This web page describes an attempt to use nearby Vandenberg Air Force Base radiosondes, VBG, for establishing a calibration of the MTP radiometer.  It proved unfeasible, due to the strong horizontal gradients of the PBL's temperature field in going from the VBG site on the west coast, 50 miles away, to the ROOF site 2 miles inland from a south coast.  The inversion layers ofter are at the same altitude, but sometiems they acn be qwuite different, as when a cyclonic coastal eddy is just south of the South Coast, moving marine air inland to the ROOF site from the southeast but not affecting the VBG site 50 miles to the west.  This problem is illustrated on this web page, which will interest only those who are serious about not repeating the false starts I experienced in my flounderings.  This web page is reached from a web page that describes successful attempts to understand the MTP calibration.

Attempt to Calibrate MTP Using VBG Radiosondes

In theory VBG temperature profiles might be useable for inferring T(z) at the ROOF site.  This may require adjustments for the different site locations, and that is the goal of this web page.  First, however, it is necessary to describe supporting meteorological measurements and the state of their usefulness for use at the ROOF site.

Meteorological Support Measurements Used to Calibrate the MTP

The closest radiosonde site is Vandenberg Air Force Base, VBG, located 50 miles to the west north-west of the ROOF site.  VBG is near the coast, and is 121 meters above sea level.  On many occasions I have verified that the inversion base altitude in the vicinity of the ROOF site agrees with that derived from the VBG RAOBs.  This was done by driving up a mountain road close to the ROOF site, and noting the cloud tops, which were compared to the altitude corresponding to the top of the surface-based high relative humidity plot as well as the temperature's inversion layer base altitude, ILB.  Only on rare occasions, when temporal changes were occuring, was there disagreement between these various altitudes.  The next figure shows temperature profiles for the first 3 days fo teh ROOF observing period.

Figure 1.  VBG radiosonde temperature profiles for June 2, 0Z to June 4, 12Z.

Relative humidity profiles are shown in the next figure.

Figure 2.  VBG radiosone RH profiles for the same period.

Interpretation of these profiles is straightforward, except perhaps for Fig. 1's June 2, 12Z temperature profile and Fig. 2's June 4, 0Z RH profile.  BOth of these two exceptions show evidence of a weak horizontal intrusion of a layer of marine air at about 1 km altitude.  This will be dealt with below, in the discussion associated with Fig. 25.

These two figures provide a means of estimating the mixing layer thickness: 1) the inversion layer base altitude, ILB, and 2) the top of the ground-based high RH layer.  These are plotted in the next figure.

Figure 3.  Thickness of the mixed layer at VBG, as measured by the inversion layer base altitude, "ILB" (red), and the top altitude of the high humidity profile, "RH TOP" (green).

Usually, the the VBG RAOBs give the same valeus for inversion layer base altitude and RH top altitude.  The one occasion of a large difference had an ambiguous temperature structure that could have been interpreted in a way that agreed with the RHtop altitude, implying that there was some horizontal mixing of air that confused the interpretation.

I want to use the VBG temperature profiles to help calibrate the MTP.  However, this is risky due to the 50 mile distance between the VBG site and the ROOF site.  At ER-2 altitudes this separation would be considered "small," but for comparisons at low altitudes this separation could be "large."  One way to check the appropriateness of using VBG soundings to calibrate the MTP is to compare cloud altitude measurements made at the nearby Santa Barbara Airport, SBA, with the VBG altitudes for ILB and RHtop.  Both sites are close to the coast, so there might be reason to expect good agreement; especially since I have had good luck in verifying VBG-based cloud top predictions with actual measurements from driving up nearby Hwy 154.  If these two altitude types agree, then since SBA is only 4.3 miles from the ROOF site the agreement would provide confidence that VBG can be used at the ROOF site.

The next figure superimposes SBA cloud altitude measurements on the previous VBG graph.

Figure 4.  The triangle symbols are cloud base and ceiling measurements at SBA.  The thick blue dashed trace is my subjective estimate of marine layer top altitude at SBA, based on times when scattered (SCT) or few (FEW) clouds were present (which should be shallow, and therefore be close to the top altitude of the clouds).  The black pattern at the top is a graph of cloud cover percentage (plus 1100).  The x-axis is UT date.

