Increases in global temperature over recent decades and the projected acceleration in warming trends over the 21 century have resulted in a strong need to obtain information about future temperature conditions. Hence, skillful decadal temperature predictions (DTPs) can have profound societal and economic benefits through informed planning and response. However, skillful and actionable DTPs are extremely challenging to achieve. Even though general circulation models (GCMs) provide decadal predictions of different climate variables, the direct use of GCM data for regional-scale impacts assessment is not encouraged because of the limited skill they possibly exhibit and their coarse spatial resolution. Here, we focus on 14 GCMs and evaluate seasonally and regionally averaged skills in DTPs over the continental United States. Moreover, we address the limitations in skill and spatial resolution in the GCM outputs using two data-driven approaches: (1) quantile-based bias correction and (2) linear regression-based statistical downscaling. For both the approaches, statistical parameters/relationships, established over the calibration period (1961–1990) are applied to retrospective and near future decadal predictions by GCMs to obtain DTPs at ‘4 km’ resolution. Predictions are assessed using different evaluation metrics, long-term statistical properties, and uncertainty in terms of the range of predictions. Both the approaches adopted here show improvements with respect to the raw GCM data, particularly in terms of the long-term statistical properties and uncertainty, irrespective of lead time. The outcome of the study is monthly DTPs from 14 GCMs with a spatial resolution of 4 km, which can be used as a key input for impacts assessments.
All Science Journal Classification (ASJC) codes
- Atmospheric Science
- Bias correction
- Continental United States
- Decadal temperature predictions
- Statistical downscaling