In my recent post whether there is a ‘hiatus’ in global warming I left out the satellite observations of the lower troposphere. The reason for that was that the analysis of these is different from that of the near-surface series, and I considered the latter were more relevant. First, most of us live at ground level most of the time, and secondly this has traditionally been the main measure by which to gauge global warming. My conclusions were that there has been a positive trend since 1998, but no trend over the last 10 years. However, the natural variability of 10-year trends is so large that this is compatible with the positive long-term trend. The indicator of global warming with the best signal-to-noise ratio, ocean heat content, shows no sign of stopping over the last 10 years.
Guest post by Geert Jan van Oldenborgh, KNMI
In this post I consider whether the same holds for the temperature of the lower troposphere (TLT) that is measured by satellites. This is an estimate of the temperature over the lowest ten kilometres of the atmosphere, with a weighing function that emphasises the 0-4km layer. The analysis below contains nothing new, the ideas have been around for more than ten years and have been published by various people (e.g., Foster and Rahmstorf, ERL, 2011). Most of the content and links to other articles can already be found in the comments on the post on the near-surface temperature.
The annual mean time series is shown in Fig. 1. It starts at the beginning of the modern satellite era, in 1979. This is in stark contrast to the global mean near-surface temperature that can be reasonably accurately reconstructed to the end of the 19th century. The short record is one of the reasons it is not used much in climate science.
The global mean TLT differs from the global mean near-surface temperature in a few key aspects. One of them is that the influence of El Niño is much larger, as can be seen from, e.g., the height of the peak in 1998, which is about 0.4 K above the trend line, against about 0.2 K in the near-surface temperature. Note that strongest effects of El Niño on temperature lag the event itself by about half a year. The 2010 peak is also higher, about 0.2 K versus only 0.1 K. On a map, this can be seen as a stronger and broader response to El Niño and La Niña in the tropics, see Fig. 2. The amplification can easily be understood due to the stronger warming at height caused by the heat of condensation of the higher rainfall. This is the upper tropospheric warming that accompanies an increase in SST in the tropics. In the deep tropics, near the equator, this causes heating well above the lower troposphere and hence is not clearly visible in the TLT, but away from the equator the warmer air descends in the Hadley circulation and enters the heights to which the TLT is more sensitive.
Because the effects of El Niño are stronger in the TLT than in the near-surface temperature, the year 1998, following the strongest El Niño on record, is indeed still the warmest year in the series. Trends starting on this exceptional year naturally are lower than the long-term trend, but not zero at 0.05 K/decade. The trend from 1999 would be equally zero if global warming would have stopped in 1998, but it is 0.12 K/decade. Both are within the uncertainties equal to the long-term trend of 0.13 K/decade.
The trend is better visible if we subtract the effects of El Niño, as many people have shown already. Subtracting 0.11 times the Niño3.4 index lagged by the same half year I obtain the series shown in Fig. 3. The El Niño of 1998 has been much suppressed, but is still visible, so the subtraction factor (determined by linear regression) is a bit too small for the biggest events. The biggest signal now is the cooling due to the eruption of Mount Pinatubo visible in 1992 and 1993. Also visible is the cooling due to El Chichon after it erupted in 1982. Apart from that this series shows a steady increase with natural variability around it, with no pause in heating. Only short trends up to 2014 are negative, starting with the 6-yr trend 2009-2014. These short trends have uncertainty margins that make them compatible with the long-term trend.
Another difference between the TLT and near-surface temperature occurs in the subtropics in the areas near the eastern coasts of ocean, where very low clouds form over the cold waters. In these areas the surface temperature is decoupled from the temperature of the air above it by these clouds. These areas are visible as white regions in the plot of the regression of TLT on near-surface temperature, Figure 4.
A third large difference between near-surface air temperature and the TLT occurs in the high latitudes. Cold air in winter there is often a thin layer on the ground, an inversion. Under these circumstances the thermometers at 1.5m height and the satellite that averages over a large part of the atmosphere naturally give quite different numbers. The variations near the ground are then also much larger than the variations higher up in the atmosphere. This is visible in the annual mean as the lighter areas over the high latitudes in Figure 4, especially Siberia. This effect implies that the Arctic amplification, the stronger trend in temperature in high latitudes, is not as strong in the TLT as it is in near-surface observations. Most of the warming there has been in winter and very low in the atmosphere.
In the global mean trend difference between the ground and the lower troposphere, the effects of the upper tropospheric warming and the vertical structure of the Arctic amplification cancel to a large extent, so that the final series look similar, but this hides the differences in the (sub)tropics and high latitudes.
I conclude that beyond the effects of El Niño there is no pause in the warming in TLT either. As there is no evidence for a long-term trend in the Niño3.4 index, this means it will revert to normal before long and by that time the trend will pick up in the full TLT series as well. I think that may well be next year (2015), due to the (very small) El Niño this winter, but that may be overwhelmed by other natural variability. The TLT series looks so similar to the near-surface temperature series only because of a cancellation between the differences in the tropics and the high latitudes. However, neither shows that global warming stopped.