Introduction
Sea-surface temperature (SST) is an essential climate variable (ECV). The
Global Climate Observing System (GCOS) project developed a list of ECVs to
focus worldwide observation efforts on a limited set of variables that are
climate relevant, technically feasible, and cost effective (Bojinski et al.,
2014). Collectively, ECVs can help develop adaptation and mitigation
strategies, assess risks, allow attribution and prediction, and support
climate services. SST is useful for monitoring El Niño events and
multi-decadal ocean changes. It is also relevant to quantification and
modeling of many other aspects of climate such as air–sea interaction, ocean
acidification to determine solubility of carbon dioxide, biophysical
processes, and marine organism distributions. However, models require not
just observations but also complete data fields, also referred to as analyses.
Today, satellites offer high spatial and temporal coverage and are,
therefore, the main source of SST observations. Additional processing is
applied to satellite data to form analyses to allow for bias corrections and
gap-filling and thereby increase spatiotemporal consistency.
The objective of this work is to describe the National Oceanic and
Atmospheric Administration (NOAA) 1/4∘
daily Optimum Interpolation SST version 2 (or dOISST.v2,
herein), an analysis that has been selected by the NOAA Climate Data Record (CDR) program as an
operational CDR. This implies that the dOISST.v2 meets the definition of CDR
put forward by the National Research Council (2004): it is of sufficient
length, consistency, and continuity to determine climate variability.
Furthermore, operational NOAA CDRs undergo a research-to-operations process
to ensure systematic production and quality assessment, thereby increasing
the data set maturity in aspects like transparency, usability, and data
preservation following metadata standards. A preliminary assessment using
the maturity matrix of Bates and Privette (2012) indicated that dOISST.v2
has a high maturity in both science and applications, but it needed
improvements in accessibility and transparency to users. As part of the
effort to address this deficiency, this paper describes the dOISST.v2 CDR
data set, in the context of its historical beginnings and evolution,
current temporal and spatial characteristics, and data set formats and access, and provides examples of applications. Much of this information is
publicly available but has not been summarized in a single document.
Historical background
Here, precursors to the dOISST.v2 that have evolved into the current CDR are
briefly reviewed to highlight the original motivation and subsequent
modifications. Historically, the widely-used name “Reynolds SST” has been
applied to all current and precursor products, and is therefore ambiguous
and not used here. Reynolds (1988) first introduced the concept of a blended
SST analysis that takes advantage of the sea truth offered by in situ data
and the high coverage of satellite data. Prior to 1980, ships were the only
source of observations, and the spatial–temporal coverage was sufficient
only for a coarse-scale analysis. Starting in late 1981, satellite-based SST
observations became available daily from an infrared instrument, the
Advanced Very High Resolution Radiometer (AVHRR), with Global Area Coverage
(GAC) resolution at ∼ 4 km. Using high-quality drifting buoys
as reference, Reynolds (1988) found that monthly analyses, based on AVHRR
SSTs alone or blended with in situ data, were slightly more accurate than
those based on in situ data alone (with drifter data withheld). However, for
infrared SSTs, notable satellite biases can occur under specific situations,
e.g., at cloud edges or dust plumes (e.g., Bogdanoff et al., 2015;
Vázquez-Cuervo et al., 2004). On a global scale, significant widespread
biases have also been observed in the presence of volcanic aerosols,
especially following the Mt Pinatubo and El Chichón eruptions, with AVHRR
SSTs cooler than in situ observations by over 1 ∘C (Zhang et al.,
2004; Reynolds, 1988, 1993). For these post-eruption periods,
reliable global SST fields could be produced if in situ data were used to
benchmark the satellite data to form blended analyses (Reynolds, 1988;
Reynolds and Marisco, 1993). This large-scale benchmarking is also referred
to as a “satellite bias adjustment”.
Reynolds and Smith (1994) adopted the optimum interpolation (OI) method to
increase the effective resolution of the blended analysis to 1∘
and the temporal frequency from monthly to weekly and later to daily. This
was the first time the name OISST was used. Along the marginal ice zone
where observations were sparse, the interpolation was relaxed to the
freezing point of seawater (-1.8 ∘C). This was slightly modified
in the follow-up version 2 (also referred to as OI.V2 in Reynolds et al.,
2002), where a regression equation was used to estimate proxy SSTs from sea-ice concentrations. NCEP continues to produce the 1∘ weekly OI.V2
for seasonal forecasting. The next section describes the original and
current versions of dOISST. The core methodology for both versions is
described in Reynolds et al. (2007). Appendix A presents minor improvements
made by Reynolds in response to feedback from users (mostly added smoothing
to reduce noise) and to extend the series back to 1981. Note that Appendix A
contains material from Reynolds (2009), an online document, and has been
included in this paper for preservation purposes.
