Last Updated: 09 June 2009
Quick Links
Abstract
The 4 km AVHRR Pathfinder Version 5 SST Project (Pathfinder V5) is a new reanalysis
of the AVHRR data stream developed by the University of Miami's Rosenstiel
School of Marine and Atmospheric Science (RSMAS) and the NOAA National
Oceanographic Data Center (NODC). In partnership with NODC and RSMAS is NASA's
Physical Oceanography Distributed Active Archive Center (PO.DAAC), which has
years of experience serving and developing earlier versions of the Pathfinder
dataset. Currently in the third year of a three-year demonstration effort,
it is hoped that this system can be implemented as an ongoing effort as part of
a broader SST climate data record system. This reprocessing uses an improved
version of the
Pathfinder algorithm and processing steps to produce twice-daily global SST
and related parameters back to 1981, at a resolution of approximately 4 km,
the highest possible for a global AVHRR data set. Temporal averages for
5-day, 7-day, 8-day, Monthly, and Yearly periods are also produced. Current
key improvements over the original 9 km Pathfinder SST data set include a
more accurate, consistent land mask, higher spatial resolution, and
inclusion of sea ice information. Additional improvements including better
flagging of aerosol-contaminated retrievals and the provision of
wind and aerosol ancillary data will be implemented in a future Version 6
reprocessing. See
http://www.nodc.noaa.gov/SatelliteData/pathfinder4km for more information.
Background
The NOAA/NASA AVHRR Oceans Pathfinder Program developed global SSTs at
9.28 km resolution beginning in the early 1990's and ended with the
releases of the
Version 4.x (V4) dataset. The V4 dataset, distributed through the JPL's
PO.DAAC (http://podaac.jpl.nasa.gov/sst/)
,
found widespread use and was among their most popular products. For
more information regarding the V4 Pathfinder data, see:
Kilpatrick, K.A., G.P. Podesta and R. Evans (2001). Overview of the NOAA/NASA
Advanced Very High Resolution Radiometer Pathfinder algorithm for sea surface
temperature and associated matchup database, Journal of Geophysical
Research-Oceans, 106 (C5): 9179-9197 MAY 15 2001.
Despite its success, some problems with the V4 data set limited its use. For
example, biases are known to exist in areas strongly impacted by atmospheric
aerosols. Additionally, the relatively crude landmask created significant
problems for coastal applications and lack of sea ice information limits the
use of the data in the ice-prone high latitudes. This new reprocessing by
RSMAS/NODC seeks to correct these deficiencies and provide a higher
resolution and ultimately more accurate SST product by the end of the
three-year effort. This global 4 km resolution dataset is termed "Pathfinder
Version 5 (V5)" since it builds directly on the V4 dataset, including the
use of the same set of matchup coefficients.
Satellites Used
The following table describes which NOAA polar orbiting satellites were used
in the Pathfinder Version 5 processing for SST. The table gives dates in
both YYYYDDD format (4 digit year followed by 3 digit day of year) and
YYMMDD format (2 digit year, 2 digit month, 2 digit day of month).
Satellite YYYYDDD YYMMDD ORBIT
--------- ------- ------ -------------
NOAA-7 1981236 - 1985003 810824 - 850103 "afternoon"
NOAA-9 1985004 - 1988312 850104 - 881107 "afternoon"
NOAA-11 1988313 - 1994256 881108 - 940913 "afternoon"
NOAA-9 1994257 - 1995021 940914 - 950121 "afternoon"
NOAA-14 1995022 - 2000285 950122 - 001011 "afternoon"
NOAA-16 2000286 - 2002365 001012 - 021231 "afternoon"
NOAA-17 2003001 - 2005155 030101 - 050604 "morning"
NOAA-18 2005156 - 2006365 050605 - 061231 "afternoon"
Note that there were no NOAA-7 data (1981-early 1985) in V4 Pathfinder,
and that there are
differences between the satellites chosen during 2000 and beyond between the
old V4 Pathfinder and the new V5 data. Changes were required due to the
detection of a digitizer error in NOAA-14 data and degradation of NOAA-16
performance. Please see the following
article for complete details on NOAA-14 digitizer problem:
Podesta, G.P., M. Arbelo, R. Evans, K. Kilpatrick, V. Halliwell, and J. Brown
(2003). Errors in high-latitude SSTs and other geophysical products linked to
NOAA-14 AVHRR channel 4 problems, Geophysical Research Letters , vol
30, no 11., 1548, doi: 10.1029/2003GL017178.
