Journal
JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES
Volume 123, Issue 7, Pages 3483-3496Publisher
AMER GEOPHYSICAL UNION
DOI: 10.1002/2017JD027577
Keywords
drought; climate models; uncertainty; drought persistence; drought length bias; drought persistence bias
Categories
Funding
- ERC DROUGHT-HEAT project [617518]
Ask authors/readers for more resources
The persistence of drought events largely determines the severity of socioeconomic and ecological impacts, but the capability of current global climate models (GCMs) to simulate such events is subject to large uncertainties. In this study, the representation of drought persistence in GCMs is assessed by comparing state-of-the-art GCM model simulations to observation-based data sets. For doing so, we consider dry-to-dry transition probabilities at monthly and annual scales as estimates for drought persistence, where a dry status is defined as negative precipitation anomaly. Though there is a substantial spread in the drought persistence bias, most of the simulations show systematic underestimation of drought persistence at global scale. Subsequently, we analyzed to which degree (i) inaccurate observations, (ii) differences among models, (iii) internal climate variability, and (iv) uncertainty of the employed statistical methods contribute to the spread in drought persistence errors using an analysis of variance approach. The results show that at monthly scale, model uncertainty and observational uncertainty dominate, while the contribution from internal variability is small in most cases. At annual scale, the spread of the drought persistence error is dominated by the statistical estimation error of drought persistence, indicating that the partitioning of the error is impaired by the limited number of considered time steps. These findings reveal systematic errors in the representation of drought persistence in current GCMs and suggest directions for further model improvement.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available