The climate data record
Reynolds et al. (2007) introduced major methodological changes to increase
the OISST resolution to the current daily, 1/4∘ grid.
The new bias correction scheme employed empirical orthogonal teleconnections
(EOTs) modes rather than the Poisson method used in the weekly OI.V2. The
use of EOTs also had the additional advantage of allowing estimation of
the bias error. Moreover, the earlier OISST analyses used operational (i.e.,
near-real-time) satellite data, but for the 1/4∘
daily OISST higher-quality input data sets reprocessed from the start of the
mission were used preferentially over operational data. Most significant was
the AVHRR Pathfinder reprocessing, which improved SST retrievals using
nighttime buoy data to compute a revised set of coefficients for each NOAA
satellite (e.g., Kilpatrick et al., 2001; Casey et al., 2010; Vázquez-Cuervo
et al., 2010). However, the cold bias associated with the eruptions of El
Chichón and Mt. Pinatubo remained a challenge even in the more recent AVHRR
SST reprocessing efforts. Another change from the weekly OI.V2 was that the
proxy SST calculation was restricted closer to the ice margin, i.e., where
sea-ice concentrations exceed 50 %, to avoid potentially erroneous SST
estimates in the more open waters, where the fit was much noisier.
Reynolds et al. (2007) referred to the above product as AVHRR-only, in
reference to the source of satellite SSTs. The same paper describes a
companion analysis that uses the same methodology but includes data from the
Advanced Microwave Scanning Radiometer on the Earth Observing System
(AMSR-E). This product, called AVHRR+AMSR, is not a CDR, due to the short
period of record (2002 to 2011) of AMSR-E data. It should be noted that the
Climate Forecast System Reanalysis (CFSR) at NCEP uses the same Reynolds et
al. (2007) methodology to generate their initial SST fields (Saha et al.,
2010, 2014), but may use different inputs (e.g., both infrared and microwave
satellite SSTs, like the AVHRR+AMSR) over time and therefore will differ
from this CDR.
As discussed in Appendix A, the 1/4∘ daily OISST was
upgraded to version 2 primarily to increase temporal stability and to
include Pathfinder AVHRR data from 1981 to 1985 that had become available
(Casey et al., 2010). The treatment of in situ data was slightly modified.
Historically, in situ observations were predominantly from ships. In the
21st century, more accurate buoy data had become increasingly dominant
over ship data, providing a better reference temperature. A constant
(∼ 0.14 ∘C) was subtracted from the ship data to
compensate for the global-average ship–buoy difference (Reynolds and Chelton,
2010; Appendix A). Modern ship measurements tend to be warmer because they
typically use intake samples that can be warmed when taken into the ship
engine room. However, there is much scatter in individual differences and
better understanding of ship bias is needed to reduce the uncertainty in
this correction. The net effect of this adjustment is that the daily OISST
tends to be cooler than the weekly OI.V2 particularly in the 1980s and 1990s
when ship data were dominant, making the long-term trend slightly steeper.
Some of the differences between the current weekly OI.V2 and the current
CDR, i.e., the dOISST.v2, including the impact on trends and climatologies,
are discussed in greater detail in Banzon et al. (2014) and Huang et al. (2016).
Data set description
The daily OISST is available in netCDF and binary (FORTRAN IEEE big-endian)
formats. In this paper, the archived netCDF files, publicly available at the
National Centers for Environmental Information (NCEI) website, are
described. However, the same data are repackaged and distributed elsewhere
for specific projects or organizations such as the Group for High Resolution
SST (or GHRSST) and Observations for Model Intercomparison Projects
(Obs4MIPs), with accompanying metadata and documentation, but are not
described here. The heritage binary format will be eventually phased out.
Examples of the four variables in a singles file: (a) dOISST.v2,
(b) dOISST anomaly, (c) error, and (d) median sea-ice
concentrations. Data are shown for 20 June 2015.
![]()
A single netCDF file contains four global gridded fields (1440×720)
pertaining to 1 day. The primary variable is the analyzed SST (units in ∘C;
Fig. 1a). Since buoys are used as a reference, this is
sometimes referred to as a “bulk” SST, at a nominal buoy depth of 1 m. The
SST input data types (AVHRR daytime, AVHRR nighttime, buoys, ships, proxy
SST; Table 1) are first averaged to 1/4∘ superobservations. The in
situ data (consisting of the buoy and adjusted ship data) are collectively
used to make large-scale adjustments to satellite data using the EOT modes.
