Artificial Intelligence improves vital global climate monitoring dataset
How do we measure the temperature of the Earth? There are actually two different kinds of measurements needed because Earth is made up of land and water. Temperature on land is recorded by weather stations all over the world. Ocean temperatures are measured by on-site instruments (in situ) like buoys, ships, and uncrewed vehicles. When combined together, the data provide a full picture of the Earth’s temperature, including change over time.
NOAAGlobalTemp is the authoritative dataset used to assess observed global climate change. NOAAGlobalTemp combines long-term sea surface (water) temperature (SST) and land surface (air) temperature datasets to create a complete, accurate depiction of global temperature trends and to identify temperature anomalies (different-from-average temperatures).
NCEI switched to an updated version of NOAAGlobalTemp today with the release of the January 2024 Global Climate Report. For the new version, NCEI scientists created an artificial neural network (ANN) method that replaces the traditional empirical orthogonal teleconnection (EOT) approach for the surface air temperatures over land and the Arctic Ocean. This new method improves the accuracy of surface air temperature reconstruction. Improvements were larger in the Southern Hemisphere—especially Antarctica—and larger before the 1950s, which is directly associated with the availability of observations.
The ANN approach outperforms the EOT method, particularly in the observation-sparse areas, which can be illustrated by an example for Antarctica (figure below). The three panels in the top row show the number of ingested observation data in the lower left corner increased from day 4, to 8, and then to 16. The observations of the last day indicate a cold anomaly in this area. In the beginning when there were no observations in the area due to data delay, while the EOT method (middle row) failed to reconstruct the temperature, the ANN method (bottom row) successfully caught the cold anomaly. This result exemplifies the robustness of the ANN approach, which works reliably even in observation-sparse areas.
Supporting Global Climate Science
NOAAGlobalTemp consists of land surface air temperature (LSAT) records from the Global Historical Climatology Network-Monthly, and sea surface temperatures (SST) from the Extended Reconstructed SST, the International Comprehensive Ocean-Atmosphere Data Set (ICOADS), and the International Arctic Buoy Programme (IABP). It has data from 1850–Present and is presented on a 5X5 grid. NOAAGlobalTemp is a key component of the Global Climate Report which is updated monthly. The global section of the Climate at a Glance tool was updated in mid-February 2024 to use the new version of the dataset.
NOAAGlobalTemp has been used by multiple science organizations such as the World Meteorological Organization and in assessments, such as the Intergovernmental Panel on Climate Change and the Bulletin of the American Meteorological Society (BAMS) State of the Climate reports. Private sector interests use the data for global climate monitoring and assessment, environmental research, and informational products and services for various industries and economic sectors, such as agriculture.
Supporting NOAA’s AI Strategic Vision
NCEI plays a critical role in each of NOAA’s strategic goals by maintaining the most comprehensive public archive of environmental data in the United States and equitably distributing scientific products that drive decision-making across sectors, supporting the new blue economy and climate-informed strategies. By improving NOAAGlobalTemp using an Artificial Neural Network system, NCEI is supporting NOAA’s Artificial Intelligence Strategy. To learn more about how NOAA scientists are using artificial intelligence to better understand the Earth’s environment, see the NOAA Center for Artificial Intelligence (NCAI).
Reference: Huang, B., X. Yin, M. J. Menne, R. Vose, and H. Zhang, 2022: Improvements to the Land Surface Air Temperature Reconstruction in NOAAGlobalTemp: An Artificial Neural Network Approach. Artif. Intell. Earth Syst., 1, e220032, https://doi.org/10.1175/AIES-D-22-0032.1.