Dr. John W. (Jack) Glendening,  Meteorologist
Naval Research Laboratory,  Monterey CA
 18 March 2002


     The automated Hollister Thermal Index Prediction (TIP) program estimates the current day's thermal soaring potential at Hollister based upon today's atmospheric conditions and forecast maximum temperature (Tmax).  It also forecasts the thermalling potential for up to two days in advance.  These analyses are sent in the morning via email to those who have subscribed to the Hollister Gliding Club's HTIP email list and are also posted on the WorldWideWeb.  Specifically, estimates of today's Thermal Index (TI) and thermal strength are computed based upon the observed 12Z atmospheric sounding at Oakland together with the Tmax forecast for Hollister by the National Weather Service (NWS) and by the Weather Channel (WxC).  The use of two Tmax predictors provides guidance on the degree of uncertainty associated with the TI estimate for that day.  In addition, forecast soundings from two different models and NWS and WxC Tmax forecasts are used to predict the TI for that afternoon and for the next 24, 36, and 48 hours.  At present, my computer runs the program automatically around 8:30 AM and if all data is available the results are emailed within 10 minutes.  If the sounding data is late in arriving, however, the email will not be sent until that data is available, to a maximum wait time of around 30 minutes.  Since the degree of uncertainty of any specific TI calculation is of importance, the email includes several outputs to aid assessment of that uncertainty.  Reading the summary numbers in the email's Subject line - or, particularly for website users, a corresponding list of summary numbers with textual descriptions given at the top of the TIP - is often sufficient for the user since they provide the principal program results.  However, more detailed information is provided in the email body if needed or desired.  A listing of all TIP forecast sites and information on subscribing to other TIP email lists can be found at the Latest TIP Forecasts webpage.  The following description also applies to TIPs used at other sites, except that at many sites no appropriate sounding observation is available so a model-predicted sounding is instead used to compute TI values for the current morning.

      The Thermal Index (TI) at a given altitude is the difference between the observed air temperature and the temperature that a parcel of air would have if it started at the surface, as occurs in a thermal, and rose adiabatically (that is, without losing heat) to that altitude.  The TI=0 height approximates the height of the tops of the thermals for the day, but the maximum thermalling height is expected to be somewhat lower due to the sailplane's descent rate and because the thermally mixed layer, or Boundary Layer (BL), is not truly adiabatic.  At present the program computes heights associated with two TI criteria, -4°F and 0°F, to judge the thermal conditions.  Ideally the maximum thermalling height could be estimated from a single TI criterion, but such a criterion is not well established.  For example, traditionally reported TI criteria for the maximum thermalling height include the TI=-5.4°F(-3°C) criterion mentioned on Kevin Ford's TI calculation website and the TI=0°F(0°C) criterion used for the Reno NWS Soaring Forecast.  Also, the maximum thermalling height will depend upon glider performance and pilot proficiency.  Actual soaring experience at Hollister may help establish a reasonable TI criterion for that site.  The uncertainty in the TI criterion is likely smaller than the uncertainly due to the average error in the forecast surface temperature maximum, which is about 4°F for Hollister as discussed below.  A more detailed explanation of soundings and the theory behind the Thermal Index is available at the TIP Sounding Analysis webpage.  [Note: The TI has units of temperature and when used on other websites is generally given in degrees Celsius, °C, not Fahrenheit, although that fact may be obscured when a website neglects to provide units when referring to the TI.  For this program, however, all TIs will be given in Fahrenheit,°F, and this is implicit when the TI units are omitted for convenience.  This choice more easily allows a change in the surface temperature maximum Tmax, which is reported in °F, to be translated into a change in the predicted TI aloft, since there is then an inverse 1:1 correspondence between the two; that is, a 1°F increase/decrease in Tmax will decrease/increase the predicted TI at a given level by 1°F.]

    The following sections describe the Hollister TIP email and the methods used to obtain its predictions - more information than you likely want to know!  The only section I consider mandatory reading is the following SUMMARY DESCRIPTION, which describes information provided in the email's Subject line and in the SUMMARY NUMBERS section at the top of the TIP.  The other sections can be read when the user desires further information.


     The TIP's purpose is to predict thermal soaring conditions for today and the following two days.  To judge those conditions, it computes two estimates of the "maximum thermalling height" from the morning (AM) sounding by using an observed or model forecast atmospheric temperature sounding together with a predicted surface temperature maximum (Tmax).  Two height estimates are provided because (1) the choice of a single criterion for the "maximum thermalling height" is uncertain and (2) providing two values gives some indication of the the sensitivity (uncertainty) of the forecast to errors in the forecast Tmax.  In program jargon, these two numbers are called the "TI=-4 height" and the "TI=0 height" and they are separated by a comma in the output.  My present guess is that the "TI=-4 height" is a better indicator of the maximum thermalling height over flat terrain.  The TI=0 is a general indicator of thermalling conditions and is more likely to represent conditions over elevated terrain.  It is important to recognize that thermal updraft strength depends strongly upon thermal depth (see below), so an increase in the TI=0 height is important not only because thermalling heights are then expected to be higher but also because thermal updrafts are then expected to strengthen.  (Cloud formation will also increase updraft strength, through the release of latent heat, but that effect is not included in the TIP forecast).  If for simplicity you only want to deal with a single number, for soaring over flat (elevated) terrain I would suggest using the number before (after) the comma [the TI=-4 (0) height] as the indicator of thermalling conditions: the higher/lower that number, the stronger/weaker the expected thermal conditions.

     Predictions based upon th afternoon (PM) sounding are also provided because most flights occur in the afternoon and conditions then may have altered from those of the morning sounding, particularly near times of frontal passage.  However, calculation of PM TI=-4 and TI=0 heights is subject to mismatches between the TI-method assumptions and the model-predicted soundings, as described below, so misleading heights can result.  Therefore, PM conditions are reported as a change in the thermalling conditions, specifically as an increase or decrease of the TI=+4 height since it is less subject to such mismatches.  PM predictions are thus made in a relative fashion and other thermalling parameters are expected to similarly change for the better/worse.  This number provides a ballpark estimate of the changes in the maximum thermalling height, but because it must be computed at the TI=+4 level instead of the TI=-4 or TI=0 level this change is only an approximation to the change at those levels.  [A subtle and perhaps confusing point is that both AM and PM soundings are used to predict predict soaring conditions at the time of Tmax (which is why the same Tmax is used for both predictions) - the AM/PM designators refer to the sounding used for the prediction and do not indicate that the predictions are intended to be for the morning and afternoon hours respectively.]

     The summary numbers given in the email's Subject line are the TIP results of primary interest, typically giving enough information that the email can be deleted without needing to open it.  Alternatively, these numbers are also given in the SUMMARY NUMBERS section at the top of the TIP,  together with a textual description for each.  The Subject line will look something like:

HolTIP: am>6178,8058|6|610 (-2) pm>-1373 MONam>5000,7000 MONpm>-1218 TUEam>4833,6636

and the corresponding SUMMARY NUMBERS section, which includes additional information, would be

AM Avg. TI HEIGHTS:  6178,8058 ftMSL @TI=-4,0degF
AM Avg. Hcrit  HGT:   6069 ftMSL    (Max. _flat_terrain_ thermalling height)
AM HGT VARIABILITY:  1870 ft     (from TI=0-+4degF)
AM Avg. Buoy/Shear:   6          (thermals may be unworkable if 5 or less)
AM Avg UPDRAFT W* :   610 fpm    (subtract glider sinkrate to get vario)
PM Avg HGT CHANGE :  -1373 ft @TI=+4degF
Tmax UNCERTAINTY  :    -2 degF   (deviation between NWS & WxC Tmax forecasts)
BL Max. Rel.Humid.:  AM= 87%   PM= 95%
AM Extensive CLOUD:  Dif= 3733 ft  LCL= 4318 ftMSL  (if Dif>0, expect OD at LCL)
PM Extensive CLOUD:  Dif= 915 ft  LCL= 4653 ftMSL  (if Dif>0, expect OD at LCL)
THUNDERSTORM CAPE :  AM= 0  PM= 0  (if CAPE>0, thunderstorms possible)
MON AM HEIGHTS:  5000,7000 ftMSL @TI=-4,0degF  &  Hcrit=  5144 ftMSL
MON PM  CHANGE:  -1218 ft @TI=+4
TUE AM HEIGHTS:  4833,6636 ftMSL @TI=-4,0degF  &  Hcrit=  4955 ftMSL

Both indicate that today's estimated thermalling height over flat terrain is 6178  ftMSL (the TI=-4 height) based upon the AM sounding (a large height for Hollister due to passage of a front just a few hours before).  Conditions are expected to weaken in the afternoon since the TI=+4 height decreases by -1373 ft.  For tomorrow (MON), condiitions are expected to be weaker, the predicted TI=-4 thermalling height being 5000 ftMSL for the AM and again lowering in the PM (-1218 ft).  For the following day (TUE),  the AM estimated thermalling height (TI=-4) is 4833 ftMSL, slightly lower but comparable to conditions on MON.

