**6.9 Recommendations**

*The
hourly scalar mean wind speed and wind direction should be used in
steady-state Gaussian dispersion
models. These statistics should be processed using the methods provided in
Section 6.2.1; unit vector processing (Section 6.2.2) may also be used to
estimate the hourly scalar mean wind direction. The standard deviation of
the wind direction should be calculated using the techniques described in
Section 6.2.1. Hourly statistics may be obtained by processing samples over
an entire hour or by averaging sub-hourly statistics. The recommended
sub-hourly averaging interval for wind data processing is 15 minutes; two
valid 15-minute averages are required for a valid hourly average. *

*For
the purposes of this guidance, a calm occurs when the wind speed is below
the starting threshold of the anemometer or vane, whichever is greater.
Calms require special treatment in such applications to avoid division by
zero in the steady-state dispersion algorithm. For similar reasons, to avoid
unrealistically high concentration estimates at low wind speeds (below the
values used in validations of these models - about 1 m/s) EPA recommends
that wind speeds less than 1 m/s be reset to 1 m/s for use in steady-state
dispersion models; the unaltered data should be retained for use in
non-steady-state modeling applications. Calms should be identified in
processed data files by flagging the appropriate records; user’s guides
for the model being used should be consulted for model specific flagging
conventions.*

*Recommended
sampling and processing strategies for the primary meteorological variables
for various applications are given in Table 6-1. *

*The
Pasquill-Gifford (P-G) stability category should be determined with Turner's
method (Section 6.4.1) using
site-specific wind speed measurements at or near 10 m and representative***
***cloud cover and ceiling height. Other approved methods for estimating
the P-G stability category, for use when representative cloud cover and
ceiling observations are not available, include the solar radiation delta-T
(SRDT) method described in Section 6.4.2, and turbulence-based methods using
site-specific wind fluctuation statistics: *_{E} (Section 6.4.3) or_{A} (Section 6.4.4). Alternative methods for determining stability
category should be evaluated in consultation with the Regional Office.

*Emperical
relationships for use in models employing boundary layer scaling techniques should
be selected in accordance with a von Karmam constant of 0.4; recmmended
empirical relationships are given in reference ***[59]**.

*Missing
data should be flagged or replaced as appropriate depending on the model to
be used. Isolated one-hour gaps in
meteorological data bases used in regulatory modeling should be filled with
estimates bases on persistence or interpolation. Application specific
procedures should be used to fill longer gaps*

*If
the recommendations in this section cannot be achieved, then alternative
approaches should be developed in
consultation with the EPA Regional Office.*

**6. METEOROLOGICAL DATA PROCESSING**

*
6.1 Averaging and Sampling Strategies *

6.2 Wind Direction and Wind Speed

6.2.1 Scalar Computations

6.2.2 Vector Computations

6.2.3 Treatment of Calms

6.2.4 Turbulence

6.2.5 Wind Speed Profiles

** ** *6.3 Temperature *

6.3.1 Use in Plume-Rise Estimates

6.3.2 Vertical Temperature Gradient

** ** *6.4 Stability *

6.4.1 Turner's method

6.4.2 Solar radiation/delta-T (SRDT) method

6.4.3
_{E} method

6.4.4 _{A}method

6.4.5 Accuracy of stability category estimates

** 6.5 Mixing Height **

6.5.1 The Holzworth Method

** ** **6.6 Boundary Layer Parameters **

6.6.1 The Profile Method

6.6.2 The Energy Budget Method

6.6.3 Surface Roughness Length

6.6.4 Guidance for Measurements in the Surface Layer

** 6.7 Use of Airport Data **

**6.8 Treatment of Missing Data **

6.8.1 Substitution Procedures

* 6.9 Recommendations *