Keywords
ensemble-based projections; prediction accuracy; Earth System Models
Start Date
5-7-2022 12:00 PM
End Date
8-7-2022 10:00 AM
Abstract
The outputs of Earth System Models (ESM) enable us to evaluate the scale of changes in climatic and environmental conditions all over the world under various scenarios of fossil-fuel emissions. For some regions, the model-to-model discrepancy in projected changes of the variables characterizing climatic and environmental conditions could be too large. The uncertainty associated with model-to-model discrepancy is reduced by combining model projections. The computational method developed for this purpose is based on the idea that simulated values of a climatic or environmental variable could be considered as the predictors for the observed values of this variable. A linear regression model predicting observed values from simulated values makes it possible to evaluate prediction accuracy using a standard statistical technique. The prediction accuracy is improved through combining model projections, that is, by using a weighted sum of the values simulated by different models as predictors of the observed values. The weights are selected to provide the best fit to the observed data, they are positive, not exceeding 1, and their sum is equal to 1. The efficiency of this method is illustrated with a case study of the median value of the mean annual air temperature over the northern part of Western Siberia and some other permafrost regions.
A computational method for combining ESM projections
The outputs of Earth System Models (ESM) enable us to evaluate the scale of changes in climatic and environmental conditions all over the world under various scenarios of fossil-fuel emissions. For some regions, the model-to-model discrepancy in projected changes of the variables characterizing climatic and environmental conditions could be too large. The uncertainty associated with model-to-model discrepancy is reduced by combining model projections. The computational method developed for this purpose is based on the idea that simulated values of a climatic or environmental variable could be considered as the predictors for the observed values of this variable. A linear regression model predicting observed values from simulated values makes it possible to evaluate prediction accuracy using a standard statistical technique. The prediction accuracy is improved through combining model projections, that is, by using a weighted sum of the values simulated by different models as predictors of the observed values. The weights are selected to provide the best fit to the observed data, they are positive, not exceeding 1, and their sum is equal to 1. The efficiency of this method is illustrated with a case study of the median value of the mean annual air temperature over the northern part of Western Siberia and some other permafrost regions.
Stream and Session
false