     The number 6 following today's AM predicted TIs is the estimated Buoyancy/Shear "B/S" ratio: if this is 5 or less the thermals may be broken up by wind shear.  The following 610 is the estimated average upward velocity of thermals "W*" in fpm (subtract the sailplane's descent rate to get the actual average rate of climb).  Following that, in parenthesis, is the difference between the NWS (National Weather Service) and WxC (WeatherChannel) Tmax predictions as an indication of the Tmax forecast uncertainly for that day: today the predicted WxC Tmax is 2°F cooler than the predicted NWS Tmax.

     The SUMMARY NUMBERS section also provides additional information not given in the Subject line.;  Hcrit provides an alternative (and more theoretically defensible) estimate of the maximum thermalling height over flat terrain; here Hcrit is close to the TI=-4 height, as is usually the case, but under some circumstances the two can differ significantly - which is the more accurate number in such cases is presently undetermined (and I would like to hear of any flights which would help such a determination).  Today's thermalling height variability, which is also the sensitivity of today's predicted TI heights to error in the forecast maximum temperature, Tmax, is rather large (1870 ft) so significant variability of the soaring conditions is expected (or the predicted soaring conditions will be greatly in error if the actual Tmax differs from the predicted Tmax).  Several parameters intended to aid in cloud prediction are included.   The greater the maximum Relative Humidity in the Boundary Layer (BL) the greater the expected cloud cover at the top of the BL, but the relationship between this number and partial cloudiness must rely upon empirical correlations based upon experience.  Today extensive cloudiness created by thermal mixing, also known as OverDevelopment, is expected to occur with a cloud base around 4318 to 4653 ftMSL since the AM and PM "Extensive Cloud Dif" values are greater than zero;  if they were negative, then no OD is expected over flat terrain.  The CAPE value would be positive if thunderstorm development were predicted,  but today the AM and PM CAPE values are zero so thunderstorms are not expected.

     The AM and PM prediction summary numbers each have strengths and weaknesses.  Both should be considered when forecasting expected thermalling conditions for a given day, but the PM forecast is particularly important due to its proximity to the flight period. AM and PM predictions in the Subject line are based upon the NWS Tmax forecast.  Today's AM values are based upon the morning observed Oakland sounding while all other values represent the average of the MAPS and ETA model TI predictions (when both model forecasts are available).

     Note that a large (greater than 2000 ft, say) difference between the TI=-4 and TI=0 heights indicates that the "maximum thermalling height" for that day is especially sensitive to changes (or errors) in the maximum surface temperature.  Also, in such cases it is more likely that better thermalling conditions will occur over elevated terrain.

     When utilizing and evaluating the TI predictions, it is important to recognize that these predictions assume relatively simple thermalling conditions, i.e. cloudless thermals developing over flat terrain, and thus the TIP cannot numerically forecast all thermalling conditions.  But relative predictions of "better" or "worse" conditions should still be useful even when complicating factors occur.

      One complicating factor is terrain.  Thermals over elevated terrain will have higher tops than those over the nearby valley.  Often this difference can be allowed for by simply using a larger TI criterion to estimate the thermal top over elevated terrain.  Such adjustments are necessarily empirical, depending for example on the height of the terrain.  So the criterion value to be associated with a specific terrain feature must be established through previous experience at that feature.

     A complex complicating factor is cloud formation.  As discussed below, cloud formation produces stronger updrafts than predicted by the TIP due to additional buoyancy produced aloft, particularly for deeper convective clouds.  The thermal strength W* and the Buoyancy/Shear ratio B/S will then be larger than the TIP predicted values.  The height of the thermal also increases. becoming the cloud top height, but the maximum soaring height is now limited by the cloud base instead of the thermal top.  In such cases the maximum soaring height is often below the maximum thermalling height predicted by the TIP, since the condensation occurs below the height reached by a dry thermal.  In the presence of convective clouds the TIP is probably best thought of as providing a minimum estimate of thermalling conditions, with thermalling conditions expected be better than predicted - here "thermalling conditions" is to be interpreted as meaning updraft strength, since the maximum soaring height is now decoupled from the maximum thermal height, which is what the TIP TI heights predict.

     Please note that the TIP is not a complete "Soaring Forecast". The TIP is a tool which provides one important piece of the soaring forecast puzzle, that of forecasting thermalling conditions, in an easily digested package.  It is a tool which uses computer automation to take some of the drudgery out of soaring predictions and provide information which pilots generally do not have the time or knowledge to obtain individually.  But the TIP only considers one aspect of soaring weather - that of thermal lift.  The TIP says nothing about lift from another sources, such as wave or shear, so pilots cannot just look at the TIP forecast and expect to know whether today is a "good" soaring day in the general sense of having some kind of strong lift.  In addition, other weather factors affecting soaring, such as ceiling height and precipitation, are not considered by the TIP's summary numbers.

     Forecasting the weather will never be like reading an airspeed indicator.  Intelligent interpretation of the TIP, based upon an understanding of its methodology and limitations, will give a more realistic expectation of expected conditions than is obtained by simply accepting its numbers without evaluation.  And of course the TIP forecasts are no better than the Tmax and sounding information from which they are calculated.  If, for example, the pilot judges that the actual Tmax will be warmer/cooler than predicted, he should then expect thermalling conditions to be better/worse than predicted by the TIP.

·   The email is intended to be read with a fixed font so that similar columns align vertically but can be read with a proportional font (except for the temperature profile plot, which requires a fixed font for proper presentation).
·   Since email recipients will often not be reading the body of the email, if a new note is added the email Subject line will be begin with the message "NEW_NOTE" for that day only.


     This section briefly describes the body of the email.  The email itself need not be read if the user is satisfied by the summary information given in the Subject line.  The information within the email allows a more detailed evaluation of the TI predictions and, in particular, allows better evaluation of the uncertainly in the predictions.  For the casual user the number of numbers there can be intimidating, so remember that the most important numbers are those given in the Subject line (and in the SUMMARY NUMBERS section at the top of the email body).  The other numbers within the email can be used when further information is desired and when one becomes more familiar with the program results.  [The most interesting part of the email body for the casual user may be the funky plot of temperature vs. height, since it gives a graphic representation of what the TIP does - showing the atmospheric structure in an "unskewed-T" format and also plotting the adiabat giving the afternoon heating due to Tmax, with the point at which the two curves cross being the TI=0 height] 

     The email body consists of the following main sections:

     The "SUMMARY NUMBERS" email section provides a textual description of the summary numbers given in the Subject line along with additional summary numbers, as described in the SUMMARY DESCRIPTION above. 

    The "NEAREST NWS TAF FORECAST" email section [not included in the sample email provided below] gives the NWS Terminal Aerodrome Forecast (TAF) nearest to the TIP location, which hopefully will be relevant to the TIP location.  This section provides information on cloud bases (notably for clouds above the BL, which the TIP does not predict), surface winds, and precipitation. 

    The "WxC HOURLY FORECASTS" email section gives hourly weather conditions predicted by the Weather Channel.  It is particularly useful for judging the time of maximum heating, which generally occurs when Tmax is largest, and whether sky conditions or wind speeds are expected to change significantly.

    The "THERMAL INDEX ANALYSES" email section consists of several subsections.  The "TODAY's ANALYSIS OF MORNING SOUNDING" subsection gives an extensive analysis of the morning atmospheric sounding observed at Oakland, providing information for those who want to gauge the uncertainty in today's TI predictions.  It gives the TI heights forecast by several possible Tmaxs: the NWS prediction, the WxC prediction, and their AVeraGe.  The resulting TI variation is an indication of TI uncertainty due to uncertainty in Tmax prediction.  The second and third subsections, titled "MAPS MODEL FORECAST TIs" and "ETA MODEL FORECAST TIs" give individual TI predictions for this afternoon and the following two days from the two different models.  For the AM soundings the TI=-4 and TI=0 heights are given, while for the PM soundings the change in the TI=+4 height is given.our summary sections are ordered by time.  If enabled, there will be an additional subsection titled "POST-ANALYSIS OF YESTERDAY's ACTUAL CONDITIONS" [not included in the sample email provided below] giving yesterday's observed Tmax at Hollister and yesterday's NWS and WxC forecast Tmax as an indication of the degree of Tmax error in yesterday's TI forecast.

