Binbin ZHou
                                                         SAIC@NOAA/NWS/NCEP/EMC

1. Motivation, Background and Goal

        The NCEP's Short Range Ensemble Forecast(SREF) provides probabilistic forecast information on the
meso-scale, and has undergone development by EMC/NCEP since 1996 (Du, et al, 2003). The SREF  system
has been run operationally at NCEP for two years. The SREF Aviation project, a subset of the NCEP SREF
system, seeks to further improve the utility of SREF products by applying ensemble techniques to aviation
weather forecasts.

     Basically, an ensemble forecast is composed of a set of models or "ensemble  members" run at the same
time. The members can be made from the same model using different initial/ boundaryconditions or physics
schemes, from a multi-model system, or from some combination of the above. The output from the various
ensemble members is used to perform statisticalcomputations to create probabilistic forecast information
such as an ensemble mean, spread (standard deviation),  probability of occurence, etc.

      The aviation weather forecast requires additional forecast products than are typically used for standard
weather forecasting. Aviation forecasting focuses on short time scales. Substantial efforts have been made
to improve the aviation weather forecast, such as utilizing more sophisticated meteorological models and
employing  more complex algorithms to compute aviation weather variables such as icing and turbulence.
Traditionally,  deterministic weather forecast approaches were used.  It has been recognized that the
computer-generated weather system does not always result in a perfect forecast  because of the chaotic
characteristics of the atmosphere.  Thus, probabilistic or ensemble-based forecast from the SREF system
can provideadditionalinformation in terms of forecast confidence and uncertainty, and can be useful to
aviation weather forecasters.

     Please keep in mind that current SREF-aviation products are still under development and very experimental.
As the SREF system is updated and SREF ensemble outputs are changed,  SREF aviation products will be
improved or make use of new algorithms as well.  Also, SREF aviation products will be upgraded based on user
feedback and requirements. We are trying to provide a useful, efficient, and user-friendly tool to aviation
weather forecasters.

    Also please keep in mind that the primary goal of SREF system is not only improve the forecast"preciseness",
but  more important, to quantify the uncertainties in the initial conditions and the models or confidence interms
of a set of statistics based on the ensemble  members. It  gives out information on what the most likely outcome
might be.

 The SREF documentation is located here.   Additional ensemble training information can be found here.

2. Ensemble models and members

  SREF runs at NCEP twice a day; first at 09Z, then at 21Z.  The forecasts run out to 63 hours,and data is output
at 3-hour intervals.

The current SREF ensemble is made up of 15 members, based on 32km Eta model and 40km RSM models.

While the current SREF system has 15 members, not every aviation product is made from all 15 SREF
members.  The Eta and RSM have different post-processed data output, and don'talways contain the
variables needed for  a particular product.  The table below lists the variousaviation products and which
of the models/physics schemes were used to derive them:
 

Variable	Model Schemes	Members	Forecast Type
Icing	All 3	15	probability
Turbulence	All 3	15	probability
Jet Stream	All 3	15	probability
Low level  wind shear	All 3	15	probability, mean&spread
Convective storm speed	All 3	15	mean&spread
Convection	Eta	10	mean&spread
10 m wind	All 3	15	probability, mean&spread
Precipitation type	Eta	10	Probability
Tropopause	Eta	10	mean&spread
Frozen height	Eta	10	mean&spread
Cloud	Eta	10	probability, mean&spread
LIFR,IFR,MVFR,VFR	Eta	10	probability
Ceiling	Eta	10	mean&spread
Visibility	Eta	10	mean&spread

3. Multiple initial conditions and boundary conditions

The current SREF system uses a breeding method to create different ensemble members. That is, using a
random perturbation field, we add or subtract from the original initial condition to create a pair of new perturbed
initial conditions.

The boundary conditions come from NCEP's global ensemble output.

4. Model timelines, Data display, Output format and Product update times

      The SREF-aviation production is divided into 3 parts:

      (1) Computation: Copy SREF files and run models, 30 minutes total

                            09Z cycle:    9:30-10:00Z
                            21Z cycle:  21:30-22:00Z

       (2) Image creation: Produce image (GIF) files from model output, about 30minutes

       (3) Distribution:  Send GIF files to the NCEP server, 30 minutes.  This also includes
              updating the html files' time stamp on the server

    The total time used to produce the new forecast is one and half hours, finishing at about 11:00Z for  the 09Z  run
and 23:00Z for the 21Z run.

