MSG Satellite in orbit

A Meteosat look at budgets

January 2004-October 2020

MSG Satellite in orbit
MSG Satellite in orbit

Meteosat Second Generation (MSG) spans 18 years of an atmosphere in evolution.

Last Updated

05 February 2021

Published on

16 December 2020

By José Prieto, Vesa Nietosvaara and Rob Roebeling (EUMETSAT) and Ivan Smiljanic (CGI)

The Earth manages its temperature, and keeps it roughly constant, by emitting thermal radiation to space and absorbing solar radiation in the atmosphere and Earth surfaces. Since pre-industrial times, surface and air temperatures have increased, faster since the 1970s. The observed warming of the Earth is attributed to the imbalance in incoming shortwave and outgoing longwave radiative flows of energy, caused by anthropogenic emissions (Stocker et al. 2013). These emissions change concentrations of well-mixed greenhouse gases, as CO2 and CH4, leading to a direct, or indirect, impact on the energy balance of the Earth, and driving climate change. As CO2 becomes more abundant in the atmosphere, weaker radiation is expected in CO2-absorption satellite channels (around 15 µm). At the same time more radiation is expected at the infrared (IR) satellite window channels (e.g. around 10 and 12 µm), due to warmer surfaces or lower cloudiness. Changes in the amount of solar reflection (shortwave outgoing radiation, SOR) result from the feedback of clouds or snow, and will affect the Earth’s radiative flows of energy contributing to a new balance. These changes, however, are more difficult to quantify, as the IPCC states in its IPCC in its Fourth Assessment Report: "The cloud feedback is likely positive but its quantification remains difficult".

The scheme in Figure 1 (right panel) shows the individual energy fluxes of the Global Mean Radiation Budget. Among others, satellite observations are particularly well suited to monitor changes in the atmospheric radiation budget at the Top of the Atmosphere (TOA), a sum of the Outgoing Longwave Radiation (OLR) and the Outgoing Shortwave Radiation (OSR). IPCC (2007) reports that, during the 1980s and 1990s, the OLR decreased (~0.7 W/m2) and OSR increased (~2.1 W/m2) over the tropics. During the first decade of the 21st century no noticeable trends in either the tropical or global radiation budget were observed (Hartmann et al., 2013).

ariation in the solar and infrared radiances
Figure 1: Variation in the solar and infrared radiances between 2005 and 2018 for different regions of the Meteosat field of view (as defined in Figure 2). The vertical bars show the errors in the derivation of the values for one of the areas ('Mediterranean'). Similar errors for other areas. On the right hand side, a conceptual depiction of the solar and infrared fluxes sampled by Meteosat.

We try here an approximation based on radiances by using four shortwave (or solar, including 3.9 µm where sun energy predominates during the day) and four longwave (infrared) channels of the SEVIRI instrument on board of Meteosat Second Generation (MSG) satellites, and look at changes in shortwave and longwave radiation flows over the observation period 2004 to 2020. To do so, we compared changes in average channel radiances (Figure 2) between the first four years (2004-2007) and the last four years (2016-2020) of the observation period for five geographical regions, including land and ocean surfaces over Africa and Europe.

The results show (Figure 1, left panel) that the Outgoing Longwave Radiation (OLR) has increased in most Meteosat longwave channels between 1% and 3%, except at 13.4 µm, with a decrease around 3%. In turn, the shortwave (solar) reflected radiation (Outgoing Shortwave Radiation, OSR) has decreased by 2-3%. Since the decrease in solar reflected radiation is strong in the southern Atlantic, without land or known changes in aerosol, a cloud cover reduction is suggested.

Figure 2 specifies the values for regions in the Meteosat-0° field of view. Variability can be the result of different soils, cloud frequency and ocean influence in the areas. For instance, the increase of sea surface temperature (SST), related to 10.8 µm, is moderate in the southern Atlantic compared with land scenes, with a higher increase in IR values. The area of southern Africa and Mozambique (yellow square) tends to lose energy to space in the balance. The tropical area south of Lake Victoria hardly changed in 13 years.

Region limits 2004-2020
Figure 2: Region limits for the 2004-2020 evolution analysis. OSR, OLR and total changes in radiance sensed by Meteosat satellites. Red in the third column indicates less energy to space in response to CO2 and possibly other terms, like reduced reflection of the incident solar radiation. Blue indicates an increase in energy to space. The averages are similar within the error term (0.15 W.m-2).
Change in the course of the year (January to December on the horizontal axis) for brightness temperatures measured by Meteosat in the periods 2004-2007
Figure 3: Change in the course of the year (January to December on the horizontal axis) for brightness temperatures measured by Meteosat in the periods 2004-2007 (brown lines) and Nov 2016-Oct 2020 (blue) around the Mediterranean. Note the different spacing (in kelvin) along the vertical axis for the three channels considered (click to expand).

On Figure 3, the average year cycle is presented for the early and the late periods, to illustrate natural variability. Oceans fill 71% of the Earth surface, so its analysis in the future (for instance with the IASI interferometer on board of Metop polar satellites) can greatly clarify the amount of warming in oceans.

The global ERA-5 atmospheric reanalysis was used in this study to subjectively compare the changes seen in satellite measurements and, respectively, the changes in model 12:00 UTC mean 2-metre-temperatures during the same periods of time.

In Figure 4 (right) we can see that most of the areas within the Meteosat disc area were slightly (in the order of 0.5 K) warmer towards 2020 compared to the period 2004-2007. Over Europe, and many parts in Africa, the difference is more pronounced. However, some areas, such as the northern Atlantic, show negative difference. The areas with the highest increases in air temperature, e.g. the Mediterranean area or southern Africa, bear resemblance to those with the highest increase in IR radiative flux.

Surface air temperature at 12:00 UTC


Figure 4: Difference in mean 12:00 UTC 2-metre-temperature between periods 2016-2019 and 2004-2007 (lighter hues in image). Average 2-metre-temperature in the period 2016-2019 (deeper red hues in image). Data from ECMWF ERA5 reanalysis.

The main source of radiation in the infrared window channels (8.7 µm and 10.8 µm in this case) is the skin temperature, which is sensitive to the surface type. Daytime skin temperature and 2-metre-temperature difference is the largest over the deserts, and smallest over vegetated land and oceans.

The warmest pixel value comparison between the two subperiods (2004-2007 and 2016-2020) (Figure 5) shows the specific areas of extreme heat in the Mediterranean area.

Change areas in the Mediterranean

Infrared 10.8 µm, 2016-2020 compare1

Figure 5: Maximum pixel values for Meteosat 10.8 µm imagery in the two periods mentioned in the text. Both four-year periods are affected by heatwaves, like the one over France in 2019. The colour scale goes from 25 to 45 °C. Surface Mediterranean waters show similar temperatures, but the patterns show warmer pixels towards the Adriatic in the second period 2016-2020.