水桶名單
修正歷史上的海洋溫度測量記錄之後得出較為一致的海洋暖化模式
By Leah Burrows
20世紀早期的海洋有些不對勁：北大西洋和東北太平洋的暖化速度是全球平均值的兩倍，而西北太平洋數十年來卻越來越冷。
哈佛大學最近發表的研究修正了數十年的海水表面溫度資料，因而解開了一道歷時許久、有關全球氣候變遷的難題。
大氣和海洋模型都難以解釋為何兩者的溫度變化不一致，這成為氣候科學上的一道謎題：為何20世紀早期海洋不同區域變暖和變冷的速度會差這麼多？
最近美國哈佛大學和英國國家海洋學中心的研究人員給出的答案，包含了小數點捨去這種平凡無奇的因素，卻也如同國際政治般地錯綜複雜。他們同時從歷史和氣候科學的觀點著手修正數十年間的氣候資料，得出了比之前更加一致的海洋暖化趨勢。
這項研究刊登在《自然》(Nature)。
人類測量並記錄海洋表面的溫度已經有數個世紀的歷史。海洋表面的溫度可以幫助船員確認船隻的航線、辨識自身所處的方位，還能用來預測即將到來的風暴。
在1960年代以前大多數測量海洋表面溫度的方法，是把水桶丟到海裡，測量裝到的海水溫度。
美國國家海洋暨大氣總署(NOAA)和美國國家科學基金會的國家大氣研究中心(NCAR)保存的海洋表面溫度記錄可以追溯到19世紀早期。這組資料庫來自漁船、商船、研究船、軍艦等各式船隻在世界各地海洋進行的觀測結果，包含了超過1億5500萬筆數據。當我們想要瞭解隨著時間經過，海洋表面溫度因為自然與人為因素而出現的變化時，這些觀測結果具有非常重要的價值。
但它們在統計學上也是猶如噩夢一般的存在。
比方說，要如何比較1820年的英國軍艦、1920年的日本漁船和1950年的美國海軍得到的測量結果？要怎麼知道他們用的水桶類型是什麼，以及他們採樣時水溫因為陽光照射而上升，或是因為蒸發而下降了多少？
舉例而言，平常天氣時甲板上的帆布水桶經過三分鐘後，可以比同條件下的木製水桶多下降0.5℃。有鑑於20世紀的全球暖化幅度大概也就1℃左右，不同測量程序產生的偏差便需要列入詳細考量。
論文資深作者Peter
Huybers表示：「這個資料庫含有的數據多達數十億位元，而每一筆數據背後都有一段曲折的故事。」Huybers是哈佛大學約翰·保爾森工程與應用科學學院的環境科學與工程教授，同時任職於該校的地球與行星科學系。「這些數據參雜了許多特異份子。」
過去許多研究的目標便是找出這些特異份子並加以矯正。比分說2008年一群研究人員發現海洋表面的溫度在1945年突然躍升0.3℃的原因為：這筆數據是測量進入引擎室的水而得到的。然而就算經過諸多修正，數據仍然難以稱得上完美，而海洋表面的溫度變化還是有無法解釋的地方。
在這項新研究中，Huybers與同事提出一種更加全面的方法來修正數據。這種新的統計方法比較了鄰近船隻的測量結果。
「我們的方法檢視了不同類型的船隻經過彼此時對海洋表面溫度的測量結果。『經過』是指它們的距離曾經在兩天之內不到300公里。」第一作者Duo
Chan表示。他是哈佛大學文理學院的研究生。「利用這種方法，我們找出了1780萬次的擦身而過，並發現某些群體具有重大偏差。」
研究人員把重點放在1908至1941年的數據。這些數據以船隻所屬國家和它們所在的「卡疊」來加以分類，因為這些海洋觀測數據是以成疊的打孔卡儲存起來。其中一疊甚至含有羅伯特•史考特與歐內斯特˙謝克頓遠征南極途中的觀測結果。
「這些數據需要經過諸多處理，才能從原本的航海日誌轉化成現代科技能讀取的檔案。要把航海日誌中的資訊變成符合打孔卡的形式，或者儲存成盤式磁帶但數量不至於無法處理，需要作出許多艱困的取捨。」共同作者，英國國家海洋學中心的Elizabeth
Kent表示。「透過我們的新方法並配合現在電腦的計算能力，可以顯示這些取捨對數據產生了什麼樣的影響，並且看出不同國家採用的觀測方法造成的偏差。這讓我們的溫度歷史紀錄離實際情形更加接近。」
研究人員找到了兩個關鍵因素，可以解釋北大西洋和北太平洋暖化情形的差異。
第一個因素跟日本船隻的紀錄方式改變有關。1932年之前，日本船隻對北太平洋海水表面的溫度測量紀錄多數是由漁船進行。這些數據分散在數個卡疊之中，最初它們記錄時是整數的華氏溫度，之後儲存時先轉成攝氏溫度再四捨五入到小數點後第一位。
後來到了二戰期間，越來越多日本船隻的紀錄是來自海軍艦艇。這些數據單獨存放在另一個卡疊之中，美國空軍數位化這些紀錄時有所刪減――他們把小數點以下的部分通通刪去，把資料存成整數的攝氏溫度。
Huybers表示之前估計太平洋海水表面溫度時，可以看到1935至1941年有相當明顯的迅速冷化現象，這大部分可以歸因於沒有找出上述刪減動作帶來的影響。把刪減造成的偏差修正之後，北太平洋的暖化趨勢變得一致許多。
破解20世紀早期太平洋暖化趨勢的鑰匙在於日本船隻的數據；而想要瞭解同一時期北大西洋的海水表面溫度如何變化，最重要的關鍵則是德國船隻的數據。
從1920年代晚期開始，大部分北大西洋的數據是由德國船隻提供。這些數據多數存放在同一個卡疊中，跟鄰近群體的測量結果比較起來明顯溫度較高。把它們修正過後，北大西洋的暖化趨勢變得較為緩和。
經過調整之後，研究人員發現北太平洋和北大西洋的暖化速率變得一致許多，而且暖化模式也更接近預測中溫室氣體濃度升高會出現的趨勢。然而兩者之間還是有所差異；而且測量結果整體呈現出來的暖化速率，仍然高過模型做出來的預測。
Huybers說：「兩者之間仍然有所差異強調出我們必須繼續探討太陽輻射如何驅動氣候、氣候受到不同因子的影響程度以及氣候本身就有的變化。在此同時，我們也得持續統整資料――透過資料科學、釐清歷史因素以及更加瞭解問題本身的性質，我敢斷定未來還會發現更加有趣的事物。」
共同研究作者還有英國國家海洋學中心的David
I. Berry。研究經費來自哈佛全球研究所、美國國家科學基金會、英國自然環境研究委員會。
