skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Questions of bias in climate models

Abstract

The recent work by Shindell usefully contributes to the debate over estimating climate sensitivity by highlighting an important aspect of the climate system: that climate forcings that occur over land result in a more rapid temperature response than forcings that are distributed more uniformly over the globe. While, as noted in this work, simple climate models may be biased by assuming the same temperature response for all forcing agents, the implication that the MAGICC model is biased in this way is not correct.

Authors:
; ; ;
Publication Date:
Research Org.:
Pacific Northwest National Lab. (PNNL), Richland, WA (United States)
Sponsoring Org.:
USDOE
OSTI Identifier:
1170094
Report Number(s):
PNNL-SA-101590
400408000
DOE Contract Number:
AC05-76RL01830
Resource Type:
Journal Article
Resource Relation:
Journal Name: Nature Climate Change, 4:741-742
Country of Publication:
United States
Language:
English

Citation Formats

Smith, Steven J., Wigley, Tom M., Meinshausen, Malte, and Rogelj, Joeri. Questions of bias in climate models. United States: N. p., 2014. Web. doi:10.1038/nclimate2345.
Smith, Steven J., Wigley, Tom M., Meinshausen, Malte, & Rogelj, Joeri. Questions of bias in climate models. United States. doi:10.1038/nclimate2345.
Smith, Steven J., Wigley, Tom M., Meinshausen, Malte, and Rogelj, Joeri. Wed . "Questions of bias in climate models". United States. doi:10.1038/nclimate2345.
@article{osti_1170094,
title = {Questions of bias in climate models},
author = {Smith, Steven J. and Wigley, Tom M. and Meinshausen, Malte and Rogelj, Joeri},
abstractNote = {The recent work by Shindell usefully contributes to the debate over estimating climate sensitivity by highlighting an important aspect of the climate system: that climate forcings that occur over land result in a more rapid temperature response than forcings that are distributed more uniformly over the globe. While, as noted in this work, simple climate models may be biased by assuming the same temperature response for all forcing agents, the implication that the MAGICC model is biased in this way is not correct.},
doi = {10.1038/nclimate2345},
journal = {Nature Climate Change, 4:741-742},
number = ,
volume = ,
place = {United States},
year = {Wed Aug 27 00:00:00 EDT 2014},
month = {Wed Aug 27 00:00:00 EDT 2014}
}
  • Here, global climate models often underestimate aerosol loadings in China, and these biases can have significant implications for anthropogenic aerosol radiative forcing and climate effects. The biases may be caused by either the emission inventory or the treatment of aerosol processes in the models, or both, but so far no consensus has been reached. In this study, a relatively new emission inventory based on energy statistics and technology, Multi-resolution Emission Inventory for China (MEIC), is used to drive the Community Atmosphere Model version 5 (CAM5) to evaluate aerosol distribution and radiative effects against observations in China. The model results aremore » compared with the model simulations with the widely used Intergovernmental Panel on Climate Change Fifth Assessment Report (IPCC AR5) emission inventory. We find that the new MEIC emission improves the aerosol optical depth (AOD) simulations in eastern China and explains 22–28 % of the AOD low bias simulated with the AR5 emission. However, AOD is still biased low in eastern China. Seasonal variation of the MEIC emission leads to a better agreement with the observed seasonal variation of primary aerosols than the AR5 emission, but the concentrations are still underestimated. This implies that the atmospheric loadings of primary aerosols are closely related to the emission, which may still be underestimated over eastern China. In contrast, the seasonal variations of secondary aerosols depend more on aerosol processes (e.g., gas- and aqueous-phase production from precursor gases) that are associated with meteorological conditions and to a lesser extent on the emission. It indicates that the emissions of precursor gases for the secondary aerosols alone cannot explain the low bias in the model. Aerosol secondary production processes in CAM5 should also be revisited. The simulation using MEIC estimates the annual-average aerosol direct radiative effects (ADREs) at the top of the atmosphere (TOA), at the surface, and in the atmosphere to be –5.02, –18.47, and 13.45 W m –2, respectively, over eastern China, which are enhanced by –0.91, –3.48, and 2.57 W m –2 compared with the AR5 emission. The differences of ADREs by using MEIC and AR5 emissions are larger than the decadal changes of the modeled ADREs, indicating the uncertainty of the emission inventories. This study highlights the importance of improving both the emission and aerosol secondary production processes in modeling the atmospheric aerosols and their radiative effects. Yet, if the estimations of MEIC emissions in trace gases do not suffer similar biases to those in the AOD, our findings will help affirm a fundamental error in the conversion from precursor gases to secondary aerosols as hinted in other recent studies following different approaches.« less