      The next two email sections are titled "TODAY's *AM* TI LISTING..." and "TODAY's *PM* TI LISTING...".  They provide detailed listings of the "best guess" AM and PM model soundings, normally MAPS model soundings using the NWS Tmax (when available), and winds are given for different heights.  The feature of interest is often the plot depicting the atmospheric temperature structure in an "unskewed-T" form, which is more easily readable than a skew-T plot, and also plots the adiabat corresponding to Tmax.  For the AM sounding, the point at which the two curves cross is the predicted TI=0 height.   The best estimate of the PM thermalling height is obtained through analysis of the PM sounding - however, to be effective such interpretation requires knowledge of the PM mixing layer structure, as discussed briefly in the PM analysis section below  At the end of each of these listing sections is "CLOUD FORMATION PARAMETERS", which gives cloudiness estimates.

     If enabled, an email section titled "POST-ANALYSIS TI LISTING"  [not included in the sample email provided below] gives yesterday's predicted PM sounding with yesterday's observed Tmax at Hollister (when available).

    The final email section "MISCELLANEOUS NOTES" can include updated information which supersedes the descriptions given on this web page.


     The results produced by the different sections will be described in detail using the sample email below.  It is hoped that much of the email's contents is self-explanatory.  Only a few of the numbers are of general interest and those are provided in the email's Subject line and in the SUMMARY NUMBERS section at the top of the TIP.   The numerous numbers provided in the analyses are mainly intended for those aficionados, such as cross-country pilots, who are particularly interested in the influence of atmospheric conditions on thermalling.   The following description is intended for those users.

     The sample email is the same as used for the Subject line and SUMMARY NUMBERS examples used previously.  To recapitulate, the Subject line for this example is:
 HolTIP: am>6178,8058|6|610 (-2) pm>-1373 MONam>5000,7000 MONpm>-1218 TUEam>4833,6636
The first pair of numbers gives the prediction for today's heights at which TI=-4°F and TI=0°F,  the buoyancy/windshear ratio B/S follows the first vertical bar, and the predicted thermal strength W* follows the second vertical bar.  These predictions are based upon the NWS temperature prediction, and the temperature difference between the NWS and WxC predictions is given in parentheses.  These predictions for the current day are followed by TI height forecasts for this afternoon, for tomorrow, and for the following day.

 Sample Email
     A sample email example is (you can open a new browser window to keep this example in view when reading the descriptions given below - you may need to widen the window to see the entire line. A fixed font is used to display the email so that plotted values will be correctly aligned):


DrJack's TIP (Thermal Index Prediction) for HOLLISTER on SUN Nov 25

****************** SUMMARY NUMBERS - based on NWS Tmax only *****************

AM Avg. TI HEIGHTS: 6178,8058 ftMSL @TI=-4,0degF
AM Avg. Hcrit HGT: 6069 ftMSL (Max. _flat_terrain_ thermalling height)
AM HGT VARIABILITY: 1870 ft (from TI=0-+4degF)
AM Avg. Buoy/Shear: 6 (thermals may be unworkable if 5 or less)
AM Avg UPDRAFT W* : 610 fpm (subtract glider sinkrate to get vario)
PM Avg HGT CHANGE : -1373 ft @TI=+4degF
Tmax UNCERTAINTY : -2 degF (deviation between NWS & WxC Tmax forecasts)
BL Max. Rel.Humid.: AM= 87% PM= 95%
AM Extensive CLOUD: Dif= 3733 ft LCL= 4318 ftMSL (if Dif>0, expect OD at LCL)
PM Extensive CLOUD: Dif= 915 ft LCL= 4653 ftMSL (if Dif>0, expect OD at LCL)
THUNDERSTORM CAPE : AM= 0 PM= 0 (if CAPE>0, thunderstorms possible)
MON AM HEIGHTS: 5000,7000 ftMSL @TI=-4,0degF & Hcrit= 5144 ftMSL
MON PM CHANGE: -1218 ft @TI=+4
TUE AM HEIGHTS: 4833,6636 ftMSL @TI=-4,0degF & Hcrit= 4955 ftMSL

******************** WxC HOURLY FORECASTS *************************************

95023 WxC hourly for 11/25 @ 9AM 10AM 11AM 12PM 1PM 2PM 3PM 4PM
Sfc. Temperature (F) = 46 49 53 56 58 58 57 55
Sfc. Wind Speed (kt) = 6 6 6 6 6 6 6 6
Sfc. Wind Direction = S S S SW W SW W W
Precip.Probabil. (%) = 20 20 20 20 20 20 30 40
Weather Type = Prtly Prtly Prtly Prtly Prtly Prtly Prtly Prtly
Cldy Cldy Cldy Cldy Cldy Cldy Cldy Cldy

******************** THERMAL INDEX ANALYSES ***********************************

TODAY's ANALYSIS of MORNING SOUNDING at OAK @ 11/25:12Z for Sfc= 230 ftMSL
NWS Tmax= 59 > TI=-4,0@ 6178, 8058 Hcrit= 6069 HgtVar=1870 B/S= 6 W*= 610
WxC Tmax= 57 > TI=-4,0@ 4833, 6843 Hcrit= 4913 HgtVar=1907 B/S= 6 W*= 509
AVG Tmax= 58 > TI=-4,0@ 5642, 7357 Hcrit= 5422 HgtVar=1920 B/S= 6 W*= 558

am NWS Tmax= 59 > TI=-4,0@ 5000, 7450 Hcrit= 5511 HgtVar=2216 B/S= 5 W*= 566 MAPSanl:11/25:12Z
pm NWS Tmax= 59 > Change@TI=+4: -1726 -4.5F MAPS12h:11/26:00Z

am NWS Tmax= 59 > TI=-4,0@ 5142, 7875 Hcrit= 5889 HgtVar=2513 B/S= 8 W*= 592 ETAanal:11/25:12Z
pm NWS Tmax= 59 > Change@TI=+4: -1020 -2.4F ETA12hr:11/26:00Z
MONam NWS Tmax= 60 > TI=-4,0@ 5000, 7000 Hcrit= 5144 HgtVar=1281 B/S=33 W*= 551 ETA24hr:11/26:12Z
MONpm NWS Tmax= 60 > Change@TI=+4: -1218 -4.2F ETA36hr:11/27:00Z
TUEam wxc Tmax= 64 > TI=-4,0@ 4833, 6636 Hcrit= 4955 HgtVar=1721 B/S=13 W*= 584 ETA48hr:11/27:12Z

**************** TODAY's *AM* TI LISTING using RAOB+NWStmax ***************

25-NOV-2001 12 UTC TI report from OAK OBSraob upper air data.
Forecast max Temp.: 59.0 F
Forecast max VirtT: 60.6 F
Forecast sfc VirtT: 60.8 F
Raob est. max temp: 54.1 F
Surface elev Temp: 48.8 F
Surface elev VirtT: 50.6 F
Surface elevation: 230 ftMSL SfcAdjust:1
Station elevation: 10 ftMSL
Lowest elevation: 404 ftMSL
Lowest elev Temp: 48.6 F
Lowest elev VirtT: 50.3 F
Lifted Index @700mb: 0.9 C
Sfc-Lift Cond Lev: 1342 ftMSL
Lift Cond Lev (LCL): 2657 ftMSL
Conv Cond Lev (CCL): 2967 ftMSL
Ford est. base of any clouds: 4369 ftMSL
Lowest elev windspeed: 5 kt
Mixing Layer windspeed: 18 kt
Convection overcast height: 6383 ftMSL
Convection TI=0 height: 8051 ftMSL