Up to date information on the status of the NOAA polar orbiting satellites can
be found at the NOAA Office of Satellite Operations
POES Status
web site and
from the
Weekly NOAA Polar
Orbiting Spacecraft Status Reports.
Times of Observation
A common question about the Pathfinder SSTs concerns the time of observation
for each pixel. This question is a complicated one because there are
several factors to consider, some having to do with the satellite orbits
and some having to do with the manner in which the data are processed.
Each satellite in the NOAA polar orbiting satellite series was launched
in a specific orbit, with specified ascending and descending node times
(the ascending node time is the time when the satellite passes from south
to north over the equator and the descending node time is when it passes from
north to south over the equator). Each orbit requires about 102 minutes to
complete, and the swaths observed by the satellite overlap, resulting in
possibly multiple observations from multiple orbits at any given spot on
Earth. This overlap is more pronounced at higher latitudes. Since the
Pathfinder algorithm combines the multiple observations (as long as they
have they same quality level), the actual time of observation of any given
pixel is not known precisely. While an average, or median, time could be
determined during the processing, it would require some significant changes
to the processing code making it difficult to carry out under the
current funding situation. To complicate matters, the actual orbits of the
satellites drift over time, introducing a temporal component as well. The
table below (assembled from information in
Table 1.2-1 of the NOAA Polar Orbiter Data User's Guide and
Weekly NOAA Polar
Orbiting Spacecraft Status Reports)
shows the Local Solar Times (LST) at launch and the values they had drifted
to by later dates for the satellites used in the Version 5 Pathfinder
algorithm. Drift rates are also given if known.
Satellite Ascending and Descending Node Times (LST) Avg Drift Rate
------------------------------------------------------------------------
NOAA-7 Ascending: 1430 (Jun 1981) to ? (Not Available) ? min/month
Descending: 0230 (Jun 1981) to ? (Not Available)
------------------------------------------------------------------------
NOAA-9 Ascending: 1420 (Jan 1985) to 2116 (Mar 1995) +3.4 min/month
Descending: 0220 (Jan 1985) to 0916 (Mar 1995)
------------------------------------------------------------------------
NOAA-11 Ascending: 1330 (Sep 1988) to 1723 (Mar 1995) +3.0 min/month
Descending: 0130 (Sep 1988) to 0523 (Mar 1995)
------------------------------------------------------------------------
NOAA-14 Ascending: 1330 (Dec 1994) to 2155 (May 2007) +3.4 min/month
Descending: 0130 (Dec 1994) to 0955 (May 2007)
------------------------------------------------------------------------
NOAA-16 Ascending: 1400 (Sep 2000) to 1657 (Jul 2008) +1.9 min/month
Descending: 0200 (Sep 2000) to 0457 (Jul 2008)
------------------------------------------------------------------------
NOAA-17 Ascending: 2200 (Jun 2002) to 2149 (Jul 2008) -0.1 min/month
Descending: 1000 (Jun 2002) to 0949 (Jul 2008)
------------------------------------------------------------------------
NOAA-18 Ascending: 1400 (May 2005) to 1338 (Jul 2008) -0.6 min/month
Descending: 0200 (May 2005) to 0138 (Jul 2008)
Parameters Produced
In addition to the SST values, the Pathfinder V5 Project is making several
other useful parameters available. Among other uses, these parameters can
be used to calculate higher-order statistics, examine in detail the quality
tests passed and failed, and to individually determine what consitutes a
"good" SST value by selecting SSTs based on their overall quality
assignment. The seven parameters currently available are listed below.
- "All-pixel" SST
- The all-pixel SST files contain values for each pixel location,
including those contaminated with clouds or other sources of
error. The Overall Quality Flag values may be used to filter
out these unwanted values. The SST value in each pixel location
is an average of the highest quality AVHRR Global Area Coverage
(GAC) observations available in each roughly 4 km bin.
- First-guess SST
- The Pathfinder algorithm uses a first guess SST based on the
Reynolds Optimally Interpolated SST (OISST), Version 2 product.
The OISST V2 is also used in the quality control procedures.
- Number of Observations
- This parameter indicates the number of AVHRR GAC observations
falling in each approximately 4 km bin.
