WindSat Environmental Data Record files from NAVOCEANO
(Click HERE
for WindSat Data
Products Users' Manual
from which this infromation was obtained)
The
Environmental Data Record (EDR) file
contains geolocated
wind vector retrievals. In addition to the selected wind vector, the
first four
ambiguities are provided (or as many as were
retrieved) along with the ranking
statistics. Other
retrieved parameters include sea surface temperature, integrated water
vapor,
integrated cloud liquid water, and rain rate. These parameters were
retrieved
in support of the wind vector retrieval.
The
file is composed of data records spaced
in time for every fourth 37 GHz VH
measurement along
scan and every scan along track. This results in approximately at
12.5-by-12.5
km grid
locally. In this dataset the WindSat
channels were averaged to a common footprint determined by the 6.8GHz
channel.
This results in a spatial resolution of approximately 50km along scan
for the
WindSat footprint. Only data points over the ocean
are included
in the EDR data records. The EDR
contains the SDR record number, which can
be used to
associate EDRs with SDRs.
Note that SDRs for all surface types are
reported.
All
WindSat files
are created under the UNIX operating
system. The EDR file is a binary file
written as a FORTRAN direct
access file, with FORTRAN 90 structure for each EDR
record. Note that in FORTRAN90 integers
are long (4 bytes) by default, unless noted otherwise.
Missing/Invalid
data: Missing or invalid
data, and the values of the spares, will be set to –9999. This
value will never
be found in valid data.
The
basic addressable unit is the 8-bit byte.
Multi-byte quantities are addressed by the most significant byte, and
hence
bytes are stored in order of decreasing significance. A byte is 8
contiguous
bits starting on an addressable byte boundary. The bits are numbered 0
through
7 starting from right to left.
Example:
Byte
n Byte
n+1 Byte
n+2 Byte
n+3
31
... 24 23 ... 16 15 ... 08 07 ... 00
The
twos complement system is used for
negative numbers. The twos complement representation of a negative
number is
formed by performing binary subtraction of each digit from 1 and then
adding 1
to the least significant digit. For example, the twos complement of
11010011 is
00101101.
The
file name contains a unique descriptor of
the data set. For example, consider the file
NPR.E068.WS.D10006.S1118.E1258
NPR
is the identifier used by NESDIS who
provides the files for NCEP
on their DDS server.
E068
indicates that channels are averaged to
a common footprint determined by the 6.8 GHz channel.
WS
identifies the data as WindSat.
D
field is the date (YYJJJ)
S
is the start time of the data in the raw
file (HHMM)
E
is the end time of the data in the raw file
(HHMM)
(Note
on start and end time: The start and end times of the data in the EDR differ from the times indicated by the file
name. This
is the result of the averaging, resampling
and
filtering within the processing code. The difference is usually less
than one
minute.)
Each
data record in the EDR
file has the following variables and attributes:
VARIABLE |
DESCRIPTION |
TYPE |
JD2000 |
Seconds since 1200Z, 1 Jan 2000; FillValue: 0.0 |
8-byte REAL |
Latitude |
Degrees latitude |
4-byte REAL |
longitude |
Degrees longitude |
4-byte REAL |
Scan_Angle |
Angle between look and flight
direction; units: Radians |
4-byte REAL |
EIA |
Earth incidence angle; units: Radians NoValue: 0.0 |
4-byte REAL |
CAA |
Compass azimuth angle; units: Radians |
4-byte REAL |
Scan_Number |
Scan number of record |
4-byte INTEGER |
Downcount_Number |
Downcount number of record |
2-byte INTEGER |
SurfaceType |
Surface Type (see below for list) |
2-byte INTEGER |
SDR_QC_Flag |
SDR Quality Control Flags (see below); FillValue = 0 |
4-byte INTEGER |
SDR_Record_Number |
SDR record number corresponding to
