Maximum Likelihood Estimates of Temperatures using Data from the Dynamically Consistent Ensemble of Temperature (Version 1.2)

Project: Maximum Likelihood Estimates of Temperatures using Data from the Dynamically Consistent Ensemble of Temperature - The primary goal of DCENT_MLE is to combine instrumental observations with physically realistic statistical models to produce maximum likelihood estimates of surface temperature anomalies and other physical quantities of the Earth. Additional goals of DCENT_MLE include correcting for biases in estimates, producing the most efficient estimates given the available data, and better quantifying uncertainties. The maximum likelihood estimation approach allows for estimated fields to be temporally and spatially complete for the entire instrumental period (since 1850) and for the entire surface of the Earth. DCENT_MLE uses source datasets primarily from the Dynamically Consistent Ensemble of Temperature project. This project has not received funding from any source.
Summary: DCENT_MLE_v1.2 is a dataset of monthly gridded surface temperatures for the Earth during the instrumental period (since 1850). The name ‘DCENT_MLE_v1.2’ reflects the dataset’s use of maximum likelihood estimation and observational data primarily from the Dynamically Consistent Ensemble of Temperature (DCENT) (Chan, Gebbie, Huybers and Kent, 2024). Source datasets used to create DCENT_MLE_v1.2 include land surface air temperatures of Chan, Gebbie and Huybers (2024), non-infilled DCLSAT, and GHCNv4; sea surface temperatures of DCSST; sea ice coverage of HadISST2; land mask data of OSTIAv2; surface elevation data of GMTED2010; and climate model output of CCSM4 for a pre-industrial control simulation. DCENT_MLE_v1.2 was generated using information from the DCENT project, the Met Office Hadley Centre, the U.S. National Oceanic and Atmospheric Administration, the E.U. Copernicus Marine Service, the U.S. Geological Survey, and the University Corporation of Atmospheric Research. Results of sensitivity tests using alternate sea ice source datasets from the Japanese Meteorological Agency (COBE-SST3) and the National Snow and Ice Data Center (modified G10010v2 appended with G02202v6) are also available.

DCENT_MLE_v1.2 uses the approach of HadCRU_MLE_v1.2 (https://doi.org/10.26050/WDCC/HadCRU_MLE_v1.2), which is described in “Improving global temperature datasets to better account for non-uniform warming” (https://doi.org/10.1002/qj.4791), but uses different source data. Additional details about DCENT_MLE_v1.2 are available in the DCENT_MLE_v1.0 information document. The primary motivation to develop HadCRU_MLE_v1.0 was to better account for spatially nonuniform warming across the planet by fitting an amplification function to observations to better account for spatially nonuniform warming trends, and by using differences in temperature climatologies and temperature anomalies between open sea and sea ice regions to better account for the impacts of changes in sea ice concentrations.

DCENT_MLE_v1.2 includes mean surface temperature anomalies for each month from 1850 to 2025 and for each 5° latitude by 5° longitude grid cell. The maximum likelihood estimation approach allows for the estimated field of surface temperature anomalies to be temporally and spatially complete for the entire instrumental period and for the entire surface of the Earth. A 5° by 5° gridded 1982-2014 temperature climatology is available, which was produced by blending an extension of the DCLSAT temperature climatology for land and sea ice regions with the DCSST temperature climatology for open sea regions. Other information of DCENT_MLE_v1.2 is available, including model parameters, the estimated amplification function, the internal variability pattern, the land area fractions, and the impacts of sea ice concentrations and the El Niño Southern Oscillation on surface temperature anomalies.

DCENT_MLE_v1.2 is an annual update to extend DCENT_MLE until the end of 2025. The median estimate of the change in global mean surface temperature change from 1850-1900 to 2025 is 1.61 °C, with a 95% confidence interval of [1.47,1.75] °C. Future versions of DCENT_MLE may become available to extend the temporal coverage beyond 2025.