All data are merged during the interpolation, using the pre-computed error
characteristics as weights. More details can be found in Reynolds et al. (2007).
Grid points corresponding to land, permanent ice shelves, and most
inland waters are not processed and assigned a missing value of -999.
Input data sets to the daily OISST version 2. The data sources are
explained in detail in Reynolds et al. (2007). For version 1, ICOADS 2.1 was
used.
Input type
Reprocessed or higher-quality data sets
Operational data sets
Satellite (AVHRR SSTs)
Pathfinder 5.0/5.1 (1981–2006)
Navy (2007–present)
In situ SSTs
ICOADS 2.4 (1981–2006)
NCEP (2007–present)
Sea-ice concentrations
GSFC NASA (1981–2004)
NCEP (2005–present)
Three other gridded fields at the same 1/4∘ spatial resolution
complement the daily analysis.
Anomalies (i.e., the daily OISST minus the 1971–2000 climatological mean;
units in ∘C; Fig. 1b) represent departures from “normal” or
average conditions. The anomalies are provided so users can easily compute
climate indices, such as the NINO3.4 (Fig. 2a). The 1971–2000 climatology is
partly based on an in situ analysis for the years that satellite data are
not available (1971–1981) and on the weekly OISST for years satellite data
are available from 1982 onward (Xue et al., 2003). A climatological mean
computed from daily OISST (1982–2011) is now available and is more suitable
to use with this data set, as explained in Banzon et al. (2014). It will be
used in the next version. User should consider that with a long-term
warming, a more recent period may produce a warmer climatological mean;
thus, when subtracted from the analyzed SSTs, it produces cooler anomalies.
The standard error (with units in ∘C; Fig. 1c) provides a measure
of uncertainty in the estimated SST, allowing users to exclude (using a
threshold) or to minimize (using weights) the importance of grid point
values with greater errors, as needed for the specific application, e.g.,
resource management, risk analysis, or assimilation into a model.
The 7-day median of daily sea-ice concentrations (expressed as a real
fraction from 0.0 to 1.0; Fig. 1d) is the basis of the proxy SST estimate in
the marginal ice zone. Aside from reducing noise, the temporal median
populates the time series in the early 1980s when satellite sea-ice
observations were available only every other day. There are no sea-ice data
from 4 December 1997 to 14 January 1998. This field is effectively also an
ice mask when the user opts to exclude areas with high ice concentrations.
(a) Temporal progression of the 1997 daily OISST in the NINO3.4
region (solid line) and the 1982–2011 climatological mean (short dash) for
the same area. The offset from the mean by plus and minus 1 standard
deviation (long dash) shows characteristic variability. (b) Same but in 2012
in the Gulf of Maine (after Mills et al., 2013). Titles show coordinates
used to compute the area weighted means.
![]()
The input data sets to dOISST.v2 are listed in Table 1 and have been
evaluated in more detail in Reynolds et al. (2007). While reprocessed inputs
are used whenever possible, only operational data sets meet the low latency
needs of the daily updates. Users should be aware that sensor problems are
typically cannot be addressed in near real time. The release date of the
dOISST.v2 was November 2008. To minimize the impact of near-real-time sensor
problems, data from two AVHRRs are used from 2007 onward (Table 2).
Platform time spans of AVHRR inputs to the daily OISST. Note that
two satellites at a time are used beginning January 2007.
Data set
Start date
End date
Platform
Sensor
Pathfinder
24 Aug 1981
3 Jan 1985
NOAA-7
AVHRR/2
4 Jan 1985
7 Nov 1988
NOAA-9
AVHRR/2
8 Nov 1988
13 Sep 1994
NOAA-11
AVHRR/2
14 Sep 1994
21 Jan 1995
NOAA-9
AVHRR/2
22 Jan 1995
11 Oct 2000
NOAA-14
AVHRR/2
12 Oct 2000
31 Dec 2002
NOAA-16
AVHRR/3
1 Jan 2003
4 Jun 2005
NOAA-17
AVHRR/3
5 Jun 2005
31 Dec 2006
NOAA-18
AVHRR/3
Navy
1 Jan 2006
31 Dec 2008
NOAA-17
AVHRR/3
1 Jan 2007
15 Aug 2011
NOAA-18
AVHRR/3
1 Jan 2009
Present
MetOP-A
AVHRR/3
16 Aug 2011
Present
NOAA-19
AVHRR/3
The analysis for the first day in the record used climatology as a first
guess. For all other days the previous analysis is used as a first guess.