- Standard Deviation
- This is the standard deviation of the observations in each 4 km
bin.
- Overall Quality Flag
- The overall quality flag is a relative assignment of SST quality
based on a hierarchical suite of tests. The Quality Flag varies
from 0 to 7, with 0 being the lowest quality and 7 the highest.
For most applications, using SST observations with quality levels
of 4 to 7 is typical. For applications requiring only the
best-available observations (at the expense of the number of
observations), use quality levels of 7 only.
For more information regarding the suite of tests, see the
Kilpatrick et al. (2001) paper cited above.
- Mask 1
- These files contains a mask code, which along with Mask 2, can be
to determine the tests in the hierarchical suite that were
passed or failed, resulting in the Overall Quality Flag.
- Mask 2
- These files contains a mask code, which along with Mask 1, can be
to determine the tests in the hierarchical suite that were
passed or failed, resulting in the Overall Quality Flag.
Additionally, in the Version 6.0 data planned for future production, wind
speeds and aerosol content will also be provided.
Data Format
The data are all stored in Hierarchical Data Format Verison 4 (HDF 4),
using the Scientific Data Set (SDS) model. HDF-SDS is a self-describing
format capable of storing multiple layers of data as well as metadata
describing the file contents. The metadata contained within the HDF files
do not conform to FGDC, COARDS, or other standards. However, complete FGDC
metadata records are available from NODC for all of the formal Pathfinder
data and climatologies (complete FGDC metadata is not currently planned
for any of the interim data).
HDF's capacity to chunk, or tile, the data is taken advantage of, with
512 by 512 element tiles. This internal compression scheme enables
both efficient access to subsets of the data as well as enormous
savings in disk volume required compared to uncompressed requirements.
Each of the seven parameter files listed above contains a mapped array with
8192 elements in longitude and 4096 in latitude, plus a vector of length
8192 identifying the longitudes and a vector with 4096 values indicating
the latitudes. There are also global tags describing the entire contents
as well as tags describing each of the 2 vectors and 1 array. The seven
parameters are stored either as 8-bit or 16-bit unsigned integers which
may be converted linearly (y = mx + b) to geophysical units using a scale
(i.e., slope=m) and offset (i.e., intercept=b) according to the following
table:
PARAMETER # BITS SCALE OFFSET UNITS
------------------------------------------------------------
"All-pixel" SST 16 bit 0.075 -3.0 Deg C
First-guess SST 16 bit 0.075 -3.0 Deg C
Standard Deviation 16 bit 0.150 0.0 Deg C
Number of Observations 8 bit 1.000 0.0 Unitless
Overall Quality Flag 8 bit 1.000 0.0* Unitless
Mask 1 8 bit 1.000 0.0 Unitless
Mask 2 8 bit 1.000 0.0 Unitless
* - Note that the offset parameter in the Overall Quality Flag files is
incorrectly set to a value of 1. The value of 0 listed here is correct.
Information for GIS Users
While not directly readable in most GIS systems, mechanisms are being
developed to dynamically convert HDF files into suitable formats.
Upon request, NODC will convert specific files into GeoTIFF format (see
http://remotesensing.org/geotiff/geotiff.html)* which is
capable of being of being read by most GIS software.
Map Projection Information
The following combination of parameters
works with software, such as GIS systems, needing this level of detail (the
four digit codes are from the GeoTIFF web site
http://remotesensing.org/geotiff/geotiff.html)*. These
pieces of information comprise the related tags in a GeoTIFF file:
Geotiff_Information:
Version: 1
Key_Revision: 1.0
Tagged_Information:
ModelTiepointTag (2,3):
0 0 0
-180 90 0
ModelPixelScaleTag (1,3):
0.0439453 0.0439453 0
End_Of_Tags.
Keyed_Information:
GTModelTypeGeoKey (Short,1): ModelTypeGeographic
GTRasterTypeGeoKey (Short,1): RasterPixelIsArea
GTCitationGeoKey (Ascii,17): "LONG/LAT E005"
GeographicTypeGeoKey (Short,1): GCS_WGS_84
GeogAngularUnitsGeoKey (Short,1): Angular_Degree
ProjLinearUnitsGeoKey (Short,1): Linear_Meter
End_Of_Keys.
End_Of_Geotiff.