     The images are produced by GRADS directly from output files in GRIB format. Each weather parameter  has
22 images so that the results can be animated. Thus, your browser needs to download all 22 of the GIF  files before
a new forecast product can be displayed.  Those who have low-speed Internet connections or  less powerful
computers will notice some delay in viewing SREF products.

   The SREF-aviation computation data are stored in GRIB-extension format. The extension is NCEP's  special
addition to regular GRIB format for storing unique ensemble data. Because of the limits to NCEP's on-line system
storage, SREF-aviation computation results are kept only 3 days. If users want these data, we can send them files
in GRIB-extension format only. A list of the weather parameters available in the  SREF-aviation GRIB-extension
file can be found here.   Background information on the NCEP GRIB-extension format is also available on our
website.

SREF BUFR files (for specific locations) are not available at present.
 

5. Ensemble variables

        There are many variables used in the ensemble forecast to create the probabilistic forecasts of mean, spread,
"spaghetti", and probability of occurrence.  In SREF-aviation, only mean, spread and probability are used.  The
"spaghetti plots" of individual ensemble members are not shown.

   Definition-ensemble mean: The average value of all members.

   Definition-spread: The standard deviation with respect to ensemble mean. The spread of a variable indicates the
forecast diversity of the ensemble mean among all the members.

   Definition-probability:  For an event, out of all members, in how many members does the event occur.  Probability
of aevent (variable) is expressedas a percentage.

    For example, a probability of icing at FL180 equals to 60 % means that 9 of the 15 members, or 60% of the runs
(9/15 x 100 = 60 %), have icing at that level.

    In the images, the mean is denoted by contours, spread by colored shading, and probability can be denoted either
by contours or shading.

6. Definition and method of computation for each variable

(1) Icing (In-Flight) probability

    In the real world, icing is a phenomenon in which ice particles or water droplets freeze on an aircraft, particularly
the wings. The icing forecast predicts where the kind of environment favorable for icing events is likely to develop.

    The icing environment depends mostly on temperature, humidity and cloud water content at certain levels. There
are many algorithms to predict such an environment, ranging from the simple to the very complex. Since the current
RSM lacks cloud water content,we use a T-RH (temperature-relative humidity) algorithm to predict icing instead
of a method  using cloud water. Thus, current icing products can't distinguish between frozen ice and supercooled
large droplets (SLD).  The new version of RSM, which has cloud water content in its post-processed output, will
be soon be implemented at NCEP. Then we can use more sophisticated icing algorithms.

   The range of T and RH for icing event is:

            -10 C < T < 0 C   and   RH > 70%

   Icing is computed at 8 flight levels, from low to high:

       FL000:    surface
       FL030:    3000 ft (900mb)
       FL060:    6000 ft (800mb)
       FL090:    9000 ft (725mb)
       FL120:  12000 ft (650mb)
       FL150:  15000 ft (575mb)
       FL180:  18000 ft (500mb)
       FL240:  24000 ft (400mb)

       At certain levels and model gridpoints, if temperature and relative humidity both meet the icing criteria above,
then the icing condition is "yes".  If M ensemble members out of all N members predict icing conditions at the same
level and gridpoint, then the icing probability is

                                                 M/N x100 %

(2) Clear Air Turbulence (CAT) Probability

     CAT is defined as a turbulence event occuring at high, cloudless altitudes.  Like icing, there are morethan a dozen
methods to predict the CAT intensities. The SREF aviation turbulence product uses the Ellrod(Gary P. Ellrod, 1992)
algorithm.

     The Ellrod algorithm classifies CAT intensity into 3 categories: LIGHT to MODERATE, MODERATE, and
MODERATE to SEVERE, and uses a turbulence intensity index to determine which categorythe event belongs to.

      The Ellrod turbulence intensity index is defined as a function of stretching deformation (DST), shearing deforma-
tion (DSH), vertical wind shear (VWS) and convergence (CVG).  These four factors implicitly reflect the effects of
wind gradients in both the horizontal and vertical directions, and the temperature gradient as well.  In detail,

       Index = VWS * [ DEF + CVG ]

        VWS = sqrt(dU*dU + dV*dV) / dz,   where U , V are wind u,v components (in m/s), respectively

       DEF = sqrt(DST*DST + DSH*DSH)

       DST = dU/dx - dV/dy

       DSH = dV/dx + dU/dy

       CVG = - (dU/dx + dV/dy)

      As pointed out by Ellrod, defining the 3-category index values for different CATintensities depends on resolution.
For different horizontal resolutions, Ellrodsuggested using different index values.  For the SREF-aviation CAT,  all
computations use the following index values:

