Impact experiments which ECMWF has carried out for the UN’s Systematic Observations Financing Facility (SOFF) have shown that weather forecasts would benefit significantly from more in-situ observations in several countries.
SOFF is a United Nations initiative which provides grant funding for the collection of weather and climate data to strengthen climate adaptation and resilient development.
It targets countries with the most severe shortfalls in observations, particularly Least Developed Countries and Small Island Developing States.
“Without data, there is no forecast. The results of the ECMWF impact experiments provide the strongest scientific evidence to date that SOFF investments to close the blind spots in the global observing system dramatically improve forecast accuracy, both locally and globally,” said Celeste Saulo, Secretary-General of the World Meteorological Organization (WMO).
Eight scenarios
ECMWF has designed and run eight tailored scenarios simulating the impact of more surface, upper-air and marine observations in a range of countries.
These observations were simulated as coming from an expanded Global Basic Observing Network (GBON), which is a fundamental element of the WMO Integrated Global Observing System (WIGOS).
The countries chosen were Least Developed Countries (LDCs), Small Island Developing States (SIDS), and Lower Middle-Income Countries (LMICs). Regional gaps, such as around Africa and in the Pacific, were also covered.
Spatial coverage of simulated surface pressure measurements used in ECMWF’s experiments on 1 June 2023.
Spatial coverage of simulated radiosonde observations used in ECMWF’s experiments on 1 June 2023.
Simulated observations were generated from ECMWF’s operational initial conditions of forecasts, also called the analysis. It has been demonstrated that the statistical characteristics of simulated data generated in this way are similar to those of real observations.
These additional, simulated observations were used to help determine an updated analysis in a process called data assimilation.
In the eight scenarios, different combinations of simulated observations were used.
Results
The study shows a clear reduction in the uncertainty of the analysis, and thus of forecasts. This is directly related to the number of simulated observations used: the more observations are added, the greater the reduction in uncertainty.
That reduction is largest where the density of new, simulated observations is greatest. This means that the biggest observed improvements are over land.
As an example, consider simulated surface and upper-air observations for Least Developed Countries (LDC) and Small Island Developing States (SIDS). The following map shows the percentage change in surface pressure uncertainty in the analysis when using these simulated observations.
Percentage change in the surface pressure analysis uncertainty when compared with the control experiment for 1 to 30 June 2023. Negative values (blue shading) indicate regions where the surface pressure analysis uncertainty is improved when compared to the control experiment. The diagonal lines superimposed on the shading indicate regions where the improvement is statistically significant at the 95% level.
The contribution of simulated surface observations on the one hand and simulated upper-air observations on the other was also investigated.
For example, one experiment compared the change in surface pressure uncertainty when both kinds of simulated observations were used and when only upper-air simulated observations were used.
The result showed that simulated surface observations make a substantial contribution to the overall effect of observations.
Percentage change in the surface pressure analysis uncertainty when comparing an experiment with simulated surface and upper-air observations with just using simulated upper-air observations, for 1 to 30 June 2023. Negative values (blue shading) indicate regions where the surface pressure analysis uncertainty is improved when both observation types are used. The diagonal lines superimposed on the shading indicate regions where the improvement is statistically significant at the 95% level.
Other experiments confirmed the positive impact of upper-air observations on, for example, short-range temperature forecasts from near the surface to around 35 km above the surface.
Improvements in the initial conditions of forecasts lead to better forecasts, too. The improvements also spread geographically in the forecast as weather systems move around the globe.
Florian Pappenberger, Director of Forecasts and Services and Deputy Director-General at ECMWF, said: “Better forecasts can benefit the people in the country and beyond. Not only can this feed into better early warning systems and save lives and livelihoods, but global weather and climate models benefit from this data in parallel, improving forecasts around the world.”
Next steps
The SOFF Steering Committee has noted the evidence of local and global improvements in forecast accuracy as a result of investments in GBON infrastructure in under-observed regions.
It has also requested the SOFF Secretariat to prepare a proposal for more work on the importance of SOFF-supported data for artificial intelligence forecasting.
In addition, ECMWF and the World Bank are partnering to investigate how improved weather forecasts can enhance socio-economic resilience to natural disasters, using the World Bank’s Unbreakable model and ECMWF’s simulations of SOFF’s investment scenarios.
The ongoing study starts by estimating the potential benefits in terms of avoided asset losses. It then converts avoided asset damages into a well-being metric, reflecting people’s individual ability to cope with disaster impacts.
With this study, the World Bank aims to provide a more comprehensive view on the socio-economic benefits of forecast skill improvements and a denser weather observation network.
“This allows us to compare the costs of additional observations to their expected benefits, and we can prioritise different scenarios based on their effectiveness,” explains Paolo Avner, Senior Economist at the World Bank’s Global Facility for Disaster Reduction and Recovery (GFDRR). “The interim findings underscore that investments in better forecasts strengthen people’s resilience against natural disasters.”
Further resources
A summary of ECMWF’s impact experiments for SOFF is included in this SOFF document.
An article about ECMWF's experiments has also been published on the SOFF website.