GCS: 4326/WGS 84
Datum: 6326/World Geodetic System 1984
Ellipsoid: 7030/WGS 84 (6378137.00,6356752.31)
Prime Meridian: 8901/Greenwich (0.000000/ 0d 0' 0.00"E)
Projection Linear Units: 9001/metre (1.000000m)
Corner Coordinates:
Upper Left (180d 0' 0.00"W, 90d 0' 0.00"N)
Lower Left (180d 0' 0.00"W, 90d 0' 0.00"S)
Upper Right (180d 0' 0.00"E, 90d 0' 0.00"N)
Lower Right (180d 0' 0.00"E, 90d 0' 0.00"S)
Center ( 0d 0' 0.00"E, 0d 0' 0.00"N)
Other systems, such as MATLAB, can read the HDF-SDS format directly
and require less detailed information to re-project the data. If using
the MATLAB Mapping Toolbox, for example, use a map legend of:
ml = [8192/360 90 -180];
Key Differences Between V5 Pathfinder and V4 Pathfinder
In addition to the provision of the seven parameters listed above (V4 only
provides SST, number of observations, and quality flag), there are three
key enhancements which distinguish the new V5 data from the older V4 data:
improved resolution, ice mask identification, and improved land mask. These
areas are discussed below.
- Improved Resolution
-
While High-Resolution Picture Transmission (HRPT) or Local Area
Coverage (LAC) data exist for some parts of the world at roughly
1.1 km resolution, the GAC data at roughly 4.6 km are the highest
resolution AVHRR data available globally. When mapped to an equal
angle grid, this resolution results in an array of 8196 by 4096
pixels. This improved spatial resolution greatly enhances the
utility of the Pathfinder V5 dataset in coastal zones and areas
of strong SST gradients.
- Ice Mask Identification
-
A tiered approach to flagging pixels compromised by sea-ice is being
taken. First, weekly ice-masks derived from the Special Sensor
Microwave Imager (SSMI) and processed by Remote Sensing Systems, Inc.
(RSS), are used to identify regions with ice concentrations exceeding
1% (in other words, if any ice is present the region is flagged as
ice covered). Currently, RSS version 5 data are being used at 25 km
resolution. These data are available back to July 1987. Prior to that
time, the sea ice information from the Reynolds OISST V2 is used.
Because SSMI sensor data are not valid near land, ice concentrations
in coastal regions are not properly identified. These gaps are
filled using the Reynolds OISST V2 data at 1-degree resolution.
Finally, because some areas appear inconsistent (such as a gap right
at the poles in the SSMI data), an automated approach to "backfilling"
the ice mask into interior regions has been developed. Therefore, the
first tier uses the SSMI data, the second tier uses the OISST V2 data,
and the third tier fills gaps using an automated algorithm. This ice
mask is then used to exclude those pixels from the calculation of the
three week running mean internal reference check. Eventually, in
Version 6 of the Pathfinder reprocessing efforts, ice masks will be
used to flag the data as poor quality. The current Version 5
approach is intended as an intermediate step to reduce the likelihood
of misclassified pixels (that is, SST pixels with high quality levels
but falling on ice mask) and develop an understanding of the impact of
the ice masking techniques on the dataset without requiring major
changes to the processing system.
- Improved Land Mask
-
A new land mask developed from MODIS data is employed in the V5
Pathfinder Project as well. This mask represents an enormous
improvement over the old 9 km land mask used in Pathfinder V4 and
earlier data. The new mask is based on a 1 km resolution dataset
derived by the USGS Land Processes Distributed Active Archive Center
(see
http://edcdaac.usgs.gov/modis/mod12q1.html for more info.) The
classifications in that USGS dataset are as follows:
SHALLOW_WATER 0
LAND 1
SHORELINE 2
SHALLOW_INLAND_WATER 3
EPHEMERAL_WATER 4
DEEP_INLAND_WATER 5
MODERATE_OCEAN 6
DEEP_OCEAN 7
The USGS dataset is at 1 km resolution so must be degraded to 4 km
before use in the Pathfinder V5 processing. A liberal approach is
taken so if any of the 16 USGS 1 km values within a given 4 km
bin are identified as water, then the entire 4 km bin is classified
as water in Pathfinder V5. Any isolated water pixels completed
surrounded by land are then set to land to avoid extremely small inland
water bodies which are likely to suffer land-contamination.