  • Regional climate change impact (CCI) studies have widely involved downscaling and bias-correcting (BC) Global Climate Model (GCM)-projected climate for driving land surface models. However, BC may cause uncertainties in projecting hydrologic and biogeochemical responses to future climate due to the impaired spatiotemporal covariance of climate variables and a breakdown of physical conservation principles. Here we quantify the impact of BC on simulated climate-driven changes in water variables(evapotranspiration, ET; runoff; snow water equivalent, SWE; and water demand for irrigation), crop yield, biogenic volatile organic compounds (BVOC), nitric oxide (NO) emissions, and dissolved inorganic nitrogen (DIN) export over the Pacific Northwest (PNW)more » Region. We also quantify the impacts on net primary production (NPP) over a small watershed in the region (HJ Andrews). Simulation results from the coupled ECHAM5/MPI-OM model with A1B emission scenario were firstly dynamically downscaled to 12 km resolutions with WRF model. Then a quantile mapping based statistical downscaling model was used to downscale them into 1/16th degree resolution daily climate data over historical and future periods. Two series climate data were generated according to the option of bias-correction (i.e. with bias-correction (BC) and without bias-correction, NBC). Impact models were then applied to estimate hydrologic and biogeochemical responses to both BC and NBC meteorological datasets. These im20 pact models include a macro-scale hydrologic model (VIC), a coupled cropping system model (VIC-CropSyst), an ecohydrologic model (RHESSys), a biogenic emissions model (MEGAN), and a nutrient export model (Global-NEWS). Results demonstrate that the BC and NBC climate data provide consistent estimates of the climate-driven changes in water fluxes (ET, runoff, and water demand), VOCs (isoprene and monoterpenes) and NO emissions, mean crop yield, and river DIN export over the PNW domain. However, significant differences rise from projected SWE, crop yield from dry lands, and HJ Andrews’s ET between BC and NBC data. Even though BC post-processing has no significant impacts on most of the studied variables when taking PNW as a whole, their effects have large spatial variations and some local areas are substantially influenced. In addition, there are months during which BC and NBC post-processing produces significant differences in projected changes, such as summer runoff. Factor-controlled simulations indicate that BC post-processing of precipitation and temperature both substantially contribute to these differences at region scales. We conclude that there are trade-offs between using BC climate data for offline CCI studies vs. direct modeled climate data. These trade-offs should be considered when designing integrated modeling frameworks for specific applications; e.g., BC may be more important when considering impacts on reservoir operations in mountainous watersheds than when investigating impacts on biogenic emissions and air quality (where VOCs are a primary indicator).« less
  • This study examined the impact of model biases on climate change signals for daily precipitation and for minimum and maximum temperatures. Through the use of multiple climate scenarios from 12 regional climate model simulations, the ensemble mean, and three synthetic simulations generated by a weighting procedure, we investigated intermodel seasonal climate change signals between current and future periods, for both median and extreme precipitation/temperature values. A significant dependence of seasonal climate change signals on the model biases over southern Québec in Canada was detected for temperatures, but not for precipitation. This suggests that the regional temperature change signal is affectedmore » by local processes. Seasonally, model bias affects future mean and extreme values in winter and summer. In addition, potentially large increases in future extremes of temperature and precipitation values were projected. For three synthetic scenarios, systematically less bias and a narrow range of mean change for all variables were projected compared to those of climate model simulations. In addition, synthetic scenarios were found to better capture the spatial variability of extreme cold temperatures than the ensemble mean scenario. Finally, these results indicate that the synthetic scenarios have greater potential to reduce the uncertainty of future climate projections and capture the spatial variability of extreme climate events.« less