The figure above contains some discrepant altitudes.  During June 3, 4, 5 and 6 the SBA cloud base altitudes are significantly higher than the VBG ILB and RHtop altitudes!  To illustrate this disagreement further, consider the next figure.

Figure 5.  VBG RH profile at 0Z, June 5 (blue), and SBA scattered cloud base altitutue at approximately 21Z, June 4 (green).

This figure shows that there are times when the temperature and RH profiles at VBG and SBA do not agree!  It cannot be argued that the 3 hour difference in times for these measurements can account for the discrepancy since the VBG RH profiles had the same shape for the preceding and following soundings (12 hours from the time in question).  This is a serious cautionary case that illustrates the potential dangers of relying upon radiosondes from 50 miles away when low altitude temperature fields are needed.  Mesoscale effects can be important at low altitudes, perhaps especially close to the coast.  On the occasion of June 3-6, there may have been a "coastal eddy" centered approximately over the "Channel Islands" that affected the Santa Barbara area but dd not extend westward enough to affect VBG.  The Coastal Eddy may force horizontal intrusions of marine air at altitudes that are otherwise dry.  The two "exceptions" to a straighforward interpretation mentioned in the descriptions of Fig.'s 21 and 22 could be cases of this occuring at VBG.

Proposed Method for Inferring True Temperature Profile Above the ROOF Site

It is important to validate the MTP's performance (i.e., calibration) by comparing MTP-derived T(z) with estimates of true T(z) profiles.  Software modifications are in progress for doing this.  However, due to the unreliable nature of the VBG temperature profiles for the ROOF site it remains to be determined how to best derive credible T(z) for the ROOF site.  One procedure that should be evaluated is described next.

Water vapor mxing ratio (and dew point temperature, DPT) should be conserved much better than air temperature when mesoscale dynamics distort the marine layer.  Thus, as the marine layer changes thickness, the DPT should remain constant.  Another property of DPT is that it should also have smaller horizontal gradients within the marine layer.  Therefore, I propose adopting the nearby SBA measurement of DPT as applying to the ROOF site.  This is Assumption #1.

I also propose to assume that the marine layer top altitude at the ROOF site is the same as at nearby SBA.  Any "coastal eddy" affects that change SBA's conditions with respect to VBG's conditions should be shared over the 4.3 mile distance between SBA and ROOF (or, cnservatvely, they should be 12 times smaller due only to the distance ratio).  This is Assumption #2.

Cloud cover is sometimes different at ROOF and SBA.  This, I shall assume, is due to small horizontal temperature gradients throughout the marine layer which are aligned perpendicular to the coast.  This is Assumption #3.

When stratus first forms, and when it last dissipates, the cloudy layer is very thin.  At this time the cloud base altitude is essentially the same as the cloud top altitude.  The shallow cloud layer at these special times is at the top of the marine layer, where the temperature is at a local minimum with respect to neighboring altitudes.  This is Assumption #4.

The SBA measurements of ceiling and cloud base are accurate for the SBA site, and when the cloud layers are shallow the ceilings and cloud bases are the same at the ROOF site.  This is Assumption #5.  (Note:  When there are two cloud layers and the lower has a less than 50% coverage and the higher cloud layer is BKN or OVC, the lower layer is used to define "ceiling" while the upper is used to define "cloud base.")

When the IR radiometer is providing quality temperatures for the ROOF altitude (in the direction that MTP views), one point in the T(z) plot will be established.  The next step is to determine (or infer) the temperature at the top of the marine layer (I'll also refer to this altitude as the inversion layer base, or "IL base").  This can be done at times when the stratus is just forming, or just dissipating (when the stratus layer is very shallow).  At these times it is possible to infer air temperature at the cloud layer using the previous assumptions:  Assumption #1 (DPT at the ROOF site is the same as DPT reported at SBA), Assumption #2 and #4 (the altitude of the shallow cloud layer over the ROOF site is the same as the SBA reported cloud base altitude), and Assumption #3 (when stratus is just appearing or disappearing, i.e., when the cloud cover is FEW, the temperature at the stratus layer equals the surface DPT).  There's a subtle effect that must be ealt with carefully:  sometimes SBA reports the FEW cloud cover condition at a quite different time than the ROOF site has a FEW cloud cover condition, so care must be taken to not use the SBA ceiling values under FEW conditions if their times differ greatly from the ROOF site's FEW condition.

