle of the debate arriving from prior study of SST bucket adjustments.)

The concept of the “nocturnal inversion layer” is distinct from but related to UHI, with the description of rural settings being as useful as the urban ones. In rural locations, a very strong nocturnal inversion at surface develops rapidly at sunset; this doesn’t happen in urban settings (where only an attenuated inversion develops at an altitude above the surface). In fancier language from Godowitch et al:

” A surface-based radiation inversion layer is a common feature in temperature profiles at night in natural environments. In contrast the nocturnal thermal structure in the highly inhomogeneous urban environment usually consists of an isothermal to adiabiatic region, the urban boundary layer (UBL) which is capped by an elevated nocturnal inversion layer (NIL).

So think of the “urban heat island” not as an “island”, but as a type of dome, keeping in mind that if you get high enough aloft (about 400 m in 1975 St Louis, the urban effect is attenuated or nonexistent.

Godowitch et al (1985) reported on a program around St Louis MO, in which they obtained vertical temperature profiles by an instrumented helicopter during 31 evening and 46 morning runs over 5-6 week periods in July-August 1974-1976, mostly under fair weather with surface wind speeds of 1-3 m s-1. Eleven flights with broken or overcast conditions were excluded from the reported calculations.

The urban (141) and nonurban (401) locations are shown below. They noted that the nonurban location was not ideal, but still had very strong contrast to the urban site. The urban site was in a commercial and multiple story building district about 3 km west of the downtown center; the nonurban site (401) was about 7 km southeast of the city at Bi-State Park airport, where it was surrounded by forested and agricultural land. (It would be interesting to do a followup comparison today.)


Their next figure shows the nocturnal evolution of inversion in deg C for both urban and nonurban settings. The nonurban inversion reaches 3 deg C, while the urban inversion is only about 1 deg C. The nonurban inversion begins to develop immediately at the end of the day, while the inversion develops later in the urban setting. (A note here – keep in mind that at a relatively low altitude, 400 m or so in this case, the differences between urban and rural settings are strongly attenuated or eliminated.) Maybe someone can comment on whether these nonurban inversions are characteristic in other environments (e.g. oceans, tropics).


Their next Figure, Figure 3 (below) shows night-time evolution of the boundary layer in nonurban (top) and urban (bottom) settings. In the nonurban setting, a slight inversion has already developed by 6 pm; in this example, it reached nearly 4 deg C by 5 am, with the height increasing to slightly over 300 m by dawn. [Note – inversion here is a vertical differential and is related to, but is distinct from the horizontal urban-rural UHI differential.]

In the urban case, at 6 pm, the lapse rate above 200 m is pretty similar to the nonurban case, but there is no inversion. Indeed, at ground level, in the urban situation, no inversion develops; in each case, there is a lapse rate (of varying steepness) for the lowermost 100 m or so. At 6 pm, there is already a surface differential (UHI) of a little over 1 deg C. At 11 pm, the two are similar at 400 m; the nonurban is a little warmer at 200 m with strong inversion to surface, while, in the urban setting, there remains a somewhat adiabatic lapse rate near surface with an attenuated inversion.

From a snail perspective, the horizontal urban-nonurban differential (UHI) has reached about 1.5 deg C. at 11 pm increasing to about 2.2 deg C by morning. This may not be representative as elsewhere in the article the UHI is said to max out earlier in the evening.


In Figure 2 above, the altitude of the nonurban inversion top h_T is shown as about 300 m. Godowitch et al estimate this height by reference to underlying physics using the formula:

h_T = 2 (K_T t)^{1/2}

with K_T being a roughness parameter which they estimate at about 0.75 m^{2} s^{-1} . They also observe that the amount of surface inversion S (in deg C) also develops according to a square-root time formula, see graphic below:

In addition to the thermal “dome” structure, they also observed a type of “inversion” structure in wind speeds in which there is a wind speed maximum )defined by height h_u becoming well-defined aloft about an hour after sunset and increasing until midnight. They note that the height of the wind speed maximum was higher in the urban setting.

While their analysis is specific to St Louis, they compare the inversion structure in St Louis to larger and smaller cities as follows:

Statistics for New York City (Bornstein 1968) which is considerably larger than St Louis in horizontal and vertical extens show an average h_u about twice that for this urban site, while the average S was about the same. Thus areal size and the vertical dimension of a city are important factors that determine the maximum heeight of the nocturnal UBL. Consequently a shallower average UBL is expected in smaller cities as observed in Cincinnati Ohilo (Clarke 1969) and Columbus Ohio (McElroy 1973) where h_u was typically 100 m or less in the downtown area.