Height *TI* Wind TI=0 Tv ------ Temperature Profile Plots ------
ftMSL degF deg kts trig degF o=Tv .=DewPt *=Adiabat 1.2 degF/division
----- ---- ------- ---- . ---- ---------------------------------------------
12000 5.6 65 | 3.3
11500 5.3 300 28 64 | 5.7
11000 4.9 64 | 8.0
10500 4.5 63 | 10.2 o
10000 4.1 63 | 12.5 * o
9500 3.4 300 26 62 | 14.5 * o
9000 2.5 61 | 16.2 * o
8500 1.5 289 27 61 | 18.0 * o
8000 -0.2 286 27 59 | 18.9 . o
7500 -0.8 58 | 21.0 . o*
7000 -1.5 290 26 58 | 23.0 . o*
6500 -3.1 56 | 24.0 . o *
6000 -4.5 299 23 54 | 25.3 . o *
5500 -5.2 54 | 27.3 . o *
5000 -5.8 53 | 29.3 . o *
4500 -6.4 305 21 53 | 31.4 . o *
4000 -6.9 52 | 33.6 . o *
3500 -7.3 52 | 35.8 . o *
3000 -7.8 51 | 38.0 . o *
2500 -8.3 295 15 51 | 40.2 . o *
2000 -8.5 50 | 42.6 . o *
1500 -8.8 50 | 45.0 . o *
1000 -9.1 50 | 47.4 . o *
500 -9.3 235 5 50 ! 49.9 . o *

Expect extensive cloud formation at LCL when LCLdiff positive
or over mountains when LCL is (approx) below mt top
(1stLine=_no_warming, 2ndLine=_with_warming)
Thunderstorms possible when CAPE is positive
BLmaxRH LCLdiff LCL CCL CAPE Zsfc= 230 TI=0@ 8051
87 3733 4318 4574 0
83 2285 5766 5880 0

**************** TODAY's *PM* TI LISTING using MAPSpm+NWStmax *****************
*NB* TI values below pm model-predicted mixing height are meaningless !

26-Nov-2001 00 UTC TI report from OAK MAPS12h upper air data.
Forecast max Temp.: 59.0 F
Forecast max VirtT: 60.9 F
Forecast sfc VirtT: 61.0 F
Surface elev Temp: 52.9 F
Surface elev VirtT: 54.9 F
Surface elevation: 230 ftMSL SfcAdjust:1
Station elevation: 518 ftMSL
Lowest elevation: 230 ftMSL
Lowest elev Temp: 52.9 F
Lowest elev VirtT: 54.9 F
Ford est. base of any clouds: 4033 ftMSL
Lowest elev windspeed: 5 kt
Mixing Layer windspeed: 7 kt
Convection overcast height: 4830 ftMSL
Convection TI=0 height: 5568 ftMSL

Height *TI* Wind TI=0 Tv ------ Temperature Profile Plots ------
ftMSL degF deg kts trig degF o=Tv .=DewPt *=Adiabat 1.2 degF/division
----- ---- ------- ---- . ---- ---------------------------------------------
12000 14.3 73 | 12.3 o
11500 12.9 312 26 72 | 13.5 o
11000 11.9 71 | 15.3 o
10500 11.0 70 | 17.0 o
10000 10.1 319 22 69 | 18.8 * o
9500 8.8 68 | 20.2 * o
9000 7.6 67 | 21.6 * o
8500 6.4 65 | 23.1 * o
8000 5.2 319 13 64 | 24.6 * o
7500 4.1 63 | 26.2 * o
7000 3.2 62 | 27.9 . * o
6500 2.2 61 | 29.6 . * o
6000 1.2 60 | 31.2 . *o
5500 -0.2 264 8 59 | 32.5 . o
5000 -1.7 240 8 57 | 33.7 . o *
4500 -2.7 229 8 56 | 35.4 . o *
4000 -3.3 210 9 56 | 37.4 . o *
3500 -4.1 207 8 55 | 39.3 . o *
3000 -4.5 210 7 54 | 41.6 . o *
2500 -4.6 211 7 54 | 44.2 .o *
2000 -4.7 215 7 54 | 46.7 . o *
1500 -4.6 217 7 54 | 49.5 . o *
1000 -4.7 221 7 54 | 52.1 . o *
500 -5.0 224 6 54 ! 54.4 . o *

Expect extensive cloud formation at LCL when LCLdiff positive
or over mountains when LCL is (approx) below mt top
(1stLine=_no_warming, 2ndLine=_with_warming)
Thunderstorms possible when CAPE is positive
BLmaxRH LCLdiff LCL CCL CAPE Zsfc= 230 TI=0@ 5568
95 1726 3842 4171 0
95 915 4653 4693 0

******************** MISCELLANEOUS NOTES *************************************

Addtional local weather information is available at the Tmax data sites:
and at the hourly data site:

A description of the Hollister Thermal Index Prediction (TIP) program
is given at:
First-time TIP users are strongly encouraged to read the second section of
that description, titled "SUMMARY DESCRIPTON", for information on how to
interpret the email's Subject line, which gives the basic TIP forecasts.
Descriptions of how some pilots use the TIP for their own soaring flights
are given at:
The latest TIP forecasts are also available on the web at:
The hourly-updated BLIPSPOT forecasts of thermal soaring parameters at
specific locations for various times throughout the day are
available on the web at:
The latest BLIPMAP area forecasts of thermal soaring parameters for
the current day over CA/NV are available on the web at:

Currently the TIP seems stable and I am only looking at it for
my own soaring needs. If you do notice a continuing problem,
please email me.

UPDATES to the description at



This email section is described in the Summary Description above.


This section is not included in the sample email presented above.
 The Terminal Aerodrome Forecast provides information on cloud bases (especially for clouds above the BL, which the TIP cannot predict), surface winds, precipitation, and other conditions that the FAA deems significant to aviation operations.  The provided location can differ from that of the TIP, so the TAF's relevance must be evaluated.  The TAF is in coded format: a short description is at while the full description is given at


 The hour-by-hour WxC forecast is primarily intended to provide the time of maximum surface temperature, but it also includes wind speed, wind direction, and weather type. The weather type can include, in increasing order of cloudiness (some vowels being omitted to save space): "Sunny", "Prtly Cldy", "Mstly Cldy", and "Cldy"(overcast). Whether the clouds are high, medium, or low altitude clouds is not specified. "Fair" sometimes means a thin high overcast.  These descriptions indicate large-scale cloudiness and not, for example, cloud formation over mountains.  The maximum surface temperature may not agree with the given WxC Tmax, but the reason for this is unclear since WxC does not reveal their prediction methods


This email section consists of subsections giving TI analysis of observed or model-predicted soundings.

     This subsection predicts the current day's thermalling conditions based upon the observed morning sounding released from Oakland for predicted temperature maxima obtained from NWS and WxC.  [For sites with no local observed sounding, or for Hollister if the Oakland observed sounding is missing, the predictions here will employ model-predicted soundings rather than an observed sounding.]  The subsection title is followed by the date and time (in UCT) of the sounding analyzed, as a check that that morning's observation has actually been used.  For the sample case, the Oakland sounding is for November 25 at 12Z=4AMpst.
     The next lines give the TI heights calculated for each of three different surface temperature predictions.  The identifier for the surface prediction source, either "NWS" or "WxC" or their average "AVG", begins each TI estimation line, followed by the predicted surface temperature maximum Tmax in degrees Fahrenheit ("Tmax=").  In the example, the predicted NWS temperature is 59°F.
     The number pair on each line, following "TI=-4,0@", give the primary program result, the calculated heights in feet MSL for TI values of -4°F and 0°F respectively.  These heights are 6178 and 8058 ftMSL for the NWS case.  Heights for two criterion values are provided to help each pilot make his own prediction of the maximum thermalling height, based upon his own criterion - however, I would suggest use of the TI=-4 height as a starting point for thermalling over flat terrain.  Comparison of the two different TI heights, computed for a for a single Tmax, will help the user evaluate the sensitivity of the maximum thermalling height prediction to the assumed TI value and also its sensitivity to the predicted Tmax.  If the TI=0 height is much larger than the TI=-4 height, such as over 2000 ft larger, then conditions in different locations, such as over the hills, have a good chance of being substantially better than those over the valley.  Further, the TI forecast will then be more sensitive to any error in the surface temperature prediction.  On days having a strong temperature inversion aloft the two numbers will both be low and relatively close together (within 500 feet for example), indicating that there is not much hope of good thermalling conditions.
    The next number on the same line ("Hcrit=") gives an alternative prediction of the maximum thermalling height expected over flat terrain.  often Hcrit is roughly comparable to the TI=-4 height, as is true for its present value of 6069 ftMSL.  This parameter is described in further detail below.
     The next number indicates the expected variability of the thermalling height, which is assumed to depend upon the atmospheric stability above TI=0.  This number is the difference between the TI=0 and TI=+4 heights in feet, here 1870 ft.   A larger number indicates that the maximum thermalling height will have greater variation from one thermal to another.
     The next number ("B/S=") is the Buoyancy/Windshear ("hot air") ratio, here being 6.  A small B/S ratio means that windshear is more likely to be significant.  At present, the best estimate, based on limited pilot reports, is that for a B/S of 5 or less the thermals are likely to be torn apart by windshear and broken.  Further discussion is given below.
     The next number ("W*=") gives the predicted thermal strength, the upward velocity of air within the thermal in feet per minute, 610 fpm for the NWS case.  Note that a sailplane's thermalling descent rate must be subtracted to give its vertical velocity while thermalling.  W* is calculated from convective theory, as described further below.
     Providing TI calculations for two different Tmax predictions, NWS and WxC, aids in evaluating the uncertainty of the maximum thermalling height resulting from uncertainty in the surface temperature prediction.  This uncertainty increases if the heights for TI=-4 and TI=0 differ greatly, since this indicates large sensitivity to a 4 °F error in the surface temperature prediction.  Note that there is an inverse 1:1 relationship between the TI index at a given height and the predicted Tmax.