The bucket list
Correcting historic sea surface
temperature measurements reveals a simpler pattern of ocean warming
Something odd happened in the oceans in
the early 20th century. The North Atlantic and Northeast Pacific appeared to
warm twice as much as the global average while the Northwest Pacific cooled
over several decades.
Atmospheric and oceanic models have had trouble
accounting for these differences in temperature changes, leading to a mystery
in climate science: why did the oceans warm and cool at such different rates in
the early 20th century?
Now, research from Harvard University and the UK’s
National Oceanography Centre points to an answer both as mundane as a decimal
point truncation and as complicated as global politics. Part history, part
climate science, this research corrects decades of data and suggests that ocean
warming occurred in a much more homogenous way.
The research is published in Nature.
Humans have been measuring and recording the sea
surface temperature for centuries. Sea surface temperatures helped sailors
verify their course, find their bearings, and predict stormy weather.
Until the 1960s, most sea surface temperature
measurements were taken by dropping a bucket into the ocean and measuring the
temperature of the water inside.
The National Oceanic and Atmospheric Administration
(NOAA) and the National Science Foundation’s National Center for Atmospheric
Research (NCAR) maintains a collection of sea surface temperature readings
dating back to the early 19th Century. The database contains more than 155
million observations from fishing, merchant, research and navy ships from all
over the world. These observations are vital to understanding changes in ocean
surface temperature over time, both natural and anthropogenic.
They are also a statistical nightmare.
How do you compare, for example, the measurements of
a British Man-of-War from 1820 to a Japanese fishing vessel from 1920 to a U.S.
Navy ship from 1950? How do you know what kind of buckets were used, and how
much they were warmed by sunshine or cooled by evaporation while being sampled?
For example, a canvas bucket left on a deck for three
minutes under typical weather conditions can cool by 0.5 degrees Celsius more
than a wooden bucket measured under the same conditions. Given that global
warming during the 20th Century was about 1 degree Celsius, the biases
associated with different measurement protocols requires careful accounting.
