-
Notifications
You must be signed in to change notification settings - Fork 1
/
1.html
81 lines (80 loc) · 11.2 KB
/
1.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
<!DOCTYPE html>
<html>
<head>
<link rel="stylesheet" type="text/css" href="sst_review_style.css">
</head>
<title>Introduction</title>
<body>
<div class="row">
<div class="column side">
<p></p>
</div>
<div class="column middle">
<p><a href="index.html">Index</a> <a href="2.html">Next - General classification of uncertainties</a></p>
<h1>1. Introduction</h1>
<p>
Measurements of the temperature of the sea surface have been made for more than 200 years for a wide variety of purposes. The earliest measurements of sea-surface temperature (SST) in the eighteenth century were taken out of pure scientific interest. Later, after the connection between SST and ocean currents was made, large numbers of measurements were made for the construction of navigational charts. In the twentieth century, the needs of weather forecasting and, to an extent, the need to produce marine climate summaries determined the quantity and quality of observations. Most historical SST measurements were not made by dedicated scientific vessels, but by voluntary observing ships (VOS) on the basis that they would contribute to the safety of life at sea. This is reflected in the geographical distribution of observations, which are largely confined to major shipping lanes.
</p><p>
Nowadays, in situ measurements of SST - those made at the surface as opposed to those
made remotely by satellites or aircraft - are used in diverse applications. They are
used directly in calibration and validation of satellite retrievals and they are
assimilated into ocean analyses [Roberts-Jones et al., 2012]. They are also used to
construct data sets of summaries of SST on regular grids and globally-complete SST
fields are created using statistical techniques to impute SSTs in regions where there
are no observations. The SST data sets and statistical SST 'reconstructions' or
'analyses' are widely used, for example as an index of global climate change
[Morice et al., 2012], as a boundary condition for climate simulations [Folland, 2005]
and reanalyses [Simmons et al., 2010], as initial conditions for decadal forecasts
[Smith et al., 2007], in studies of hurricane formation [Saunders and Harris, 1997] and
in studies of the impact of climate change on marine ecosystems
[Sheppard and Rayner, 2002].
</p><p>
As the demands for SST measurements have changed, so have the instruments used to make them, and so have the ships and other vessels from which the measurements were made. The first systematic observations were made using buckets to collect a water sample. Buckets made of wood, canvas, tin, leather, brass, rubber and plastic of designs as various as the materials employed in their construction have all been used to measure the temperature of the surface layers of the ocean. There are two problems with this approach. The first is that during the collection and hauling, the temperature of the water sample can be modified by the combined actions of latent and sensible heat transfer and the warmth of the Sun. Even in the best conditions, an accurate measurement requires diligence on the part of the sailor; that is the second problem. Improvements to minimize the physical effects were made to bucket designs during the 1950s, but as ships became larger and faster, the making of bucket measurements became not just thankless, but dangerous.
</p><p>
After the advent of steam ships in the late nineteenth century, it was routine to measure the temperature of the sea water that was circulated through the steam condenser. Condenser inlet measurements and later, engine room inlet (ERI) measurements, were often recorded in ship logbooks, but they were not entered into meteorological logs until the 1930s. The convenience of using measurements that were made as a matter of routine, and the attendant reduction in the risk of losing a bucket or sailor overboard, meant that ERI measurements became the preferred method for measuring SST on board ships during the latter half of the twentieth century. That is not to say that the method was without its difficulties: modification of the temperature of the water between inlet and thermometer was still a problem and it was now compounded by the varying depth of the measurements.
</p><p>
Since the 1970s, a growing number of ships have been fitted with dedicated sensors either outside or inside the hull. These have been joined by a growing array of moored and drifting buoys which make automated measurements that are relayed by satellite. At present, around 90% of all SST observations come from buoys. In calm conditions drifting buoys measure at a nominal depth of between 10 and 40 cm depending on their size. However, wave motion means that in some conditions the buoy will be submerged for part of the time and report temperatures that are representative of something like the upper 2 m.
</p><p>
Moored buoys are fixed platforms, akin, in some ways, to meteorological stations on land. They come in a variety of shapes and sizes. Most are a few meters in height and width, but the largest in regular use are the 12 m Discus buoys designed to weather the wilder climates of the northern oceans. There are two loose groupings of moored buoys: the Global Tropical Moored Buoy Array (GTMBA) and a more diverse group of coastal moorings mostly around the US. The GTMBA has regular arrays of moorings in the tropical Pacific, Atlantic and Indian Oceans. The majority of moored buoys measure SST at a nominal depth of 1 m. Some measure slightly deeper and some moorings make measurements at a range of depths.