this EDR record number |
4-byte INTEGER |
sstErr |
Square root of the estimated retrieval error
covariance for the 1st ranked ambiguity; NoValue = 255; scale_factor = 0.05 |
1-byte INTEGER |
wspdErr |
Square root of the estimated retrieval error
covariance for the 1st ranked ambiguity; NoValue = 255 scale_factor = 0.05 |
1-byte INTEGER |
vaporErr |
Square root of the estimated retrieval error
covariance for the 1st ranked ambiguity; NoValue = 255; scale_factor = 0.05 |
1-byte INTEGER |
cloudErr |
Square root of the estimated retrieval error
covariance for the 1st ranked ambiguity; NoValue = 255; scale_factor = 0.002 |
1-byte INTEGER |
SST |
Sea surface temperature in K |
4-byte REAL |
Water_Vapor |
Integrated columnar water vapor
reported in mm |
4-byte REAL |
Cloud_Liquid_Water |
Integrated columnar cloud liquid
water in mm |
4-byte REAL |
Number_of_Ambiguities |
Number of wind vector ambiguities
retrieved |
2-byte INTEGER |
Selected_Ambiguity |
Ambiguity selected by the
ambiguity selection algorithm |
2-byte INTEGER |
Wind_Speed (1) |
Wind speed for 1st
ranked ambiguity, at 10m height, in m/s |
4-byte REAL |
Wind_Speed (2) |
Wind speed for 2nd
ranked ambiguity, at 10m height, in m/s |
4-byte REAL |
Wind_Speed (3) |
Wind speed for 3rd
ranked ambiguity, at 10m height, in m/s |
4-byte REAL |
Wind_Speed (4) |
Wind speed for 4th
ranked ambiguity, at 10m height, in m/s |
4-byte REAL |
Wind_direction (1) |
Wind direction for 1st
ranked ambiguity, at 10m height, in oceanographic degrees |
4-byte REAL |
Wind_direction (2) |
Wind direction for 2nd
ranked ambiguity, at 10m height, in oceanographic degrees |
4-byte REAL |
Wind_direction (3) |
Wind direction for 3rd
ranked ambiguity, at 10m height, in oceanographic degrees |
4-byte REAL |
Wind_direction (4) |
Wind direction for 4th
ranked ambiguity, at 10m height, in oceanographic degrees |
4-byte REAL |
Chi_Squared (1) |
Chi-squared probability for 1st
ranked ambiguity |
4-byte REAL |
Chi_Squared (2) |
Chi-squared probability for 2nd ranked ambiguity |
4-byte REAL |
Chi_Squared (3) |
Chi-squared probability for 3rd
ranked ambiguity |
4-byte REAL |
Chi_Squared (4) |
Chi-squared probability for 4th
ranked ambiguity |
4-byte REAL |
Model_Wind_Speed |
NOGAPS (or, prior to 2006, GFS) model wind speed, in m/s |
4-byte REAL |
Model_Wind_Direction |
NOGAPS (or, prior to 2006, GFS) model wind direction,
in oceanographic degrees |
4-byte REAL |
EDR_QC_Flag1 |
EDR Quality Control Flags 1 (see
below); FillValue = 0 |
4-byte INTEGER |
EDR_QC_Flag2 |
EDR Quality Control Flags 2 (see
below); FillValue = 0 |
4-byte INTEGER |
Rain_Rate |
14 km resolution rain rate in mm/hr |
4-byte REAL |
phiErr (1) |
Square root of the estimated retrieval error
covariance 1st ranked ambiguity; NoValue = 255; scale_factor = 0.2 |
1-byte INTEGER |
phiErr (2) |
Square root of the estimated retrieval error
covariance 2nd ranked ambiguity; NoValue = 255; scale_factor = 0.2 |
1-byte INTEGER |
phiErr (3) |
Square root of the estimated retrieval error
covariance 3rd ranked ambiguity; NoValue = 255; scale_factor = 0.2 |
1-byte INTEGER |
phiErr (4) |
Square root of the estimated retrieval error
covariance 4th ranked ambiguity; NoValue = 255; scale_factor = 0.2 |
1-byte INTEGER |
Explanation
of variables:
JD2000 is the time of the measurement in
Julian Day
2000 format, the number of seconds since January 1, 2000 noon UTC. JD2000 can be converted to ‘seconds
past midnight’ by SPM = (JD2000 +
43200) modulus 86400
Latitude and Longitude
is the location of the pierce point projected onto the earth (degrees).
Range
of the longitude is ± 180°.
Scan_Angle
is the angular scan position (radians) of the data, where Scan_Angle
equal to –90° is referenced to the X-axis of WindSat
(0° is approximately along the spacecraft velocity vector; 90°
is under the
warm load). Units are radians.