     This subsection gives TI estimates based upon soundings from NOAA FSL's MAPS model, providing forecasts for the morning and afternoon of the current day.  The format of the morning prediction is similar to that of the previous subsection.  For Hollister, comparison between the AM model prediction and the prediction obtained from the observed AM Oakland sounding in the previous section provide an estimate of the model's initial state. 
     The afternoon prediction gives the relative change expected, based upon the change in TI=+4 heights.  This technique is employed because TI=-4 values are often invalid for PM soundings, as discussed further below.  In the above sample, thermalling conditions are expected to significantly diminish in the afternoon, the change in the TI=+4 height being -1726 ft.

  This subsection is similar to the previous one except that a different numerical model is utilized, the NWS's ETA model.  An advantage of this model is that it provides forecasts for tomorrow and for the day after, but this is achieved through use of a coarser grid mesh so ETA predictions are expected to be less accurate than those of the MAPS model.  The ETA 24 hour prediction likely reasonable, but the 48 hour prediction is more speculative.  Since models tend to have biases, qualitative results - such as whether tomorrow's thermal conditions are forecast to be better or worse than today's, is likely to be more valid than the actual numbers produced.  The ETA model is intended primarily to forecast broad scale weather events and its resolution is marginal for evaluating changes in the Boundary Layer.  Further details and more caveats are given below.

At present no TIP contains this subsection, due to a disk failure on my machine
This subsection is not included in the sample email presented above.
     This subsection is only present when a separate program is utilized to obtain the actual surface temperature maximum observed at Hollister on the previous day.  It gives a TI analysis of the previous day for those who actually flew yesterday and want to find out why yesterday's thermals did not agree with that forecasted or for those interested in the scientific accuracy of the TI predictions, such as prediction dependence upon Tmax accuracy or how well the morning sounding represents that afternoon's observed atmospheric conditions.  The first sentences give yesterday's observed Tmax and reviews the previous day's predictions from that morning's sounding with the NWS and WxC temperature predictions.  The actual observed Tmax is then used to provide two TI analyses, using yesterday's morning and yesterday's afternoon observed Oakland soundings.  The former indicates the "best" prediction that could have been obtained from the morning sounding while the latter gives the prediction that might have been obtained if clairvoyance allowed the afternoon atmospheric conditions to be known early in the morning.  Differences between the two analyses indicate the effect of atmospheric changes during the day, since the same Tmax is used for both.


     This section provides a detailed TI analysis of the current day's morning sounding.  For Hollister the observed sounding at Oakland is normally used, but if it is missing then a model-predicted sounding will be used instead.  The sounding and Tmax used in the analysis is noted at the end of the header line.  Following a parameter header, the TI as a function of altitude is listed and plotted.  The TI listing has several possible uses.  It allows the user to find the maximum thermal height predicted by TI values other than the TI criteria used for the program summary line.  It provides a listing of the "trigger" temperature for different altitudes, which is the surface temperature that must be exceeded for the TI=0 height to reach the indicated altitude (and the surface temperature must be 4°F warmer than the trigger temperature for the TI=-4 height to reach that altitude).  The listing also gives winds at various altitudes.  The parameter header provides several cloud parameters and a moist convection stability index, described below.
     The funky, but quite useful, text graphic plots the morning temperature profile ("o") and the predicted afternoon adiabatic profile ("*") created from mixing to the surface Tmax; these two lines intersect at the TI=0 height.  This graphic shows details of the atmospheric structure, such as the depth and strength of surface based inversions which restrict thermal development.  In addition, dew point (".") temperatures are also plotted; cloud formation is likely when the difference between the temperature and dew point is less than 4°F.  Analysis of plotted soundings is described further, with several examples, in the TIP Sounding Analysis webpage
    At the end of each of this section is "CLOUD FORMATION PARAMETERS", which gives estimates for cloudiness created by locally-generated convection.  These predictions do not describe larger-scale clouds that move into a region.  As described in its caption, extensive cloud formation (OverDevelopment) is expected over flat terrain when LCLdiff is less than zero, in which case the cloud base is expected to be the LCL (Lifting Condensation Level). There are two lines for each prediction,the proper choice depending upon whether the day's heating is expected to be important.  For AM predictions the first line is most likely to be applicable, whereas for PM predictions the second line is generally applicable.  If LCLdiff is negative, cloud formation is still possible over surrounding elevated terrain if its height is larger than LCL.  Another cloud formation parameter is the CAPE (Convective Available Potential Energy), which is positive when thunderstorm formation is predicted.    Also given is the maximum Relative Humidity in the BL, BLmaxRH, with a higher value indicating a greater likelihood of greater cloud cover; users may be able to empirically correlate this parameter to the the degree of partial cloudiness experienced at an individual site.


     This section is identical to the previous section except that the sounding analyzed is for the afternoon of the current day and will always be a model-predicted sounding.  NotaBene: the PM sounding will typically contain a model-predicted mixed layer but the TI methodology assumes that the sounding analyzed exists prior to the initiation of mixing -  so TI values below the top of the predicted mixed layer are invalid and should be ignored, as noted in the cautionary note below the section heading.  This issue is discussed in more detail below.


At present no TIP contains this section, due to a disk failure on my machine
This section is not included in the sample email presented above.
     This section is identical to the previous section except that the sounding analyzed is for the afternoon of the previous day.  One use is to aid users who are trying to empirically "calibrate" the thermal heights over elevated terrain features.  Specifically, pilots who reach the top of a dry thermal over an individual peak or ridge can use the next day's post-analysis sounding to note the TI associated with that height - if such TI values are reasonably constant, then that TI value will provide a useful prediction of the thermalling height over that individual feature.


     This section gives appropriate webpage links and provides updates which supersede the description given on this webpage.


     The amount of variation between the different estimates may be dismaying to the meteorologically naive or to those who are only comfortable with a single answer: if so, welcome to the world of weather prediction!  These models are designed to predict large-scale weather patterns and are least accurate near the ground.  The results are better than nothing, but how much better is yet to be determined.  Judgment by the user is needed to determine the most probable forecast.

ANALYSIS DETAILS (Intended for those who want detailed answers, not for the casual reader.  information here is correct, but information about some new features may be missing or cursory.)

Morning Predictions
     Use of the morning (AM) Oakland sounding for TI predictions has the advantage that "real" (observed) atmospheric conditions are used in the calculations.  Using the AM sounding also has the advantage that unambiguously valid TI heights are generally obtained because the morning atmosphere is unmixed, as the TI method assumes -  whereas use of the PM sounding can produce invalid TI heights, as discussed in the next section.  However, use of the AM sounding to predict that day's conditions requires the assumption that those conditions remain unchanged during the day except as influenced by surface mixing.  If the conditions aloft do change, there will likely be a mismatch between the sounding and the Tmax used in the TI calculation, since the Tmax reflects the expected afternoon conditions, and incorrect TI numerical predictions will result.  For example, subsidence can cause atmospheric warming aloft during the afternoon, creating a temperature inversion above the surface and causing surface temperatures to rise.  The predicted Tmax, if correctly made, will include that additional warming effect.  But the AM sounding will not include that warming, because it has not yet happened.  As a result, thermalling heights will be overpredicted.
   The AM prediction also has the advantage that the W* and B/S parameters can be calculated a the AM sounding, when Tmax is known, but not from the PM sounding.  However, if the atmospheric conditions change from those of the morning sounding the predicted B/S and W* parameters will be in error.  The sense of this error is that if the AM height predictions are too large, as for the example of the last paragraph, then the predicted B/S and W* will also be too large.