The procedure just described allows for the determination of air temperature at a known altitude above the ROOF site, and this altitude can reliably be identified as the coldest altitude in the ROOF site's T(z) profile (below the capping inversion) - i.e, the IL base.  We now have surface temperature and the IL base temperature at a known IL base altitude.  Two more challenges remain:  1) how to fill in the profile between the surface and the inversion layer base, and 2) how to merge temperatures immediately above the IL base to the synoptic scale "free atmosphere" as reported by the VBG sounding.

The first challenge may need to take into account temporal context.  For example, during the formation phase the entire marine layer is cooling and air at the surface is not "connected" to air at the top of the marine layer by an adiabatic relationship since surface air parcels are not convectively circulating to the top of the marine layer.  Cooling should be occurring faster near the surface than at the top of the marine layer, due to the greater thermal IR emissivity of the ground compared to the air.  Inevitably, some subjectivity will enter the process of "drawing a trace" of T(z) between the surface and the IL base.  During stratus dissipation it may happen that a stratight line between the surface temperature and the IL base temperature is superadiabatic.  If this is the case, the it is easy to construct a trace in which the bulk of the superadiabaticness is confined to the lowest 100 meters (an empirical observation).  If the straight line is subadiabatic, then the bulk of the vertical gradient should be present at the lowest altitudes, due to the fact that most ground types have a greater thermal IR absorption coefficient than air.  Inevitably, some subjectivity will enter into drawing the trace for this condition.

The second challenge, merging the lower IL with the synoptic scale temperature field, can be done using the assumption that the percentage of air at any level within the transition region is specified by the difference between actual air temperature and a projection of the free atmosphere's vertical temperature gradient downward into the IL, which is to be compared with the temperature difference between this projection and the temperature at the IL base, and finally that the shape of this transition region has a characteristic "scale height" which should be preserved over distances as short between VBG and SBA (since these are set by a competition between the synoptic scale compression of downwelling air that is part of the global circulation pattern and the vigorousness of surface-based convection activity).

Example of Deriving T(z) Above ROOF Site

Consider the case of June 4.9 UT.  According to the SBA meteogram for June 4, 22Z, the cloud ccover condition was FEW at an altitude of 3300 feet, and the surface DPT was 12.8 C.  Thus, at 3300 feet above SBA the temperature was 12.8 C according to my assumptions.  My ROOF field log states that at 21Z cloud cover was SCT, while at 22.5Z it was CLR; thus, cloud conditions at the ROOF site changed in the same way as at SBA.  I conclude that at 22Z the temperature was 12.8 C 3100 feet above the ROOF site (recall that the ROOF site is at an altitude of 210 feet, while the SBA site is at 10 feet, which explains the need to subtract 200 feet from the cloud base altitude at SBA to get the expected cloud base altitude at ROOF).  Let's adopt a surface temperature of 18.0 C, which is compatible with both the SBA surface temperature record (after a 1.0 C downward adjustment that was typical for this date) and the TB's (assuming EH = 88 97 and 92%, which is a compromise solution for this date, giving closeness of TBs during the night and day hours and giving agreement with the night ROOF surface readings).  We now have two points on the T(z) trace that we're trying to create.  These are shown on the next figure, which also shows the VBG profile for June 5, 0Z.

Figure 6.  Illustration of first step in constructing a T(z) profile using the Assumption Set (described above) for June 4, 22Z.. Dashed green trace is the VBG profile which was demonstrated to NOT apply to the ROOF site at this time. Blue square is the ROOF surface temperature.  Red square is inferred from the SBA report of FEW clouds at 3300 feet when the surface DPT was 12.8 C.  The thin dashed line is an adiabat line pegged to the surface temperature.

VBG's T(z) is superadiabatic from the surface to 0.38 km, yet the two points so far created for ROOF are sub-adiabatic averaged over the lowest 1 km.  At 22Z, which is 2 PM local standard time, it is reasonable to expect a superadiabatic layer for the first 0.1 km when there are no clouds.  At the ROOF site the coud cover was BKN from 16Z to 20Z, then SCT at 21Z, and finally CLR at 22Z.  Therefore, it is not necessary to impose a ground-based superadiabatic layer at the ROOF site.  The surface temperature at the ROOF site had been stable (at 18 C) for the previous 3 or 4 hours, so there is no need to hypothesize a (night time) gound-based inversion.  A sub-adiabatic lapse rate above the surface is reasonable in this situation.  Simply joining the two temperature points is a conservative and acceptable hypothesis.  This profile would be consistent with convection being slightly more important than thermal equilibrium in establishing the lapse rate throughout the lowest 1 km (partially) mixed layer.  I admit to being uncomfortable with this hypothesis, but I cannot justify a more complicated model.