EIA
contains the earth incidence angles of the 37 GHz channels. Units are
radians.
CAA is
the compass azimuth angle at the pierce point (radian). This represents
the WindSat look-direction relative to the
local meridian.
North equals 0°, and the CAA
increases clockwise.
Scan_Number
is the number of BAPTA spins since the
start of the
data file. This number is incremented when a sync mark is crossed. (The
BAPTA is the scan drive mechanism on WindSat.)
Downcount_Number
is a short integer containing the 37 GHz VH
sample
number associated with this measurement for this scan. Downcount
numbers count downward and reset once per scan. The fore swath data
covers a downcount range of 800 to 1116.
For the EDRs,
the downcounts decrement by four, because
the pixel locations
correspond to every fourth 37 GHz sample.
SurfaceType
is an integer value where:
0
= LAND
1
= NOT USED
2
= NEAR COAST
3
= ICE
4
= POSSIBLE ICE
5
= OCEAN
6
= COAST
7
= SPARE
SDR_QC_Flag
is a bit pattern indicating errors in the processing or quality control
flags.
This flag is set in the SDRP and passed
through the EDRP for informational
purposes. The table below defines
the flag.
SDR QC Flag contents:
Bit Number |
Contents |
0-7 |
Reserved |
8 |
Forward/aft scan (bit set to 1 for
forward part of scan, 0 for Aft scan) |
9 |
Ascending/descending orbit flag (1
for ascending, 0 for descending) |
10 |
Reserved |
11 |
Gains Applied (1 = Gains applied
to data, 0 = gains could not be applied) |
12 |
Glare angle (sun glint) invalid
because no sun vector of LOS doesn't pierce earth if set to 1 |
13-18 |
Glare Angle (glare angle foes 0 to
30 for 0 to 60 degrees, 31 is > 60 deg, 32 is invalid) |
19-23 |
Cold Load Flag (1 if RFI or moon contamination present in scan). This
flag indicates that the cold sky calibration data was contaminated with
RFI or lunar intrusion.
There is one bit for each frequency band (6.8: bit 19;
10.7: bit 20;. . . 37.0: bit 23). A
mitigation algorithm is used to modify the cold load temperature used
for calibration. |
24-28 |
Frequency specic
warm load flag indicating presence of thermal gradients. There is one
bit for each frequency band (6.8: bit 24; 10.7: bit 25;. . . 37.0: bit 28). A mitigation algorithm is
used to modify the warm load temperature used for calibration. |
29 |
Attitide Transient; the satellite attitude
is changing too rapidly for accurate resampling. |
30-31 |
Reserved |
SDR_Record_Number is the number of the corresponding
SDR record for a given EDR
record. Having this value enables easy
comparison
with SDRs or SDR-based
match-
up data sets.
sstErr,
wspdErr, vaporErr, cloudErr are the
estimated retrieval error covariances for
the first rank
ambiguity for SST, wind speed, water vapor and cloud liquid water
respectively.
They are as unsigned one-byte integers.is
a real
field reserved for later use. A value of 255 is used to denote that the
retrieval error covariance is invalid or not meaningfull
(such as when a retrieval is not
performed). Multiply
by sstErr by 0.05 to convert to K.
Multiply wspdErr by 0.05 to convert to
m/s. Multiply
vaporErr by 0.05 to convert to mm. Multiply
cloudErr by 0.002 to convert to mm.
SST is the retrieved sea surface
temperature in
K.
Water_Vapor is
the retrieved integrated columnar water vapor reported in mm.
Cloud_Liquid_Water is the retrieved integrated
columnar cloud liquid water in mm.
Number_of_ambiguities is a short integer containing the
number of wind direction
ambiguities retrieved. When set to 0, this indicates that no
directional
ambiguities were retrieved.
Selected_ambiguity is a short integer containing a
pointer to the ambiguity
selected by the ambiguity selection algorithm. It contains values 0,1,2 and 3 where 0 value points to first ranked
ambiguity, 1 points to 2nd
ranked etc. This
variable should be used together with number of ambiguities.