Afternoon Predictions
     The afternoon (PM) predictions are of particular interest because most soaring flights occur in the afternoon.  Use of a PM sounding for TI predictions allows the inclusion of atmospheric changes which occur after the AM sounding time.  Although the PM forecasts must be based upon model forecast soundings, rather than observations, in general the forecast PM soundings have agreed well with the observed PM sounding and also with thermalling experience.  However, difficulties arise in the computation of summary numbers from those soundings, of which users should be aware when utilizing the PM forecasts.
     The TI method assumes that the PM temperature profile will be an adiabat, representing the layer mixed by the thermals, extending upward from the surface Tmax to the unmixed atmosphere and all TI values are intended to be referenced to an unmixed sounding. Numerical TI estimates are therefore invalid for heights below the top of the model-predicted mixing layer.  The model calculates its own PM Tmax, which can differ from the TI-assumed Tmax, and its own PM mixing layer,  with a temperature lapse rate which may not be exactly adiabatic.  As a result, the TI-assumed adiabat can intersect the PM sounding in the model-predicted mixed layer or even not intersect the model PM sounding at all.  Predicted PM TI=-4 and TI=0 heights, particularly the former, are susceptible to being below the model-predicted mixing height should that occur invalid and misleading PM height estimates result.  The PM TI=+4 height, being higher in the atmosphere, generally is above the model-predicted mixing layer and remains valid.  Therefore the PM predictions are given as the change of the TI=+4 height from AM to PM, in feet.
    Thus PM predictions must made in a relative fashion, by assuming a bettering/worsening of all thermalling conditions (i.e. the TI=-4 and TI=0 heights and the B/S and W* parameters) similar to the change in the TI=+4 height.  The PM height change at the TI=-4 and TI=0 heights obviously will differ from that at the TI=+4 height, but the TI=+4 height change is at least a reasonable approximation to those height changes.
     An additional difficulty with the PM prediction is that calculation of both the B/S and W* parameters require knowledge of the surface heat flux and at present this cannot be estimated from the PM sounding, so no B/S or W* parameters are given for the PM times.
      A large change between AM and PM forecasts generally reflects unsettled and changing atmospheric conditions, indicating that the forecast uncertainty is large.   In such cases examination of the detailed PM sounding plot is suggested to evaluate visually the forecast depth of the mixing layer, within which the temperature profile is approximately adiabatic.  This requires some knowledge on the part of the user, but this is a case where human intelligence will provide a better evaluation than the simplicities used in a computer analysis.  The information necessary for such interpretations is provided in the TIP Sounding Analysis webpage.

TI Uncertainty
     The uncertainty in a given prediction of maximum thermalling height results primarily from uncertainty in:
      (1) the predicted surface temperature maximum,
      (2) the difference between the sounding and actual conditions while soaring,
      (3) the TI criterion used to estimate the maximum thermalling height.
Prediction error can also result from assumptions inherent in the analysis.  For instance, the TI prediction method assumes flatland conditions.  Further, the Hollister TI prediction is based upon the Oakland sounding released over the coastal plain.  Atmospheric profiles over the mountains differ from those over the plain since, for example, upslope flow alters environmental conditions there.  Thermalling heights in mountainous terrain are often significantly higher are than estimated by a simple TI criterion due to local variations induced by the terrain.  The TI also assumes non-condensing convection (though the density effect of water vapor is considered).  Cloud formation during the day releases latent heat which provides additional buoyancy aloft and increases updraft intensity,  producing higher thermalling heights than calculated by TI prediction.  Large cloud systems can also significantly alter the atmospheric environmental conditions.  Use of a single sounding also assumes time-invariant conditions.  During rapidly changing conditions, such as occurs with a frontal passage, the atmospheric profile during the day may differ considerably from that of the early morning hours.  Of course, the TI estimates consider only thermal lift, not frontal or convergence lift.

Tmax Uncertainty
     The degree of uncertainty of the predicted temperature maximum is important for evaluating TI prediction uncertainty.  By noting differences between the NWS and WxC temperature predictions over several weeks, the user will gain some appreciation of the amount of error inherent in such predictions at Hollister.  I have made a statistical comparison of observed (as reported by Hollister's Maze middle school website) to predicted Hollister temperatures for a 8 month period.  The NWS predictions average 1.6°F warmer than the the observed temperature maxima and thus tend to predict thermalling heights that are too high; however, their day-to-day variability (a standard deviation of 9.6°F) is close to that actually occurring at Hollister (having a standard deviation of 9.9°F).  The WxC predictions have a smaller bias than the NWS, averaging only 0.3°F warmer than the observed maxima, but they exhibit less day-to-day variability (a standard deviation of 7.5°F) than observed at Hollister and thus tend to miss days having stronger-than-average heating.  The average absolute error of the daily NWS and WxC forecasts are 3.8°F and 4.3°F respectively, values which are somewhat larger than the TI criterion uncertainty.  The absolute error of the average of the NWS and WxC predicted temperatures is somewhat less (3.7°F) than the error of either alone, indicating that the two forecast models err for different reasons, so for Hollister a good strategy for dealing with two differing surface temperature predictions is to simply average them, as given by the AVG result.   At one time the TIP used the AVG  Tmax instead of the NWS Tmax for its summary results, but subsequent TIP use at other sites found the WxC Tmax predictions to be significantly in error at inland sites, particularly in the summer, so to avoid misleading predictions at other sites the summary numbers now rely only on the NWS Tmax.
     An overall conclusion is that the Tmax forecast error is large compared to typical temperature variations.  A historical analysis of Hollister TIs finds that the average Tmax forecast error corresponds to a TI forecast error of roughly 850 ft.  It should be noted that the historical analysis only considers 8 months, including summer months when Tmax can be difficult to forecast because it is greatly affected by non-local phenomena such as the sea breeze and fog/stratus created by the ocean - so the average error during the other months may be lower.  Still, we need better Tmax predictions!  Tell your congressman to give the NWS more money!  It is likely that the principal error in the TI forecasts is due to Tmax error.
     The source of the NWS temperature prediction for Hollister is a NWS website while the source of the WxC prediction is a Weather Channel website.  I have found that the WxC often "updates" it's surface temperature prediction around 8:15 AMpst, after the TI prediction program has run, so on some days a later calculation would give a different TI result based upon a possibly more accurate temperature prediction.  Should be TI program be run at a later time, such as 8:30, to incorporate this update?  At present I believe it best to have as early a prediction as possible, but that may change.

Evaluation of TI Prediction Uncertainty
     The TIP email provides many outputs to aid in the evaluation of TI uncertainty.  For the same sounding, comparison of TI predictions from differing Tmax values indicates the TI uncertainty due to Tmax uncertainly.  For the same Tmax value, comparison of TI predictions from differing model soundings indicates the TI uncertainty due to model error.  In addition to those uncertainties, there is uncertainty resulting from possible cloud formation or terrain effects - but these cannot be presently evaluated.

Critical Height (Hcrit)
     More exactly described as the Height of Critical Updraft Velocity, this parameter estimates the maximum thermalling height over flat terrain under cloudless conditions.  Hcrit is obtained from an averaged profile of thermal updraft velocity vs. height (as obtained from aircraft measurements by Lenschow and Stephens) by assuming the thermal strength W* to be the maximum updraft velocity in the BL and computing the height at which the updraft velocity drops below 225 fpm (a rough estimate of the sink rate of a sailplane or hang glider actively turning and maneuvering to remain with in a thermal).  The intent is to obtain a better estimate of the maximum thermalling height than is provided by the TI=-4 height, since the latter often has no meaning for a PM sounding; in addition, the TI=-4 height seems to overestimate the maximum thermalling height under weaker conditions, and Hcrit is expected to be smaller than the TI=-4 height under such conditions.  But the superiority of Hcrit over the TI=-4 height as a predictor of the maximum thermal soaring height has not yet been been demonstrated in practice so I would be interested to learn of any flights under conditions when the two differ significantly, which would help such an evaluation.  Note that if W* is less than 225 fpm, then Hcrit is predicted to be the surface.