</p><p>
SST measurements from ships and buoys together with near-surface measurements made
by oceanographic cruises have been gathered in digital archives. The largest and most
comprehensive of these is the International Comprehensive Ocean-Atmosphere Data Set
(ICOADS, Woodruff et al. [2011], Freeman et al. [2017]). The latest release of ICOADS,
release 3.0, contains individual marine reports from 1662 to 2014, but air and sea
temperature measurements only start to appear in the 19th Century. Metadata giving
information about some of the measurements and the ships that make them is also
provided and has been complemented by information from bulletins such as <a href="http://www.wmo.int/pages/prog/www/ois/pub47/pub47-home.htm">WMO
publication 47</a> and in
<a href="ftp://esurfmar.meteo.fr/pub/Pub47/PUB_47_export_esurfmar_database_active_vos_v3a.csv">near real time</a>.
</p><p>
Other digital archives exist. <a href="http://coaps.fsu.edu/RVSMDC/index.shtml">Research vessel</a>
(RV) data are gathered at the Research Vessel
Surface Meteorology Data Center at Florida State University. <A href="http://www.whoi.edu/">Woods Hole Oceanographic
Institute</a> maintains an archive of research mooring data and the
<a href="http://www.oceansites.org/data/index.html">OceanSites website</a> provides
links to other
mooring data. The <a href="http://www.pmel.noaa.gov.tao/global/global.html">Pacific Marine Environmental Laboratory</a>
maintains an archive of water
temperature measurements from the GTMBA at a range of depths and time resolutions that
are not available in ICOADS.
Near-surface measurements from other sub-surface sources such as the Argo array of
autonomous profiling floats also exist. Both near-surface and sub-surface measurements
are gathered together in the Met Office Hadley Centre Integrated Ocean Database
(HadIOD, Atkinson et al. [2014]) together with estimates of uncertainty and bias
adjustments where available.
</p><p>
Despite being comprehensive, ICOADS is incomplete. Large archives of paper records exist around the world and many of these have yet to be digitized. It is not possible yet to know exactly how many undigitized records remain because there is no definitive catalogue of global archives. What is known is that many archives that have been identified are far from being exhausted. The potential for reducing the uncertainty in SST analyses as well as in reconstructions of other marine variables is clear, but funding, particularly sustained funding for the efforts to identify, image and key the data has proved difficult to find. Nonetheless, there have been some successes such as a project to crowd source the keying of Royal Navy logbooks from the First World War. Volunteers on the OldWeather.org project keyed pages from the logbooks online. In the first three years of the project, more than 1.6 million weather observations were digitized, by around 16,400 volunteers.
</p><p>
The observing network was not created with a single purpose in mind. It was certainly not intended to meet the stringent criteria demanded for monitoring long-term environmental change. Nonetheless, historical SST measurements have been widely used in such studies. In a 2010 paper, Jones and Wigley [2010] identified uncertainties associated with pervasive systematic errors in SST data sets as an important uncertainty in the estimation of global temperature trends. The obvious gulf between the ideal and the reality leads naturally to questions about the reliability of the SST record. Often this question is couched as a yes/no dichotomy: "are SST records reliable?" A more useful question is "How reliable are they?" Although historical measurements were not made for climate research, or any single purpose, it does not mean that it is impossible to derive from them a record that is useful to a particular end. However, it does mean that special care must be taken in identifying and, as best as possible, quantifying uncertainties.
</p><p>
In using SST observations and the analyses that are based on them, it is important to understand the uncertainties inherent in them and the assumptions and statistical methods that have gone into their creation. In this review I aim to give an overview of the various components that contribute to the overall uncertainty of SST measurements made in situ and of the data sets that are derived from them. In doing so, I also aim to identify current gaps in understanding.
</p><p>
<a href="2.html">Section 2 provides a general classification of uncertainties</a>. The classifications are not definitive, nor are they completely distinct. They do, however, reflect the way in which uncertainties have been approached in the literature and provide a useful framework for thinking about the uncertainties in SST data sets. The uncertainties have been tackled in ascending order of abstraction from the uncorrelated errors associated with individual observations to the generic problem of unknown unknowns. In this review quoted uncertainties represent one standard deviation of the relevant distribution unless otherwise stated. <a href="3.html">Section 3</a> applies this framework to analyze progress and understanding under each of the headings. Some shortcomings of the presentation of uncertainties are discussed in <a href="4.html">section 4</a> along with possible solutions. <a href="5.html">Section 5</a> reviews how some analyses have used knowledge of likely errors in SST data sets to minimize their exposure to uncertainty. <a href="6.html">Section 6</a> briefly discusses SST retrievals from satellites and how these have been used to understand the in situ record. The review <a href="7.html">concludes</a> with a summary of possible future directions.
</p>
<p><a href="index.html">Index</a> <a href="2.html">Next - General classification of uncertainties</a></p>
<br><br><br><br>
</div>
</div>
</body>
</html>