Figure 7.  Illustration of next step in constructing a T(z) profile using the Assumption Set (described above) for June 4, 22Z.  The thick blue line is a simple joining of the surface and FEW cloud level temperature estimates for the ROOF site.

This most crucial segment of the derived T(z) profile exhibits a constant lapse rate between the ROOF surface and the top of the quasi-mixed layer.  The next figure adds an inversion layer above the mixed layer.

Figure 8.  Illustration of next step in constructing a T(z) profile using the Assumption Set (described above) for June 4, 22Z.  The thick blue line segments include an inversion layer with the same thickness as the nearby VBG IL, and the profile above te IL is meant to match the VBG profile.

This completed profile is my best estimate for the ROOF site's T(z) for June 4, 22Z.  It is almost certainly defective, as there must be additonal structure within the lowest 1 km quasi-mixed layer.  There might actually be two ILs, one having a base altitdue of 370 meters (corresponding to the VBG IL base), and the second having its IL base at 1.0 km, similar to what is shown in the figure.  Such a profile would correspond to a coastal eddy intrusion at the upper part of the lowest 1 km air mass.  This is illustrated in the next figure.

Figure 9.  Illustration of an alternative way to join the surface and 1.0 km points.

The double IL profile in this figure cannot be ruled out.  It consists of two adiabatic layers within the lowest 1.0 km, and if the water vapor mixing ratio is approximately the same within this region, having DPT ~12.8 C, clouds would form only at the 1.0km level, as observed.  This profile is consistent with the limited observations and cannot be rejected except upon the weak argument that it's greater complication  is unjustified by the meager data available.

Can the MTP measurements be used to favor one model over the other?

Figure 10.  Observed and predicted TB versus applicable altitude for June 4, 22 Z. The predicted TBs are based on the single inversion layer model of Fig. 28.

This figure shows observed TBs versus applicable altitude, TB(Hb), that favor the double inversion layer model (of Fig 29).  Disregard the offsets for now, and notice the shapes of the TB(Hb), especially TB3(Hb).  The slope of "observed TB3(Hb)" is much steeper than the conservative model.  Keep in mind that in using TB(Hb) data one must give priority at the low altitudes to the channel with low high absorption coefficient (Channel #3), etc.  And it is also permissible to offset TBs belonging to a channel such that they match averaged over their overlap Hb region.  Taking these guidelines into account  yields the next figure.

Figure 11.  Offset adjusted observed TBs versus applicable altitude, with subjective weighted trace (thick black) which is implied by the observations (note the different scale from previous figures).

The subjective weighted average trace in this figure gives more weight to TB3 at low altitudes, etc, and reveals a super-adiabatic layer near the ground (in the first 100 meters, which is commonly observed).  There is a hint of an inversion at about 1000 meters, where the FEW cloud layer existed.  It is clear from this that both temperature structure models are "wrong" and that the mixed layer below cloud level was much colder than I had hypothesized.  Even the 12.8 C point at 1000 meters that was inferred to exist may be incompatible with the observed TB data.  However, such a critique of the procedure for deriving T(z) profiles from meteogram data is premature, since the MTP has not yet been fully calibrated.

This sub-section illustrates some of the power, but also some of the limitations, of invoking physics to derive a temperature profile with only meteogram type information.  Mesoscale effects are so important at low altitudes that if one's goal is to calibrate an MTP system using T(z) information it is probably necessary to locate it near a RAOB launch site.  In spite of these limitations, however, there are reasonable bounds to the T(z) profile that can be expected as output from a properly calibrated MTP.  The previous figures illustrate how the MTP might be subjected to a reality check using meteogram data. Although it may not be possible to calibrate an MTP this way, it should be possible to validate its performance to some extent.  I propose to use the T(z) generating procedure described in this section for that purpose.

At this point I conclude that another strategy must be developed for the calibration of MTP gains.

Return to the calling web page.

This site opened:  May 31, 2001.  Last Update:  August 14, 2001