Values of 1, 2, and 4 from the USGS dataset are selected to
represent land; the remaining values (0, 3, 5-7) are classified as
water. The result is a global array (8192 x 4096 pixels) of
land/water classifications. This file, in which land points have a
value of 2, and water a value of 0, may be accessed by following
this link:
pfv50_land.m04.hdf. In the Pathfinder V5 processing, SSTs and
quality levels are not calculated for pixels falling on land. These
areas are assigned an SST pixel value of 0 (SST=-3.0) and quality
flag of 0.
-
File Naming Conventions
The file names in the Pathfinder V5 dataset contain a great deal of
information. Therefore, they tend to be rather long and somewhat
complicated at first examination. Each set of temporally-averaged data
uses the following convention:
OBSDATE.[BITCODE][RESO][AVGPERIOD][DAYNIGHT][VERSION]-TYPE(-BITS).hdf
Where:
- OBSDATE = The date(s) the observations were made. OBSDATE may
be in one of the following forms:
- Daily Files: YYYYDDD, with:
- YYYY = 4 digit year of observation
(1985-200x)
- DDD = 3 digit day of year (001-365, or
001-366 for leap years) For help converting
between calendar dates and day of year,
please see our Calendar
Date Conversion Chart
- 5-Day Files: SYYYYDDD-EYYYYDDD, with:
- SYYYYDDD = starting 4-digit year and
3-digit day
- EYYYYDDD = ending 4-digit year and 3-digit
day
- 7-Day Files: YYYYWW, with:
- YYYY = 4 digit year of observation
- WW = 2-digit week number between 01 and 52
- 8-Day Files: SYYYYDDD-EYYYYDDD, with:
- SYYYYDDD = starting 4-digit year and
3-digit day
- EYYYYDDD = ending 4-digit year and 3-digit
day
- Monthly Files: YYYYMM, with:
- YYYY = 4 digit year of observation
- MM = 2-digit month between 01 and 12
- Yearly Files: YYYY, with:
- YYYY = 4 digit year of observation
- BITCODE = Indicates the bit length of the pixel values in the file.
"s" is for 16 bit files, "m" is for 8 bit files (see BITS
below)
- RESO = Approximate resolution in km. Set to "04" for 4 km
files
- AVGPERIOD = Indicates the averaging period used to create the
file. AVGPERIOD may be in one of the following forms:
- Daily Files = "d"
- 5-Day Files = "5"
- 7-Day Files = "w"
- 8-Day Files = "8"
- Monthly Files = "m"
- Yearly Files = "y"
- DAYNIGHT = Indicates nighttime (1 or 4) or daytime (2 or 3)
observations. Within Pathfinder, all satellite used
prior to NOAA-17 are "afternoon" satellites, which means their daytime
observations fall around 2 pm in the afternoon. NOAA-17 is a "morning"
satellite, meaning its daytime observations are made around 10 am.
Thus, two codes are needed for both day and night. This convention is
consistent with NASA Aqua (an afternoon satellite) and NASA Terra (a
morning satellite), which also use these codes.
- VERSION = Indicates the version of the Pathfinder algorithm used to
create the file. Currently, this can be one of:
- pfv50 = Formal Pathfinder Version 5.0
- pfrt = Interim Pathfinder Version 5.0 (i.e., "Pathfinder Real Time")
- pfv51 = Formal Pathfinder Version 5.1
- pfrtn17 = Interim Pathfinder Version 5.0 from
NOAA-17 (these appear only in 2004 interim file names)
- TYPE = Indicates the type of data stored in the file. TYPE
may be one of the following:
- sst: Pathfinder all-pixel SST
- bsst: OISST V2 first-guess SST field
- sdev: Standard deviation
- num: Number of observations
- qual: Overall quality value
- msk1: Quality mask 1
- msk2: Quality mask 2
- BITS = Number of bits in each pixel. Only present for 16-bit
files (-16b). Note that this piece is dropped completely for
all 2003 and more recent data, as well as the V5.1 data
for 1981-early 1985.
- hdf = Indicates HDF-SDS Version 4 file format
Examples:
- 1990182.s04d1pfv50-sst-16b.hdf
- A daily, nighttime all-pixel SST file with 16-bit pixel values
at 4 km resolution from 1990, day 182.
- 1994181-1994185.m0451pfv50-num.hdf
- The number of observations in a 5-day, nighttime mean at 4 km
resolution using observations from 1994, days 181-185.