Wind_Speed
is a four element real array containing up to four wind speeds
corresponding to
the wind direction ambiguities. The wind speeds are listed in order of
the wind
direction ambiguity
ranking. If fewer than four ambiguities
are retrieved,
the remaining fields are set to -9999. For example, when the number of
ambiguities is 2, the 3rd and 4th elements of
this array
are set to -9999. Units are m/s and all wind data are referenced to
10-m
height.
Wind_Direction
is a four element array containing up to four wind direction
ambiguities. The wind
directions are place in the array in order of rank. If fewer than four
ambiguities are retrieved, the remaining fields are set to 0. For
example, when
the number of ambiguities is 2, the 3rd and 4th
elements
of this array are set to 0. Units are degrees with 0° representing
north and
increasing clockwise. The wind direction convention is oceanographic,
i.e.,
wind direction of 90° is blowing toward the east. As with wind
speed, the
direction is referenced to 10- m height.
Chi_Squared
is a four element array containing the chi-squared, or similar metric,
of the
wind vector retrieval, which is used for ranking the ambiguities. These
values
can be used to calculate chi squared probabilities using 14 degrees of
freedom
when the 6.8 GHz measurements are available and 46.9 otherwise. The
ratio of
the chi squared probabilities for the different
ambiguities provides an estimate of the
relative
probabilities for each ambiguity. The chi squared probabilities are not
accurate measures of the relative probabilities for different
retrievals since
the appropriateness of the estimated measurement error covariance
matrix and
the accuracy of the forward model, which are
used to
compute the chi squared, varies significantly with wind speed and
atmospheric
conditions.
Model_Wind_Speed
is the wind speed that corresponds to the wind direction used for
initialization
of the ambiguity selection routine.
Model_Wind_Direction is the wind direction used for
initialization of the ambiguity
selection routine.
EDR_QC_Flag1 is a 32 bit long integer used to
describe
the “quality” of a number of radiometric and radiometrically
determined geophysical parameters, as well as indicating the conditions
under which
the retrieval was performed. The quality is indicated by the setting of
specific bits based on the range in which these various parameters
fall. As a
general rule, a bit set to “1” means “bad”,
while a bit set to “0” means
“good”. EDR_QC_Flag1 is described in more detail at the end
of this document.
EDR_QC_Flag2 is a spare retrieval quality
flag. It is currently
reserved for use in retrieval algorithm development.
Rain_Rate
is the retrieved rain rate over the ocean in mm/hour at a horizontal
spatial
resolution of 14 km.
phiErr is a four-element array
containing the
estimated wind direction retrieval error covariance for each ambiguity
stored
as an unsigned one-byte integer. A value of 255 is used to denote that
the
retrieval error covariance is invalid or not meaningful (such as when a retrieval is not performed). Multiply by 0.2 to
convert to
degrees.
EDR_QC_Flag1
description (NRL-2):
This
EDR QC Flag 1
description provided here applies to the NRL-2 data release, which used
WindSat GDPS version 1.9.0. EDR
QC Flag 1 definitions are consistent with those from the previous data
release.
EDR_QC_Flag1
is populated as described in the
table below. The flags that are most likely to indicate conditions that
will
degrade retrievals are: EDR rain flag, ice
flag, RFI flag, land
contamination, inland lakes and seas and
sheltered bodies
of water and beam averaging threshold (bits 3, 5, 6, 7, 9 and 19).
These flags
are grouped together in the “Low confidence flag”, bit
1, which along with bit 0 can be used to screen retrievals for most
uses. Additionally
the “Low confidence flag” is set when the Chi_Squared value for the first
rank retrieval exceeds a threshold value. This threshold is currently
set to
48.1 when the 6.8 GHz measurements are available and 46.9 otherwise
(these
values are chosen based on chi squared probabilities).
There
are also low confidence bits for each
of the individual EDR's which will be set
when conditions
may reduce the accuracy of the retrieva l
for that EDR. An example would be high
water vapor may reduce wind
direction accuracy even if the water vapor retrieval accuracy is
acceptable.
The criteria for setting the low confidence bits for the individual EDR's may vary with the retrieval algorithm and
is likely
to change as the retrieval algorithms and brightness temperatures are
improved.
The
contents of the flag are given in the
table at the end of this document. Additional information for some of
the flags
is given below.
EDR Rain Flag: Observations where the retrieved
integrated
cloud liquid water is greater than 0.2
mm are
flagged for rain.