“There are gigabytes of data in this database and
every piece has a quirky story,” said Peter Huybers, Professor of Earth and
Planetary Sciences and of Environmental Science and Engineering at the Harvard
John A. Paulson School of Engineering and Applied Sciences (SEAS) and senior
author of the paper. “The data is rife with peculiarities.”
A lot of research has been done to identify and
adjust for these peculiarities. In 2008, for example, researchers found that a
0.3-degree Celsius jump in sea surface temperatures in 1945 was the result of
measurements taken from engine room intakes. Even with these corrections,
however, the data is far from perfect and there are still unexplained changes
in sea surface temperature.
In this research, Huybers and his colleagues proposed
a comprehensive approach to correcting the data, using a new statistical
technique that compares measurements taken by nearby ships.
“Our approach looks at the differences in sea surface
temperature measurements from distinct groups of ships when they pass nearby,
within 300 kilometers and two days of one another,” said Duo Chan, a graduate
student in the Harvard Graduate School of Arts and Sciences and first author of
the paper. “Using this approach, we found 17.8 million near crossings and
identified some big biases in some groups.”
The researchers focused on data from 1908 to 1941,
broken down by the country of origin of the ship and the “decks,” a term
stemming from the fact that marine observations were stored using decks of
punch cards. One deck includes observations from both Robert Falcon Scott’s and
Ernest Shackleton’s voyages to the Antarctic.
“These data have made a long journey from the
original logbooks to the modern archive and difficult choices were made to fit
the available information onto punch cards or a manageable number of magnetic
tape reels,” said Elizabeth Kent, a co-author from the UK National Oceanography
Centre. “We now have both the methods and the computer power to reveal how
those choices have affected the data, and also pick out biases due to
variations in observing practice by different nations, bringing us closer to
the real historical temperatures.”
The researchers found two new key causes of the
warming discrepancies in the North Pacific and North Atlantic.
The first had to do with changes in Japanese records.
Prior to 1932, most records of sea surface temperature from Japanese vessels in
the North Pacific came from fishing vessels. This data, spread across several
different decks, was originally recorded in whole-degrees Fahrenheit, then
converted to Celsius, and finally rounded to tenths-of-a-degree.
However, in the lead-up to World War II, more and
more Japanese readings came from naval ships. These data were stored in a
different deck and when the U.S. Air Force digitized the collection, they
truncated the data, chopping off the tenths-of-a-degree digits and recording
the information in whole-degree Celsius.
Unrecognized effects of truncation largely explain
the rapid cooling apparent in foregoing estimate of Pacific sea surface
temperatures between 1935 and 1941, said Huybers. After correcting for the bias
introduced by truncation, the warming in the Pacific is much more uniform.
While Japanese data holds the key to warming in the
Pacific in the early 20th century, it’s German data that plays the most
important role in understanding sea surface temperatures in the North Atlantic
during the same time.
In the late 1920s, German ships began providing a
majority of data in the North Atlantic. Most of these measurements are
collected in one deck, which, when compared to nearby measurements, is
significantly warmer. When adjusted, the warming in the North Atlantic becomes
more gradual.
With these adjustments, the researchers found that
rates of warming across the North Pacific and North Atlantic become much more
similar and have a warming pattern closer to what would be expected from rising
greenhouse gas concentrations. However, discrepancies still remain and the
overall rate of warming found in the measurements is still faster than
predicted by model simulations.
“Remaining mismatches highlight the importance of
continuing to explore how the climate has been radiatively forced, the
sensitivity of the climate, and its intrinsic variability. At the same time, we
need to continue combing through the data---through data science, historical
sleuthing, and a good physical understanding of the problem, I bet that
additional interesting features will be uncovered,” said Huybers.
This research was co-authored by David I. Berry from
the UK National Oceanography Centre. The research was supported by the Harvard
Global Institute, the National Science Foundation, and the Natural Environment
Research Council.
原始論文：Duo Chan, Elizabeth C. Kent, David I. Berry, Peter
Huybers. Correcting
datasets leads to more homogeneous early-twentieth-century sea surface warming. Nature, 2019; 571
(7765): 393 DOI: 10.1038/s41586-019-1349-2
引用自：Harvard John A. Paulson School of Engineering and Applied Sciences. “The
bucket list.”
沒有留言:
張貼留言