- 199035.m04w3pfv50-qual.hdf
- The overall quality flag values for the weekly, daytime mean SST
for 1994, week 35, at 4 km resolution.
- 1991169-1991176.s0483pfv50-bsst-16b.hdf
- The first-guess SST field used to create the 8-day mean SST for
year 1991, days 169-176. The data have a spatial resolution of
4 km and are stored as 16 bit unsigned integers.
- 199004.s04m1pfv50-sdev-16b.hdf
- The standard deviation of the monthly mean nighttime SST for
April 1990, at 4 km resolution, stored in 16-bit format.
How to Access the Data
Work is underway to make these Pathfinder V5 data available through a
wide variety of mechanisms, in an attempt to serve the broadest
possible array of users. These mechanisms at NODC include:
- Direct FTP Access: This is a
good approach if you know what you want and have a large number
of files you want to download. Simply point your favorite ftp client to
ftp://data.nodc.noaa.gov/pub/data.nodc/pathfinder and
begin downloading files.
- Simple Web Access: Simply click on
http://data.nodc.noaa.gov/pathfinder and begin browsing
through the file hierarchy. Clicking on any of the files will prompt
you to download that file or will launch any application you might have
associated with HDF files. If you want to just take a look around and
see what is there, or if you know which file or small number of files
you want to download, then this is a good approach.
- OPeNDAP Access: These data are also served using
OPeNDAP
(formerly DODS)
server so you may access all of these data using that distributed data
access system. The base URL is
http://data.nodc.noaa.gov/cgi-bin/nph-dods/pathfinder In
addition, you may simply follow the links provided in the table
on the Available Data page to
access the OPeNDAP directory listings. For a listing of OPeNDAP clients
which you can use to access the 4 km Pathfinder data (as well as any
other OPeNDAP-enabled data sets), please see the
OPeNDAP web site and follow the link
called "Download". Examine the list of "Clients" for various mechanisms to
access and display the data directly using OPeNDAP. For example, the
OPeNDAP Data Connector is an
excellent standalone client which will allow you to quickly search, access,
browse, and even generate images of the Pathfinder SST data.
- Live Access Server (and GoogleEarth): Our colleagues at
NOAA's South West Fisheries Science Center, Environmental
Research Division have established an excellent Live Access Server
which serves many datasets, including some of the AVHRR Pathfinder V5
dataset. Go to:
OceanWatch Live Access Server
- this customized interface includes GoogleEarth access. They also have
a more traditional Live Access Server interface - for direct access to
the AVHRR Pathfinder V5 data through it go to
http://las.pfeg.noaa.gov/OceanWatch/servlets/dataset?catitem=18.
- Low Resolution Browse Images: For 1985-2006, Simply click on
http://data.nodc.noaa.gov/pathfinder/Version5.0/browse_images
and begin browsing through low resolution JPEG images for every data file
on the system. Browse imagery for interim 2007 data are also available, at:
http://data.nodc.noaa.gov/pathfinder/Version5.0_interim_NOAA18/browse_images
and begin Note the browse images for 1985-2001 are more coarse, showing 1 out of
every 32 pixels, than those for 2002-2007, for which 1 out of every 16 pixels are
displayed.
- Climatologies: Corresponding climatologies based on the
1985-2001 base period are available via HTTP, FTP, and OPeNDAP. More
information on these climatologies is found in the
README.txt document. The URLs for the climatologies are the same as for the other Version5.0
data, but with _Climatologies appended:
To see a table listing the most current information on the data presently
available at NODC , please see the Available Data
page.
Much of the data for Pathfinder V5 are also available through the PO.DAAC's
data access systems.
Known Problems
-
The beginning of 1985 in the Version 5.0 Pathfinder record has sparse data and missing days. The NOAA-9 satellite was still in its pre-operational phase at this time. Because of this, we recommend that users use Version 5.1 data from NOAA-7 for early 1985, then switch to Version 5.0 for the rest of that year as follows:
-
Daily data: Use Version 5.1 for days 1-9 (1985001 through 1985009).
-
5-day data: Use Version 5.1 for the first two 5-day periods (1985001-1985005 and 1985006-1985010).
-
7-day data: Use Version 5.1 for the first 7-day period (198501).
-
8-day data: Use Version 5.1 for the first 8-day period (1985001-1985008).
-
Monthly data: Use Version 5.0 for all months in 1985.