Buoyancy/Shear Ratio (B/S)
     A shortcoming of the TI index is that it indicates the height to which mixing should occur but not all mixing is equally useful to glider pilots.  Mixing can be produced both by thermals and by wind shear, but only thermals produce the relatively large updrafts needed for soaring.  To help evaluate the degree to which the day's mixing is convectively driven, a thermal "hot air" parameter "B/S" (sic) has been added to predictions in the email body, representing the ratio between Buoyancy and Shear production of turbulence.  A small B/S value indicates shear is likely a significant problem - at present the best guidance I have, based upon a limited number of pilot reports, is that on days with B/S of 5 or less the thermals are likely to be too broken to be usable.  At B/S of 10 or above shear is likely not a significant factor.  Actual experience which would indicate those criteria should be altered, or confirming those estimates, would be appreciated.   Because I believe the "B/S" ratio can be a significant modifier of thermalling conditions, it is included in the Subject line following the two TI heights.
     For those interested in more scientific detail, the B/S ratio is not an empirical approach but is based upon the non-dimensional number used to distinguish between "buoyancy dominated" and "shear dominated" BLs (and those in between).  It is the ratio of the "buoyant production of turbulent kinetic energy" to the "shear production of turbulent kinetic energy" with both being well defined terms.  However, the cross-over criterion between "workable" and "unworkable" thermals must be determined empirically (and for that matter there is no sharp cut-off between the two cases).

Thermal Strength (W*)
     The program's method of estimating the thermal strength W* is unique within the soaring community, to my knowledge, and so will be discussed in some detail.  Note that this prediction is intended to forecast the upward velocity of air within the thermal and will never be negative.  It is well established, both theoretically and experimentally, that vertical motion in a cloud-free convective boundary layer is proportional to the cube root of the product of a buoyancy constant, the boundary layer depth (or thermal depth), and the surface heat flux. This produces a quantity with the units of velocity, known as the convective velocity W* ("w star").  It makes physical sense that a thermal's strength will depend upon the amount of heat entering into the atmosphere at the ground, and the thermal height is an important factor because a rising thermal bubble will achieve a higher velocity if it accelerates for a longer time.  The proportionality constant between W* and vertical velocity in a thermal should be approximately one, its exact value depending upon such factors as the area over which the vertical velocity is averaged (since a thermal's core is stronger than its periphery) and thus will depend upon the thermalling radius of the glider, for example. 
     Of the needed factors, the buoyancy constant is known and the thermal depth can be assumed to be the TI=0 height.  [Note: the remainder of this paragraph descibes the surface heating estimate utilized by TIPs and does not apply to BLIPMAPs, whcih utilize the known model surface heating so no such calculation is required.]  However, surface heat flux predictions are not available on the web for input to the TIPs (although the forecast models do need to compute it when predicting the maximum temperature).  Therefore I have attempted to back that number out of the available data, though there will be some error in the result, by using the fact that the total heat input into the boundary layer can be determined by vertically integrating the temperature difference between the early morning sounding and the afternoon temperature resulting from adiabatic mixing produced by the predicted surface temperature (this requires the assumption that there is no net transport of heat by either horizontal or vertical mean winds and will be in error if that assumption is not met, such as when atmospheric conditions above the mixing layer are changing).  A relationship between that total heat flux and the maximum heat flux during the day, the factor of interest, is needed so I have assumed the heating to be a sinusoid extending over the daylight hours. 
     For now I have simply set the problematic proportionality factor to one and it is gratifying to find that without having to resort to any empiricism whatsoever the predicted vertical velocities are already quantitatively very realistic, when compared to the "Soaring Index" (SI) results presently accepted for the Reno area .  With further experience the proportionality constant might be adjusted slightly, but the present results are considered very reasonable and more accurate than the "SI" predictions for Hollister under weak convective conditions.  I expect the present predictions to be accurate within a factor of two, though of course a range of thermal strengths occur at any one time.  One should note that W* follows the rule that deep thermals tend to be strong thermals, so larger TI=0 heights will generally be associated with stronger thermal lift. 

Cloud Parameters
     Several cloud parameters are provided in the header accompanying the detailed sounding analysis.  There is great potential to misunderstand these predictions!  All these parameters apply only to clouds which develop locally due to convection, not to clouds which move into the area or which occur above the Boundary Layer.  The "Sfc-Lift Cond Lev" is the height at which small clouds might develop due to thermal convection from the surface; this is the lowest level at which cloud bases might occur, but usually unreasonably low because it considers only the surface dew point. The Lifting Condensation Level, LCL, is a height at which the base of more extensive clouds would develop due to surface heating, since the moisture is averaged over a layer thickness near the surface.  The Ford cloud base estimate is a similar prediction using a layer of different depth and generally predicts a somewhat higher cloud base than the LCL. The Convective Condensation Level (CCL) is similar to the LCL but also evaluates limitations resulting from the actual atmospheric temperature profile and so should always be higher than the LCL.
     Also included is an index of moist instability called the "Lifted Index (LI)": the smaller, or more negative, the number the greater the chance of strong convective clouds ("towering cumulus") developing.  Of the LI values available on the web, I chose the LI for 700 mb (around 11,000 ftMSL) because it is more appropriate to local conditions than the "normal" LI, which uses a temperature difference evaluated at 500 mb (around 18,000 ftMSL) to forecast the large thunderstorms that develop in the Midwest.  But while there is guidance available on use of the "normal" LI, since that has been "calibrated" by experience (see Chris Ruf's website, which is also a good general source for soaring meteorology - but note that all stability indexes described there are for very large convective systems).  I am not aware of any guidance on what LI@700 values are associated with what kinds of clouds, even though LI@700 is a commonly reported parameter.
     The above paragraphs apply to scattered/broken clouds which develop locally due to convection.  It is also possible to use sounding profiles of temperature and dew point to predict overcast/broken clouds which either move into the the area or develop locally, as described in the TIP Sounding Analysis webpage.

Neglect of Cloud-Generated Buoyancy
     Clouds mark thermals, but they also add buoyancy to the thermals through the release of latent heat of condensation.  The TIP predictions, however, assume that thermals are driven entirely by heating at the earth's surface, so this release of heat aloft is not included in the TIP buoyancy estimates.  With cloud formation the maximum height to which a glider can climb now becomes limited by the cloud base height, not by the top of the thermal (which is the top of the cloud), so the actual soaring height cannot be equated to the thermal height, as the TIP TI height prediction assumes.  This dissociation is particularly apparent when maximum lift is found at cloud base, trying to suck the glider into the cloud - clearly the glider is then not at the top of the thermal!   (It should be noted that the above description applies to convective clouds in their growth stage - at later times the clouds can "overdevelop", forming an overcast which blocks sunlight from reaching the surface, which in turn reduces the surface buoyancy and weakens the thermals.)
     When convective clouds form, therefore, the actual thermal top, the updraft strength W*, and the Buoyancy/Windshear ratio B/S are all larger than the TIP prediction due to buoyancy generated aloft.  Further, these parameters are increased more by deep cloud convection than by shallow puffy cumulus.  However, the cloud base is expected to be below the maximum thermalling height predicted by the TIP since the condensation initially occurs in a dry thermal, below the thermal top; the deeper the cloud, the larger the difference between the cloud base and the predicted maximum thermalling height.
     Because cloud-generated buoyancy is so significant, the best soaring conditions often occur when clouds form so neglect of this effect is a significant deficiency in the TI method.  Unfortunately, inclusion of cloud-generated buoyancy would be difficult since cloud formation is hard to forecast accurately and since even relatively small amounts of condensation can significantly affect the thermal strength.  It is probably best to regard the TI forecasts as a forecast of "minimum" thermalling conditions in the absence of clouds, with cloud formation generally expected to strengthen thermalling conditions.  Here "thermalling conditions" should be interpreted as meaning updraft strength, since the maximum soaring height is now determined by the cloud base rather than by the maximum thermal height.

