Investigating Land-Atmosphere Coupling Strength Biases in CMIP5 Climate Models Using Remote Sensing Data
Abstract
In the development of global climate models, a key approach is to contrast model simulations with observations, especially the remote sensing data for global analyses. However, remote sensing data are typically subject to instrumental issues, signal noises, and retrieval uncertainties which degrade their values as a point of reference for model simulations. Here, we describe a robust, global triple collocation analysis that eliminates the impact of observational error in remotely sensed data products on correlation-based coupling strength metrics and thereby provides an unbiased global map of land-atmosphere coupling strength between multiple surface and near-surface variables (i.e., soil moisture SM, evapotranspiration ET and air temperature Ta). These derived unbiased coupling strength estimates are then applied to benchmark coupled land-atmosphere general circulation models (GCMs) within the Coupled Model Intercomparison Project 5 (CMIP5). Results demonstrate significant bias in CMIP5 GCM's representation of land-atmosphere coupling strengths. These biases are then compared against the model simulated climate (e.g., precipitation and air temperature) in these GCMs. Preliminary results show linkages between the coupling strength biases and the simulated precipitation differences. Specially, the over-coupling of SM/ET relationship in GCMs are correlated with model under-predicted precipitation. While initial results are based on a (linear) correlation-based quantification of coupling strength, prospects for generalizing the approach using a non-parametric, information-based approach will also be discussed.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.H53F..03L
- Keywords:
-
- 1833 Hydroclimatology;
- HYDROLOGY;
- 1840 Hydrometeorology;
- HYDROLOGY;
- 1843 Land/atmosphere interactions;
- HYDROLOGY;
- 1866 Soil moisture;
- HYDROLOGY