DAAC Home > Resources > News > Nature Scientific Data Publication Highlights Daymet V4 Model Improvements


Nature Scientific Data Publication Highlights Daymet V4 Model Improvements

Submitted by ORNL DAAC Staff on 2022-01-05
Image Media
Annual timeseries of daily cross-validation
Image Media

Annual timeseries of daily cross-validation mean absolute error (MAE), averaged over all station days for each year, comparing results for Daymet V4 and Daymet V3. (a) MAE for estimation of daily maximum temperature. (b) MAE for estimation of daily total precipitation. Image from Thornton et al. 2021.

A paper published this past July, 2021 in Nature's Scientific Data describes Daymet V4:  

Thornton, P.E., Shrestha, R., Thornton, M. et al. Gridded daily weather data for North America with comprehensive uncertainty quantification. Sci Data 8, 190 (2021). https://doi.org/10.1038/s41597-021-00973-0 

Daymet V4 represents several improvements from previous versions, and the data description provides complete documentation for updated methods.  

Daymet V4 is a 40-year daily meteorological dataset on a 1 km grid for North America, Hawaii, and Puerto Rico and provides temperature, precipitation, shortwave radiation, vapor pressure, snow water equivalent, and day length. The dataset includes an objective quantification of uncertainty based on strict cross-validation analysis for temperature and precipitation results. The paper describes reductions in the timing bias of input reporting weather station measurements; improvement to the three-dimensional regression model techniques in the core algorithm; and a novel approach to handling high elevation temperature measurement biases. Cross-validation analyses with the underlying weather station data demonstrates the technical validity of new dataset generation methods, and quantifies improved accuracy.

Daymet Version 4 standard products were initially released in December of 2020,

A lower latency daily product, Daymet Version 4 Monthly Latency: Daily Surface Weather Data is also available.

The complete list of available Daymet collections is available at https://daac.ornl.gov/daymet/ and access to all information for Daymet products and services is available at https://daymet.ornl.gov  .

Data for calendar year 2021 will be added to all Daymet V4 datasets by March 31, 2022.


Publication Date