New Paper on Marine Air Temperature Biases

A new paper from National Oceanography Centre researchers “Quantifying Daytime Heating Biases in Marine Air Temperature Observations from Ships” has been published in the Journal of Atmospheric and Oceanic Technology. The paper, produced as part of the GloSAT project, outlines the technical framework under which previously unused marine air temperature (MAT) can be used in the creation of global climate datasets.

The primary objective of the GloSAT project is to produce the first dataset combining MAT, with air temperatures recorded over ice and land. Air temperatures from across the global oceans have been recorded as far back as 1699,  but MAT has not previously been used in global temperature datasets, with sea-surface temperatures (SST) generally preferred by all major data-producing centres (e.g. NASA GISS, Berkeley, HadCRUT, NOAA).

The main reason MAT has not previously been used is due to the problem of daytime heating bias. This occurs because ships absorb incoming solar radiation throughout the day, raising the temperature of the ship (and the surrounding air) above the ambient air temperature. This means that air temperature measurements taken on-board ships do not reflect the true air temperature over the ocean. The recorded MAT cannot, therefore, be used in the construction of accurate climate datasets without adjustment for these heating effects.

An additional issue with daytime heating bias is that it can vary widely between individual ships - imagine the difference between a 21st century cargo ship and a 19th century sailing vessel. This new paper explores how the expected warm bias can adequately be removed from different types of ship, estimating it using a physics-driven approach. The figure below gives an example of adjusted temperatures.


Measured air temperature (black), background night-time marine temperature (red) and air temperature following adjustment (blue) for the Raphael recorded in October 1884


The ability to use MAT instead of SST in a combined dataset potentially enables the extension of the surface temperature record back as far as the late 18th century. This would add 60 years to the current record which is limited by the low number of SST observations before 1850.

The spatial coverage of the extended air temperature record from the 1790s until the 1850s will mostly be limited to the Atlantic and Indian Oceans, based on the historic trading routes of that time. However, by combining marine data with newly updated sources of air temperature over land and comparison against new global climate model outputs, scientists will be able to study the climate during a period of limited anthropogenic climate forcing.