They provide the following summary of site-specific parameters:

If you go to Google Scholar and see studies that cite Godowitch 1985, you’ll not see a lot of studies (none by Parker) but in a quick browse they tend to be empirical and interestng. They include studies of Moscow and Melbourne. HEre are a few notes from their abstracts.

M. Shahgedanova, Burt and Davies 1997 report for Moscow (1990) report:

The urban-rural temperature differences ranged mainly between 1° and 3° C, with an absolute maximum of 9⵸°C. In summer, the heat island intensity exceeded 3°C on 29 per cent of all early morning observations, confirming the widely held view that anticyclonic conditions generate strong heat islands. Temperature variations within the city were small, with a notable exception of the urban park; in winter, the lee periphery of the city was often warmer than the urban centre.

Morris and Simmonds (J Appl Met 2001) study Melbourne using as “nonurban” comparanda the three airport sites of Melbourne International Airport, Moorabbin Airport, and Laverton Airport (a; BoM station Nos. 086282, 086077, 087031, respectively).

Relevant discussions at Pielke Sr include Pielke and Matsui 2005 and two blog entries here and here.


  1. Anthony Watts
    Posted Jun 16, 2007 at 1:02 PM | Permalink

    “Temperature variations within the city were small, with a notable exception of the urban park; in winter, the lee periphery of the city was often warmer than the urban centre. “

    Like I said in the Parket thread, it appears all Parker did was prove wind spreads around the UHI effect a bit.

  2. Steve McIntyre
    Posted Jun 16, 2007 at 1:13 PM | Permalink

    I get the impression that in many settings there might be a relatively distinct gradient at the outskirts of the city which tapers off to something more gradual towards the center.

    The issue for these urban airports is then whether some of them have been folded into the urban “dome”. It looks to me, as I’ve said before, like airport and airport suburbs where you have an urbanization trend might well have the worst of all possible worlds and an index that stayed downtown in a park-like setting might well be more reliable. I recall seeing some comment like that about Vienna.

  3. Gary Kobes
    Posted Jun 16, 2007 at 1:43 PM | Permalink

    Steve, I lived in St Louis from 1969-2001. If I remember correctly an earlier study of the “urban heat island” was done in St. Louis around 1970-71. It used an instrumented fixed wing aircraft. I don’t recall any further information but something may be out there.

    You queried about changes since the study. The site just west of downtown was most likely on or near the campus of St. Louis University. The City of St Louis is quite old and at one time had very dense population. About 900,000 in the 1940’s and declined to about 600,000 at the time of the Godowitch study and today is around 300,000. Most of the structures are masonry and stone. Although there are a fair number of trees in yards and parks, the urban area is basically one big heat sink. In July and August temps and humidity are just about unbearable. I would say the area of this site is fairly typical of the city. Alsthough the population has declined, most of the structures remain and some are being renovated.

    As for the Bi-State Airport, from a casual observation standpoint (private pilot), the area is little changed in the immediate airport environs. I would characterize it as an “urban edge” airport with slowly encroaching residential and industrial but landfrom limitations that will impede dense development.

    The northern airport, St. Louis Regional in Alton, IL is also an urban edge airport. Much open farmland to the the east and south with generally suburban residential use on the west and north.

  4. Anthony Watts
    Posted Jun 16, 2007 at 1:57 PM | Permalink

    RE2 a good example of the airport being folded into the UHI dome is Petaluma Airport, which Russ Steele surveyed yesterday. Look at the Google Earth map and the homes to the south.


    Of course the MMTS tied to the wooden deck and just east of 2 A/C units on the wall and the acres of tarmac have an effect too.

    The prevailing wind direction in Petaluma is from the south to southwest as prevailing westerlies push through the Golden Gate and spread out into the coastal mountain valleys.

    Here’s GISTEMP

    Russ notes in station survey that station has been moved from Fire Station to Airport, MMS notes it but GISS hasn’t updated records, they still refer to it as Fire Station #2

  5. steven mosher
    Posted Jun 16, 2007 at 5:18 PM | Permalink

    Ok, when can we get back to the simple cartoon diagrams where sunlight comes in,
    and watts stay trapped in the air forever and burn us up eventually? That story is much
    easier to understand than this.

    Just kidding. Thanks SteveM. My head is hurting so I think I’m having fun.

    The boundary layer/heating issue remined me of this odd experiment/demostration I had the
    pleasure to watch.