SDR Rain Flag: Rain is considered to be present
by the
condition
37V -
0.979*37H < 55.0 OR
1.175*18V –
30.0 > 37V OR
18H >
170.0 OR
37H > 210
This rain flag was
developed for Stokes vector analysis and model development. It was
based on
similar flag used for SSM/I cal/val.
The flag was tested against the SSM/I rain
flag. Because this flag was developed
for model development,
it is conservative, i.e., sensitive, to ensure that as little rain as
possible
was in the training data sets. As such, the algorithm flags not only
rain, but
also flags some regions of high cloud liquid water or high columnar
water
vapor.
Geographic Boundaries Flag: Geolocation
places SDR surface point in one of several
geographic
regions that might be unsuitable for surface wind (or other edr)
retrievals due to ice, salinity variations, etc. These include Lakes
and
surrounding lakes, 4.)
Also included are
regions where Salinity is far from the norm and/or surrounded by land:
1.) Top
of Baja (above
Warm Load Anomaly: Not used at this time.
Land Contamination: Prior to performing retrievals,
individual WindSat pixels are averaged
together to reduce the
measurement noise and to match antenna beams at all
frequencies. This flag indicates regions
where the
pixels with significant land contamination have been averaged in with
ocean
pixels. This results in erroneous brightness temperatures.
RFI: This flag indicates when RFI is
likely to affect the 10.7 GHz brightness temperature measurements. The
primary
known source is the downlink from some TV broadcast satellites which
reflects
off the ocean surface. The regions flagged are around Europe in the
descending
pass and off the eastern coast of
Satellite Attitude Transient: There are anomalies in the WindSat data caused by rapid changes in the
satellite
attitude. These transients cause poor spatial matching of the WindSat channels because static along-scan and
along-track
increments are used for beam resampling.
About 0.5
percent of the data is affected.
EDRP QC Flag 1 contents (NRL-1):
Bit Number |
Contents |
Values |
0 |
Retrieval Status
|
0=OK; 1=Retrieval not performed or
retrieval failure for all EDR's |
1 |
Low confidence |
|
2 |
Reserved |
|
3 |
No 6.8 GHz flag |
6.8 GHz channels not available or
not used in the retrieval; SST retrievals are less
accurate particularly for SST < 15
degrees C |
4 |
EDR Rain flag |
Rain flag generated by the EDRP (currently based on retrieved cloud liquid water) |
5 |
SDR Rain flag |
Rain flag based on TBs (or from SDRP) |
6 |
Ice flag |
|
7 |
Land Contamination |
Threshold exceeded for land
contamination |
8 |
reserved |
|
9 |
Inland lakes and sheltered bodies of water |
Great Lakes, Lake Victoria, Red
Sea, etc. |
10 |
Salinity out-of-bounds or unknown |
|
11 |
reserved |
|
12 |
10 GHz RFI
flag |
|
13 |
Sun glint |
Threshold exceeded for sun glint
angle |
14 |
Satellite Attitude Transient |
|
15 |
Cold load anomaly flag |
Cold load correction applied to
one or more channels |
16 |
Warm load anomaly flag |
Ignore. |
17-18 |
Faraday Rotation Correction |
00=No correction, 01=SEC Based 10=Geoloc based, 11=reserved |
19 |
Beam Averaging Threshold |
Insufficient data was available
for reliable beam averaging to be performed |
20 |
Wind speed too low for accurate wind direction retrieval |
Wind speed is < 5 m/s |
21 |
Wind speed too high for accurate retrieval |
Wind speed is > 25 m/s |
22 |
Wind speed quality |
Low confidence for wind speed
retrieval |
23 |
Wind speed retrieval status |
No retrieval for wind speed |
24 |
Wind direction quality |
Low confidence for wind direction
retrieval |
25 |
Wind direction retrieval status |
No retrieval for wind direction |
26 |
SST quality |
Low confidence for SST retrieval |
27 |
SST retrieval status |
No retrieval for SST |
28 |
Water vapor quality |
Low confidence for water vapor
retrieval |
29 |
Water vapor retrieval status |
No retrieval for water vapor |
30 |
Cloud liquid water quality |
Low confidence for cloud liquid
water retrieval |
31 |
Cloud liquid water retrieval status |
No retrieval for cloud liquid water |