Turbulence intensity	LIGHT-MODERATE	MODERATE	MODERATE-SEVERE
Index (x0.00001)	4	8	12

 

       Please note that the turbulence intensity index is computed between 2 levels, i.e. within a layer instead of  at a
single level. In SREF-aviation products, the turbulence intensity is evaluated within 8 continuous layers, from low
to high:

        FL210-FL180  or   450-500mb
        FL240-FL210  or   400-450mb
        FL270-FL240  or   350-400mb
        FL300-FL270  or   300-350mb
        FL330-FL300  or   275-300mb
        FL360-FL330  or   225-275mb
        FL390-FL360  or   200-225mb
        FL420-FL390  or   175-200mb

(3) Sky (Cloud) Cover Probability

       Sky or cloud cover is divided into 4 categories, clear, scattered, broken, and overcast. Each category has a
range of cloud amount in fractions or percentages, as listed below:

Cloud Cover	Clear	Scattered	Broken	Overcast
Cloud amount	0/8	1/8 - 4/8	5/8-7/8	8/8
Cloud cover %	0	1-50	51-100	100

      Eta and RSM models treat cloud cover as a percentage instead of a category or fraction.   SREF-aviation
products use the seamless range of percentages shown in theabove table. So the probability of a particular cloud
cover category is the percentage of ensemble members predicting a cloud cover percentage  that falls in the
category.  For example, when the probability of a clear sky at one gridpoint is 50%, that means half the ensemble
members predict clear (0% cloud cover) at that gridpoint.

(4)Flight Restriction (LIFR, IFR, MVFR, VFR) probability

LIFR:     Low Instrument Flight Rules
IFR:       Instrument Flight Rules
MVFR:  Marginal Visual Flight Rules
VFR:     Visual Flight Rules

Restriction categories are defined by cloud base height (ceiling),  surface visibility and cloud amount category, as
shown below:

Category	Ceiling (ft)		Visibility(mile)		Cloud Category
LIFR	less than 500	and/or	less than 1	and	Broken or Overcast
IFR	500 to 1000	and/or	1 to 3	and	Broken or Overcast
MVFR	1000 to 3000	and/or	3 to 5	and	Broken or Overcast
VFR	more than 3000	and	more than 5	or	Clear or Scattered

Here is the pseudo code that describes the computation:

    For all members N, find how many cloud amounts are broken and overcast, say M.
             If M=0
            VFR probability = N/N=100 %, i.e. no broken and overcast clouds
   Else
            (N-M) members belong to VFR,  some of M belong to either LIFR, or IFR, or MVFR, or VFR,

          Do the following:
              Within M, find how many members L1 with cloud base <500 ft or Visb <1 mile,
                        LIFR probability = L1/N
             Within M, find how many members L2 with 500<=cloudbase<1000 ft or 1<=Visb<3 mile,
                       IFR probability = L2/N
             Within M, find how many members L3 with 1000<=cloudbase<3000 ft or 3<=Visb<5 mile,
                       MVFR probability = L3/N
            Then the Left in M, i.e. M-L1-L2-L3 members are belong to VFR,
                       VFR probability = [(M-L1-L2-L3) + (N-M)] / N = (N-L1-L2-L3) / N
          Done
    End if

  Note that in the computation, "AND/OR" means either "AND" or "OR".  When using "AND" the evaluation is
more conservative. We use "OR" in our product.

  Ceiling and visibility are also computed and listed for reference. (See sections 9 & 10).

(5) Jet Stream probability

   Jet stream probability is computed at 3 levels: 4,500 ft, 18,000 ft and 34,000 ft with 3 wind speed thresholds at
each level.

      At 4,500 ft, the thresholds are 20, 40, 60 knots
      At 18,000 and 34,000 ft the thresholds are 60, 80, 100 knots

      e.g., at 18,000 ft and a threshold of 80 knots, the image will show the probability of wind
      speed > 80 knots at 18,000 ft.

(6)Tropopause

        (I) Tropopause height: mean and spread in feet
       (II) Temperature at tropopause height level: mean  and spread in C

(7)Frozen-level height

   The first height above the ground where air temperature is below 0 C: mean and height spread.

(8)Total Cloud cover

      (I)  Mean and Spread:  averaged cloud total amount (%) and spread (%).
              example: cloud contour = 80 and shade color is yellow, which indicates that the average  cloud  amount is
                                 80%,  and standard deviation (diversity among members) is 20 %

      (II) Max cloud amount:  maximum cloud amount from all ensemble members.