    Drag Reduction with Boundary Layer Heating

    The benefits of reducing the drag of either a new or existing aircraft configuration are obvious. An aircraft’s endurance is directly proportional to the lift to drag ratio. Decreased drag also translates into faster top speed, quicker acceleration, shorter take-off distances and lower direct operating costs in the form of fuel savings. In order to project military air power, or on the commercial side, receive better range and fuel economy, reducing drag during the cruise portion of a flight is the most critical. During cruise, the drag of the aircraft primarily comes from profile drag (skin friction), induced drag (drag due to lift), compressibility drag, separation drag and interference drag. Of these, skin friction (from the “wetted” elements of the aircraft) typically accounts for more than 50% of the total. By applying active surface heating in the turbulent regions of the aircraft’s boundary layer, the skin friction is reduced as a function of the ratio of the skin temperature to the ambient temperature. The result is an effective drag reduction method that can be retrofitted to existing aircraft.

    basically, We had a surface ( say wing leading edge) and the flow was turbulant. By HEATING THE SURFACE
    you could reduced drag. Kinda cool. Roger might like it.

    Also, note this neat application of surface roughness/groving to reduce drag


    Click to access 341.pdf

    I wonder if Roger P has looked at agricultural groving of the land surface and its effect on boundary

  6. gb
    Posted Jun 17, 2007 at 9:20 AM | Permalink

    ‘Maybe someone can comment on whether these nonurban inversions are characteristic in other environments (e.g. oceans, tropics)’

    I don’t think inversions will appear so often above the ocean. Land has a low heat capacity and can cool down relatively quickly after sunset. The sea surface doesn’t cool down so quickly.

    I haven’t read the article yet but I would expect that inversions mostly happen on calm days. If the wind is too strong the turbulence in the atmospheric boundary layer will be intense and this leads to strong turbulent mixing and a more uniform temperature. Also obstacles (buildings) can create perhaps more mixing and a more uniform temperature.

  7. Julian Williams
    Posted Jun 17, 2007 at 11:20 AM | Permalink

    There is a massive body of work on inversions and boundary layers, mainly concerned with their impact on the dispersion of atmospheric pollution.

    As a rule, true inversions occur on calm days (wind less than 3 m/s). During the day the boundary layer can be up to 1500 m high, and at night it can get as low as 50 m. At dusk and dawn, the temperature gradient starts to reverse, and there is no inversion for a while, although a mixing height of around 800 m is often employed. The ground roughness (buildings, trees, hills) will also tend to distort the inversion.

  8. Steve Sadlov
    Posted Jun 18, 2007 at 11:48 AM | Permalink

    In rural settings, radiative upward cooling is less inhibited by local issues such as short distance thermal gradients and the turbulent flows arising from man made structures. The above results vis a vis nocturnal inversion behavior do not surprise me.

  9. steven mosher
    Posted Jun 18, 2007 at 1:41 PM | Permalink


    I need to buy a vowel. Can I get the 5th grade version of your comment.

  10. Steve Sadlov
    Posted Jun 18, 2007 at 1:54 PM | Permalink

    RE: #8 – Out in the truly rural country the overall pattern of heat flux is more uniform as you move from one location to another and there are fewer rigid flat structures jutting up into the air flow. As a result, radiative flux dominates at night. In the city, heat flux varies widely – a hot spot here (e.g. a large building, mall, biz park, data center, etc) a heat sink there (a park, pond, lake, etc). As a result you end up with a mini “sea breeze” effect. Lots more chaotic flow of heat. Radiation, convection and conduction all happening at the same time based on solely the anthropgenic dissipation effects. Add a bit of wind, and the difference between city and country is even more pronounced.

  11. Steve Sadlov
    Posted Jun 18, 2007 at 1:56 PM | Permalink

    Another note, Hans Erren had photos of fog over the Benelux region of Europe. It was apparent in them that the urbanization patterns were affecting the radiation pattern (and thereby, the pattern of the radiation fog).

  12. steven mosher
    Posted Jun 18, 2007 at 2:31 PM | Permalink

    SteveS Thanks.

  13. Posted Jun 19, 2007 at 9:42 AM | Permalink

    Speaking of old ideas, has everyone already seen or have an explanation of the effect of global dimming, and subsequent reduction thereof, on global temperatures?


    John M Reynolds

  14. JP
    Posted Jun 22, 2007 at 8:52 AM | Permalink

    Operational weather forecaster have always been interested in the Boundary Layer. Temp inversons can play havoc when forecating cieling and visibilities. The art of forcasting hourly temperatures relies heavily on when the surface temp warms to an amount equal to that of the top of the boundary layer. Even being off by an hour can mean the difference between VFR conditions and IFR. This is particulary so in Central Europe. Forecaster who pay attention to the local geography and land use which surround the reporting site, realize that these micro scale changes have a large impact on fog and low level cloud dispersion; these changes in turn have a direct impact of surface temps.

    There were many occaisons when I forecasted in Germany that our airfiled remained in IFR conditions (0/0), while 2km down the road it was sunny and clear.

%d bloggers like this: