NOAA satellite launch in 1998 opened new era

Niels Bormann, Stephen English, Anthony McNally


The meteorological satellite sounding community recently celebrated the 20th anniversary of the start of a new era in meteorological satellite observations: on 13 May 1998, the US NOAA-15 satellite was launched, beginning the Advanced TIROS Operational Vertical Sounder (ATOVS) series of satellite observations that has been hugely influential for numerical weather prediction (NWP) and continues to this day.

ATOVS was the first major enhancement in polar satellite instrument observing capability since the first operational system was launched 20 years earlier in 1978. For the first time, the microwave sounding capability provided information not just on temperature but also on humidity, and the vertical resolution of the temperature information was significantly improved. Previously the infrared sounder was the primary sounding instrument, but this was restricted to cloud-free areas. The new microwave sounders provided data of comparable quality, but in clear and cloudy regions. ATOVS suites were subsequently launched on many other satellites, including NOAA-16 through to 19 and the European Metop satellites, and they inspired similar instruments on other US, Chinese, and Russian satellites.

Making the most of the data

ATOVS was launched shortly after the introduction of 4D-Var at ECMWF in November 1997. One of the motivations for developing 4D-Var was that it makes better use of satellite radiance data. ATOVS in many ways showcased the value of this strategic development. A year after launch, on 5 May 1999, ECMWF started the direct assimilation of these new observations as raw radiances, resulting in very significant improvements in forecast skill.

The start of assimilating ATOVS data was also the start of learning how to make best use of these observations. We can now treat random and systematic uncertainties better and have extended the data usage over land. The development of the all-sky assimilation approach made it possible to better extract dynamical information from the humidity and cloud information contained in the data. All this means that we are now extracting more information from these observations than when they first arrived 20 years ago, ensuring that ATOVS data remain leading contributors to forecast skill.

Improvements in forecasts and reanalyses as seen by NOAA-15. The plots show standard deviations of differences between observations and short-range forecasts from the high-resolution operational system (top) and from the ERA5 reanalysis (bottom) for radiances from channel 8 of AMSU-A on the NOAA-15 satellite. This is a microwave temperature-sounding channel peaking around the tropopause. The standard deviations show a combination of errors related to the observations as well as the short-range forecasts. One main contributor from the observations is the instrument noise, which independent monitoring shows has been stable at around 0.13 K throughout the entire period. The gradual reduction in standard deviation reflects improvements in the short-range forecasts and also improved processing of the observations in the ECMWF system.

Benefits of a long life span

NOAA-15 not only carries the first in a major new generation of instruments but has also proven one of the most durable satellites. Indeed, ECMWF is still assimilating data from NOAA-15 20 years on – a remarkable feat given the satellite was originally planned to have a life span of only five years. The observations give us a unique view of the evolution of ECMWF’s operational system, not least owing to the observations’ stable noise characteristics. Comparisons between observations and model equivalents calculated from short-range forecasts show a gradual reduction in standard deviation, reflecting the various improvements in ECMWF’s Integrated Forecasting System over the years (not only in terms of the use of the NOAA-15 observations, but also in terms of other aspects of the forecasting system). Similarly, our reanalysis activities benefit immensely from this longevity and stability. Since in reanalyses the assimilation system is fixed, standard deviations of differences between observations and short-range forecast equivalents are a lot more stable over time, with only a small reduction due to general improvements of the observing system in the last two decades. The ATOVS suite is one of the backbones of any satellite-era reanalysis.

Still a crucial observing system

Even today, the microwave sensors of the ATOVS suite continue to be leading contributors to forecast skill in ECMWF’s operational assimilation system. At the time of writing, ECMWF assimilates data from six microwave temperature sounders and four humidity sounders of the ATOVS family from several satellites. We expect to add one further instrument for each type later this year after EUMETSAT’s Metop-C is launched, carrying the last flight models of this successful suite of instruments. Since ATOVS there has been an acceleration of new satellite capabilities, for example the hyperspectral infrared sounders with much better vertical resolution, and refined ATOVS-inspired microwave sensors on US, Chinese and Russian satellites.

Twenty years ago, the use of satellite data in NWP stepped up a gear through the technological advances provided by ATOVS. This is certainly a reason to celebrate. Happy birthday NOAA-15 – and happy birthday ATOVS!