     (III) Min cloud amount:  minimum cloud amount from all ensemble members.

              example:  Min cloud amount at a region = 60%, which indicates the ensemble mean  cloud  coverage for
                                   this region is AT LEAST  60 %.

(9)Visibility

        Defined as the horizontal visibility at the surface. Visibility is not a model variable but is computed from post
processing output, based on surface humidity, rain, etc. Two notes:

      (a) Eta and RSM do not consider the haze directly, but haze forms in high humidity. Thus SREFvisibilities
             implicitly include low visibility caused by haze.

      (b) Dust storms cannot be detected in the Eta or RSM.
 

(10)Ceiling

       Conditional ceiling, defined as the height of the lowest layer of cloud when the sky cover is broken or
overcast;  the mean and spread are computed in feet.  If no cloud or sky cover is less 50%, then there is no
ceiling, then its value is defined as model default 14200 m or 46580 feet. The model cloud is defined at each
grid point and level  and is not continuously distributed in the vertical.  The cloud base height is computed
from the lowest level of  cloud at each gridpoint in each ensemble member. But after averaging all the
ensemble members, the cloud  base may no longer correspond exactly to any of the model levels.

(11)Cloud top

      The height of the highest layer of cloud; but currently not available from the ensemblemodels.

(12)Convection

       Convection in the SREF is measured by the model convective cloud amount.  Convective cloud is created each
ensemble member from its convective scheme (BMJ, KF, or SAS).  This definition is not the standard used in
aviation, where convection is defined as the mass convergence ordivergence.  In the SREF product, the location of
convection is expressed in terms of convective cloudamount mean and spread.

(13)Convection Speed and direction

    Defined as convective storm's velocity field.  Mean (contour), spread (filled) and vector flow (direction) are
plotted in one picture.

(14)Low level wind shear (LLWS)

        LLWS here is defined by the vector change in the layer between the surface and 2000 feet level. Since winds are
defined at fixed pressure levels with 25mb intervals, the winds at different location may be at different pressure levels.
Furthermore, 2000feet level will not just at a pressure level. Therefore, to compute the LLWS, we must first find the
2000 feet level at each grid, then using 10m wind and the 2000 feet wind to get the wind change.

  First serch 2000 feet + 10m height (ie 2030 feet)  from the surface upward level by level, after find it, say between level-5 and level-6,
then compute the U and V components of the wind at this level by linear interpolation of level-5 wind and level-6 wind.
  Second step is computing the wind vector change over 10m and 2030 feet. The LLWS is defined as

           LLWS = SQRT [ (U₂₀₃₀ - U₁₀)² + (V₂₀₀₀ - V₁₀)² ]

   Where 2030 means 2030 feet, 10 means 10 meter.

The LLWS mean and spread thus can be obtained from 15 ensemble members.

    The severe LLWS is defined by (1) LLWS > 20 knotes over 2000 feet, or
                                                       (2) LLWS > 0.16 over any 200 feet layer withn the 2000 feet

  The probability of severe LLWS is also computed, but please note that, the pressure level intervals are

fixed 25 mb, or 600 feet, much larger than 200 feet, so, in current case, only case(1) is the threshold for severe LLWS here.

(15)Precipitation type

        Probabilities of three kinds of precipitation are defined: rain, snow and freezing rain are calculated.  

(16)Surface wind speed and direction at 10 m

        The mean of surface wind speed is computed from average  of total wind speeds Wi
      where
                Wi =  SQRT(Ui²+Vi²)   

      The products include :
               Wind speed mean, spread (wind direction is also depicted in the same image)
               The probability for wind speed exceeding 10 knotes
               The probability for wind speed exceeding 20 knotes
               The probability for wind speed exceeding 30 knotes

       The calculations of the mean and spread of wind direction(North is reference) are a little bit complex. First, compute
the mean and spread of U and V components respectively, then the mean and spread of wind direction can be expressed as:

          mean of wind direction = arctg ( Umean / Vmean)
          spread of wind direction = arctg (Wspread /Wmean)

 where
            Wspread = SQRT ( Uspread² + Vspread²)
            Wmean =  SQRT ( Umean² + Vmean²)

(17) Fog occurence probability

        Fog is detected at a grid point in an ensemble model if following cloud base and
cloud top threshold is satisfied:


            Cloud base < 10 m And Cloud top < 400 m

In general, the top of ground fog or radiation fog is less than 200m which is
mainly controlled by surface cooling and weak turbulence, while sea fog or advection
fog is wind-driven and can be thicker than radiation fog. So 400 m is used for the
cloud top threshold.