For the daily update, a 1-day delayed analysis is produced. Two weeks later,
after more data have become available, the analysis is repeated to produce
higher-quality “final” product. The final and preliminary runs can be
identified in the global attributes of the netCDF file, and the preliminary
filename also contains the word “preliminary”. Only the “final” product
is archived.
Basic characterization
The daily OISST is available for the full period of record from
September 1981 to the present. The data set is similar to other global daily
SST analyses in that monthly, seasonal, and multi-year averages can be computed
on global, regional, and local scales. For climate applications, the daily
OISST is unique because it extends from late 1981 to the present and
therefore spans over 30 years, often cited as the minimum period needed to
distinguish interannual variations from long-term variations. The
characteristic seasonal SST cycle, represented here by the 1982–2011
climatological mean, varies by location. In the tropics, it is exemplified
by the NINO3.4 region (Fig. 1a), where the seasonal signal is weak. The
start of the 1997 El Niño event is marked by SSTs that are more than 1
standard deviation greater than the climatological mean for over 3
months. A stronger seasonal cycle occurs in the temperate zone, as seen in
the Gulf of Maine (Fig. 2b). The SSTs over the entire year 2012 exceed the
climatological mean plus 1 standard deviation. The daily progression shows
particularly elevated May–June temperatures, which initiated a season of
anomalous lobster catch (Mills et al., 2013). Of course, these atypical
events can also investigated by examining the anomalies.
Long time series are ideal for computing multi-decadal trends. On an annual
scale, the 1982 to 2014 global linear trend using dOISST is ∼ 0.12 ∘C
per decade. The wintertime trend is slightly smaller
(0.09 ∘C per decade using only January monthly averages; Fig. 3a)
than in summertime (0.14 ∘C per decade using only July monthly
averages; Fig. 3b). The 1/4∘ resolution data allow
trends to be computed on more local scales, but comparisons should always be
made with in situ measurements, if available. The daily SST information can
be used to generate other climate-relevant parameters. For example, the
number of days that SSTs are above a threshold, also known as degree days,
is an indicator of thermal stress for corals. In fact, many ecological
responses to a changing ocean can be modeled in terms of the cumulative
effect of daily temperatures on growth, reproduction, recruitment, and the
like.
Global OISST trends (1982–2014) using (a) January monthly means
only and (b) July monthly means only.
![]()
Global validation of the dOISST using buoy and ship data is not an
independent assessment because in situ data are used to make the product,
although the amount of satellite data incorporated is much greater.
Comparisons with other SST analyses would have the same issue since most
analyses also use in situ data. With that caveat, Reynolds and Chelton
(2010) showed that, relative to buoys, the dOISST.v2 and other analyses all
exhibit regional variability in performance, reflecting their methodological
differences. For the dOISST.v2, the root-mean-square error relative to the
buoys is about 0.3 ∘C. Reynolds and Chelton (2010) also found
that the product degrades in quality during prolonged periods of no data,
e.g., seasonally cloud-covered areas such as the Gulf Stream in winter.
Analyses that included microwave data had better results when infrared
retrievals were not available, because of the added data coverage. However, when
infrared data were available, the addition of microwave data reduced the
quality of the resulting analyses because microwave SSTs are less accurate.
In any case, an analysis is not necessarily the best source of SST for a
single point in space or time because it is a smoothed product. The
advantage is that where there are no observations, an analysis provides
interpolated values and, over time, a consistent long-term record. It should
be noted that the analysis is also useful as a reference field for
identifying bad data and is therefore used in several satellite algorithms
for quality control. It also serves as a first guess or as ancillary data
for computing parameters that require a known temperature field.
Argo data, which are not used as an input to dOISST, can also be used for
validation. However, Argo observations are available only after 2000 and
are located deeper (∼ 4 m) than surface buoy measurements. As
agreed on by the GHRSST community, most SST analyses do not incorporate Argo
data in order to have a common independent validation data set.
Martin et al. (2012) used Argo data for the year 2010 to compare different near-real-time
SST data sets, including the preliminary dOISST (defined in Sect. 3.1), and
their collective median. All data sets had a standard deviation below 0.7 K
relative to Argo data. Most, including dOISST, had an overall bias between
0.2 to 0.4 K. Certainly, as the Argo data set grows, it might be possible to
withhold only a portion for validation and use the rest in the analysis.
For a future version of dOISST, Argo data could improve data coverage in
areas with sparse ship and buoy data such as the Tropical Pacific, where
moored buoy data were degraded for some years.