Model Forecasts
     Forecast soundings are obtained from the research-mode FSL "MAPS" model, which predicts out to 36 hours at best, and the operational NWS "ETA" model, which predicts out to 48 hours.  Both models are intended to provide predictions of larger scale weather systems and have limitations which affect their ability to predict accurate soundings near the surface.  For example, the vertical resolution of the available ETA model data is around 500 feet at 3000 ftMSL, so changes in TI heights of less than 500 ft do not have great significance (the Eta model actually uses a finer grid resolution, but its output is degraded to supply values at constant pressure levels).  MAPS has a somewhat better vertical resolution of 350 ft at 3000 ftMSL
     At present the program analyzes soundings forecast for Oakland's coordinates.  I will discuss this choice in detail, since it might seem that a location closer to Hollister would be more appropriate.  In addition, this illustrates pitfalls that can result from unwary acceptance of information available on the web.  For the NWS models forecast soundings are only available at certain coordinates, coinciding with airport locations.  For the MAPS model, however,  sounding data is available for any grid point location at the FSL sounding website, which is the program's MAPS data source.  The first point I wish to emphasize is that the model horizontal resolution is 20 km for MAPS and 32 km for Eta, which greatly limits their ability to resolve the influence of our local terrain.  Use of a 20 km grid, for example, means that a forecasted variable is expected to be the average of that variable over a 20x20 km area.  However, to avoid the introduction of numerical noise the topography must be averaged over several grid intervals, so the net result is that the model topography only resolves features which are over 40 km in width whereas our coastal mountains are about 40 km wide so the resolution is marginal at best.  The second point is that the models use "envelope" topography, which means that instead of employing the average elevation over each 20x20 km area the maximum elevation is used.  This has advantages for predicting air flow, since winds channeled by the Sierra mountains, for example, are affected by the maximum elevation of that mountain ridge, not by its 20x20km average elevation.  However, this choice leads to difficulties when trying to forecast the atmospheric structure near the surface, upon which a TI calculation is dependent.  (In fact, when the MAPS model forecasts its own surface temperatures it uses a different grid, one more representative of the average elevation.)   If the naif attempts to forecast Hollister conditions by simply using the model grid point closest to Hollister's coordinates, he will obtain a sounding having a "surface" of over 1000 ft, considerably above Hollister's actual surface elevation of 230 ftMSL (and even above it's traffic pattern!).  The error in surface elevation is large relative to the average TI=-4 height, which is around 3000 ftMSL. and is not an appropriate surface elevation to predict TI heights at Hollister. This result is a combined consequence of the envelope topography and the coarse horizontal resolution which greatly broadens the Diablo mt range, as seen on the 20km MAPS topography.  One might obtain a lower elevation, 660 ftMSL, by choosing the next grid point towards the ocean, but the ocean influence is then exaggerated.  For the present, I have decided that it is best to forecast for the coordinates of Oakland airport, which is located roughly the same distance from the ocean as Hollister and also surrounded by higher terrain, since the model surface elevation there is 540-590 ft (depending upon the model) and surface temperature predictions made for the Hollister elevation are then more appropriate.  The difference in location between Hollister and OAK parallels the ocean and this horizontal shift is considered to have less effect on TI predictions than other factors.  Also, from the scientific point of view a comparison between forecasts and observations both made for OAK is a more meaningful test of forecast accuracy.
     Because models have biases, they are often better at forecasting changes than absolute values. Hence the TI predictions in the model forecast sections include changes from the model TI prediction for the current morning (when an observed sounding is available for comparison)..
     I hope that that recognition of the various model limitations does not make the user overly pessimistic concerning the model forecasts.  I believe that they can be of value, but on the other hand the user should not expect too much or accept them blindly for the models are not magical.  The models tend to provide large-scale "broad brush" differences while smoothing out the small-scale variations which also exist in the atmosphere.  The uncertainties resulting from model limitations are one reason why the TIP provides several forecasts rather than a single one.  I do think you will find the forecasts useful, but they are subject to error.  At some later time I intend to do a historical analysis evaluating the accuracy of the forecasts and look for model biases; once such biases are known they can be used to better evaluate future TI predictions.

Adjustment of Surface Elevation
     Because the surface elevations of the sundry observed and forecast soundings differ,  a method of surface elevation adjustment has been incorporated into the model.  The compensation is not perfect, requiring some assumptions when the assumed surface lies below the bottom of a sounding, but is better than nothing.  The resulting TI adjustments lessen the difference between TI heights obtained for the morning observed sounding and those obtained from the model analyses for that time, but are relatively small for the Hollister predictions,

Day After Tomorrow Tmax
     To make comparisons between "tomorrow" and "day after tomorrow" TI predictions more meaningful, since NWS Tmax predictions are being used for the previous days but only WxC Tmax predictions are available for the "day after tomorrow", I am adjusting the "day after tomorrow" Tmax by assuming that the difference between the NWS and WxC Tmaxs will be the same on the "day after tomorrow" as it was "tomorrow". A "wxc" identifier (rather than "WxC") indicates that this adjustment has been applied to the WxC Tmax.

TI Analysis Method
     Further information on the TI analysis methodology can be found at Kevin Ford's TI calculation description, since the TI calculation is based upon the Ford subroutine.  The Ford calculation has been altered to allow TI calculation at a surface elevation which differs from that of the sounding, but this requires some assumptions about the atmospheric structure when the assumed surface elevation lies below the bottom of the sounding.

Program Description
     The program was originally developed to obtain TI analyses of observed morning soundings from Kevin Ford's TI calculation website, but that site's data sources were unreliable so the program was converted to instead gather the sounding data itself and compute the TI internally, using a modified version of the Ford routine.  This also allows TI calculations to be made from forecast and archived sounding data, all data except for the MAPS soundings now being obtained from the Storm Chaser's Weather Machine.  If any sounding data is missing when the program is first run, the program will continue to try to gather data for 20 minutes before sending an email with analyses of the available data.  If temperature data is not immediately obtained from either the NWS or WxC site, the program continues to query that site for 30 minutes prior to abandoning its attempts.  If temperature data is not available for either site, then an error email is sent.  Analysis of 8 months of historical sounding data has shown the actual computations of TI, W*, etc. to be robust, so long as valid data is obtained from the web.  However, the program depends upon a chain of events (available data, sites running, etc.) and that chain may break on any given day.  In particular the program relies upon the formats from all web sites remaining unchanged.  If the program "breaks" and errors occur, I will try to fix the problem or make the program smarter.  The program is a combination of Unix and Perl scripts, using the "Curl" open source software to interact with the web, running on an SGI computer.

     The TIP treatment of cloud influences is weak, due to (1) the complex dependence of soaring upon clouds, since clouds aid soaring by increasing thermal updraft strength through the release of buoyancy aloft but also limit soaring since a cloud base can prevent pilots from ascending to the top of a thermal (2) clouds formation is difficult to predict, due to a cloud's sensitive dependence upon humidity which is inherently difficult to predict using numerical models.

Contacting the Author
     Any opinions on whether the reference TI criteria should be changed, or adjusted for soaring locations not directly above Hollister, would be of interest if based upon actual gliding experience at Hollister.  Also, I will attempt to answer any email queries containing questions not covered in the description above.  I should warn such inquirers, however, that I am a research meteorologist, not a forecaster, and moreover my specialty is the atmospheric boundary layer where thermals develop, so while I will try to give short answers I will not make complex meteorological problems simplistic.  My decision not to give a single number supposedly representing predicted thermalling conditions at Hollister is an example of that mindset. I can be reached at   Or you can read about what I do for a living at the DrJack home page kindly provided by Webbnet.


     To establish a baseline providing perspective on the variation to be expected for TI forecasts at Hollister, I performed a historical TI analysis using the observed temperature maxima and Oakland sounding data that I have been archiving since April of this year.  This analysis is given below for those interested.  I also have the TI results for individual days, should anyone have an interest in a particular day over this period.
                ANALYSIS OF HOLLISTER DATA FOR 2000      
         Average of indicated number of days in each month
         followed by daily variation (±standard deviation) 
     [Uses OAK sounding and observed Tmax at Hollister Maze MS]            

MONTH DAYS      Tmax      TI=-3.6F        TI=0         W          SI
----- ----   ---------   ----------   ----------   --------   --------
 Apr     7   76.0 ±7.3   4885 ±1338   5724 ±1545   545 ±101   356 ±158
 May    30   75.7 ±9.9   3837 ±1594   5009 ±2212   465 ±138   329 ±274
 Jun    24   78.4 ±9.4   2681 ±1244   3227 ±1632   360 ±120   255 ± 85
 Jul    21   75.5 ±6.6   2649 ± 663   2966 ± 807   360 ± 72   156 ±113
 Aug    30   81.3 ±7.3   2395 ± 724   2805 ±1115   363 ± 90   192 ± 75
 Sep    18   81.8 ±9.0   3425 ±2220   4310 ±2368   426 ±175   292 ±225
 Oct    27   71.7 ±6.6   2717 ± 980   4030 ±1528   395 ± 95   285 ±146
 Nov    15   63.3 ±6.1   2756 ± 973   4073 ±1442   378 ± 89   235 ±165
 ALL   172   76.0 ±9.5   3008 ±1422   3847 ±1871   400 ±124   256 ±184
 Return to the top of this page
 Link to TIP Sounding Analysis webpage
 Link to the latest TIP Forecasts, a listing of all TIP sites, and TIP email subscription information
 Link to TIP user's experiences, to learn how other pilots use the TIP
 Link to DrJack home page