-
Thanks to Scott Heron of NOAA's Coral
Reef Watch project for finding a bad quality flag file: the file
http://data.nodc.noaa.gov/pathfinder/Version5.0/Daily/2000/2000294.m04d1pfv50-qual.hdf
has a patch of erroneous quality flags extending from Florida, USA through
parts of the Gulf of Mexico and down to northernm South America. The
quality flags appear uniformly too high. These region is coincident with
an area that is missing data anyway, so there are no corresponding SST
values but users should take note of this time and region.
You can see the problem in the preview graphic:
http://data.nodc.noaa.gov/pathfinder/Version5.0/browse_images/Daily/2000/QualityFlag/2000294.m04d1pfv50-qual.hdf.jpg.
- Users of the data, especially in high latitudes, are cautioned to put
a "reality check" on the data even after applying the quality flags. In
some cases, the quality checks that would normally exclude unreasonable SST
values may fail (the two most relevant checks are the ones on the
brightness temperatures and the comparison against the OISSTv2 analysis).
In high latitudes where the SSTs can drop to as much as -1.8 or -1.9
degrees C these checks can sometimes fail, resulting in SSTs with values as
cold as -2.925 degrees C (pixel values between 1 and 15) that still have
high quality levels. Since surface ocean temperatures can not become this
cold, these data should be masked out. This problem occurs in the time
series data as well as in the resulting climatologies.
- A problem involving spurious SST gradients has been detected by
Dave Ullman and Peter Cornillon at the University of Rhode Island. This
problem manifests itself as a jump in SST value every 18 lines and is not
clearly visible in a single image. Instead, it was detected using edge
detection algorithms which examine sequences of data. This problem is also known to
exist in the older 9 km Pathfinder SST data and has also been seen in MODIS SST
data produced at the University of Miami.
The problem is related to the mapping procedures by which the
equal-area space bins in which the data are generated are mapped into the
equal-angle, uniform grid space in which the data are distributed. Future
versions of the Pathfinder dataset will correct this problem.
- Biases related to aerosol contamination from Saharan dust storms have been
identified in both 9 km and 4 km Pathfinder fields. Techniques to better mask
these contaminated pixels and possibly correct them are being investigated.
- A problem with the landmask was detected and reported by Dohyung Koo at
the Southampton Oceanography Centre in the UK. He found two areas which are
incorrectly classified as land off the east coast of South America, around the
Patagonian Shelf (around 35-50S and 70-50W). An image illustrating these
"false islands" is shown below.
Another problem has also been identified by Clive Mason (Bedford Institute of
Oceanography) near the South East Shoal, as illustrated in the image
(with thanks to Dr. Mason) below:
A third land mask problem was identified in the Antarctic. The last column
(column = 8192)
in the land data file is incorrectly set and indicates water not land. Pixels
in rows 3827 through the last row (4096) are affected. The arrow on the figure
below points to the area affected, but the area is too small to notice on a
global plot.
Improvements to the land mask are being developed and will be implemented in
a future Pathfinder Verison 6 reprocessing. However, these problem are evident
in all of the Pathfinder Version 5.0 data files now available.
Advice on Using Pathfinder V5 Data
Under development.
Example Read and Display Software
All of the Pathfinder SST data and ancillary fields are stored in HDF (version
4) Scientific Data Set format, or HDF-SDS. Many tools, libraries, and data
analysis and visualization environments are available which work directly with
this powerful, self-describing format:
- General Tools: HDF is developed and maintained at the National
Center for Supercomputing Applications. For access to many tools and
libraries, see their HDF4 web
site *. Remember as you are working with these files that they are
HDF-SDS, version 4. They are not HDF-raster and they are not HDF Version 5.
- Matlab: Matlab has many tools for working with HDF files and
a complete implementation of the HDF library, which enable efficient and
direct access to these data. For a GUI interface, examine the help
information on "hdftool", or simply do an "hdftool filename" to
start working with a file. You can also do a "help hdf" for general help
in using HDF files in Matlab, and doing a "help hdfsd" will show you a list
of functions for working with HDF-SDS files. There is also a simplified
function called "hdfread" so you can examine the help information on that
as well.
- Example Code in Matlab, Fortran, and IDL is available on the Pathfinder Web Site
under the
Example Code page.
Acknowledgements
Under development.
References
Under development.
|