Power spectra of three analyzed SST fields from the first 2
months of 2016. See text for explanation of data sets used. At smaller
scales, the spectrum of the RSS product that ingests high-resolution inputs
(1 km MODIS SSTs) continues to display the same spectral slope into the
smaller-scale range. MODIS SSTs are available only from 1999.
![]()
In terms of feature resolution, i.e., the ability of an analysis to
reproduce mesoscale ocean features and capture SST gradients, Reynolds et
al. (2007) showed dOISST performs well. Reynolds and Chelton (2010) also
found that feature resolution of an analysis is not necessarily related to
the grid size. To illustrate this point here, the power spectral densities
of three SST data sets examined by Reynolds and Chelton (2010) are shown
(Fig. 4). The plot is similar to Reynolds and Chelton (2010) except that the
latest versions of the three data sets (from the first 2 months of 2016)
are used and the spectra are smoother because they are the average of
several areas rather than a single area, in order to provide a global
representation of each data set. The three products shown differ in grid
resolutions: dOISST is on a 1/4∘ grid, the Operational SST and Sea
Ice Analysis (OSTIA) is on a 1/20∘ grid
(10.5067/GHOST-4FK01; UK Met Office, 2005), and the Remote Sensing Systems
(RSS) analysis is computed on a 1/11∘ grid
(10.5067/GHMWI-4FR01; Remote Sensing Systems, 2008). All spectra indicate identical
large-scale (and perhaps seasonal) SST patterns at wavelengths larger than
1500 km. All spectra also show similar slopes (of -2) in the mesoscale
range down to about 300 km wavelength. At shorter scales (< 200 km),
the power spectra for dOISST and OSTIA are nearly identical even when the
latter has a finer grid size. In comparison, the RSS product resolves
finer-scale features as indicated by the more gradual drop off.
In general, the higher-resolution SST analyses combine higher-resolution
infrared data (with low spatial coverage due to inability to penetrate
clouds) and lower-resolution but more spatially complete microwave data,
which have quasi-all-weather coverage (e.g., Vázquez-Cuervo et al., 2013).
The ability to provide good feature resolution in a particular area is
constrained by the availability of finer resolution (∼ 1 km)
infrared data and the use of a methodology that preserves that information.
Thus, even though the average spectra may show an ability to resolve fine
features, products tend to be smoother in areas where only microwave data are
available (Vázquez-Cuervo et al., 2013). High-resolution artifacts may also
be created due to insufficient data (Reynolds and Chelton, 2010).
Conclusions
A long-term sea-surface temperature climate data record consisting of in
situ and satellite data blended daily on a 1/4∘ grid
is available for climate monitoring, modeling, validation, and a wide range
of other applications. The data set uses AVHRR data from 1981 to the present,
bias-adjusted relative to situ data. This produces a time series that is
more consistent than satellite infrared retrievals alone. This CDR is
produced, distributed, and generated by the NCEI, a newly formed entity that is a merger of
three NOAA data centers including the National Climatic Data Center (NCDC),
which originally produced this data set.
Compared to the precursor weekly OISST at NCEP, the CDR has many updates
including higher spatial resolution, reprocessed inputs, and adjustment of
ship data to match buoys. The CDR is also used as an ancillary field in
reprocessed and operational satellite algorithms including the Pathfinder
AVHRR SST, Tropical Rainfall Measuring Mission (TRMM) rain rate, and
Aquarius salinity. The CDR version of the dOISST.v2 is available in netCDF
format 10.7289/V5SQ8XB5 (Reynolds et al., 2008).
Data availability
The dOISST.v2 data set described in this paper is available at the National
Centers for Environmental Information, under the name “NOAA Optimum
Interpolation 1/4 Degree Daily Sea Surface Temperature (OISST) Analysis,
Version 2” with 10.7289/V5SQ8XB5. The data are also available in a
different format at the Physical Oceanography Distributed Active Archive
Center (PODAAC) at the Jet Propulsion Laboratory (JPL) under the name
“GHRSST Level 4 AVHRR_OI Global Blended Sea Surface Temperature Analysis”,
with 10.5067/GHAAO-4BC01. The two other SST products used for
comparisons in Fig. 4 are also available at the PODAAC. The “GHRSST Level 4
mw_ir_OI Global Foundation Sea Surface Temperature analysis” has
10.5067/GHMWI-4FR01. The “GHRSST Level 4 OSTIA Global Foundation Sea
Surface Temperature Analysis” has 10.5067/GHOST-4FK01.