Errors in climate model daily precipitation and temperature output: time invariance and implications for bias correction

Abstract. When correcting for biases in general circulation model (GCM) output, for example when statistically downscaling for regional and local impacts studies, a common assumption is that the GCM biases can be characterized by comparing model simulations and observations for a historical period. We demonstrate some complications in this assumption, with GCM biases varying between mean and extreme values and for different sets of historical years. Daily precipitation and maximum and minimum temperature from late 20th century simulations by four GCMs over the United States were compared to gridded observations. Using random years from the historical record we select a "base" set and a 10 yr independent "projected" set. We compare differences in biases between these sets at median and extreme percentiles. On average a base set with as few as 4 randomly-selected years is often adequate to characterize the biases in daily GCM precipitation and temperature, at both median and extreme values; 12 yr provided higher confidence that bias correction would be successful. This suggests that some of the GCM bias is time invariant. When characterizing bias with a set of consecutive years, the set must be long enough to accommodate regional low frequency variability, since the bias also exhibits this variability. Newer climate models included in the Intergovernmental Panel on Climate Change fifth assessment will allow extending this study for a longer observational period and to finer scales.

Download Full-text

Errors in climate model daily precipitation and temperature output: time invariance and implications for bias correction

Hydrology and Earth System Sciences Discussions ◽

10.5194/hessd-10-1657-2013 ◽

2013 ◽

Vol 10 (2) ◽

pp. 1657-1691 ◽

Cited By ~ 1

Author(s):

E. P. Maurer ◽

T. Das ◽

D. R. Cayan

Keyword(s):

Bias Correction ◽

General Circulation ◽

Daily Precipitation ◽

Extreme Values ◽

Circulation Model ◽

The United States ◽

Historical Period ◽

Late 20Th Century ◽

Precipitation And Temperature ◽

Gcm Biases

Abstract. When correcting for biases in general circulation model (GCM) output, for example when statistically downscaling for regional and local impacts studies, a common assumption is that the GCM biases can be characterized by comparing model simulations and observations for a historical period. We demonstrate some complications in this assumption, with GCM biases varying between mean and extreme values and for different sets of historical years. Daily precipitation and maximum and minimum temperature from late 20th century simulations by four GCMs over the United States were compared to gridded observations. Using random years from the historical record we select a "base" set and a 10-yr independent "projected" set. We compare differences in biases between these sets at median and extreme percentiles. On average a base set with as few as 4 randomly-selected years is often adequate to characterize the biases in daily GCM precipitation and temperature, at both median and extreme values; 12 yr provided higher confidence that bias correction would be successful. This suggests that some of the GCM bias is time invariant. When characterizing bias with a set of consecutive years, the set must be long enough to accommodate regional low frequency variability, since the bias also exhibits this variability. Newer climate models included in the Intergovernmental Panel on Climate Change fifth assessment will allow extending this study for a longer observational period and to finer scales.

Download Full-text

Probabilistic Projections of Climate Change for the Mid-Atlantic Region of the United States: Validation of Precipitation Downscaling during the Historical Era*

Journal of Climate ◽

10.1175/2011jcli4091.1 ◽

2012 ◽

Vol 25 (2) ◽

pp. 509-526 ◽

Cited By ~ 27

Author(s):

Liang Ning ◽

Michael E. Mann ◽

Robert Crane ◽

Thorsten Wagener

Keyword(s):

United States ◽

General Circulation ◽

Large Scale ◽

Daily Precipitation ◽

Circulation Model ◽

The United States ◽

Monthly Precipitation ◽

Atlantic Region ◽

Daily Precipitation Data ◽

Training Interval

Abstract This study uses a statistical downscaling method based on self-organizing maps (SOMs) to produce high-resolution, downscaled precipitation estimates over the state of Pennsylvania in the mid-Atlantic region of the United States. The SOMs approach derives a transfer function between large-scale mean atmospheric states and local meteorological variables (daily point precipitation values) of interest. First, the SOM was trained using seven coarsely resolved atmospheric variables from the National Centers for Environmental Prediction (NCEP) reanalysis dataset to model observed daily precipitation data from 17 stations across Pennsylvania for the period 1979–2005. Employing the same coarsely resolved variables from nine general circulation model (GCM) simulations taken from the historical analysis of the Coupled Model Intercomparison Project, phase 3 (CMIP3), the trained SOM was subsequently applied to simulate daily precipitation at the same 17 sites for the period 1961–2000. The SOM analysis indicates that the nine model simulations exhibit similar synoptic-scale features to the (NCEP) observations over the 1979–2007 training interval. An analysis of the sea level pressure signatures and the precipitation distribution corresponding to the trained SOM shows that it is effective in differentiating characteristic synoptic circulation patterns and associated precipitation. The downscaling approach provides a faithful reproduction of the observed probability distributions and temporal characteristics of precipitation on both daily and monthly time scales. The downscaled precipitation field shows significant improvement over the raw GCM precipitation fields with regard to observed average monthly precipitation amounts, average monthly number of rainy days, and standard deviations of monthly precipitation amounts, although certain caveats are noted.

Download Full-text

Meteorological modes of variability for fine particulate matter (PM2.5) air quality in the United States: implications for PM2.5 sensitivity to climate change

Atmospheric Chemistry and Physics ◽

10.5194/acp-12-3131-2012 ◽

2012 ◽

Vol 12 (6) ◽

pp. 3131-3145 ◽

Cited By ~ 98

Author(s):

A. P. K. Tai ◽

L. J. Mickley ◽

D. J. Jacob ◽

E. M. Leibensperger ◽

L. Zhang ◽

...

Keyword(s):

Climate Change ◽

General Circulation ◽

Transport Model ◽

Circulation Model ◽

Multiple Linear Regression Model ◽

The United States ◽

Chemical Transport Model ◽

Modes Of Variability ◽

The Us ◽

Cyclone Frequency

Abstract. We applied a multiple linear regression model to understand the relationships of PM2.5 with meteorological variables in the contiguous US and from there to infer the sensitivity of PM2.5 to climate change. We used 2004–2008 PM2.5 observations from ~1000 sites (~200 sites for PM2.5 components) and compared to results from the GEOS-Chem chemical transport model (CTM). All data were deseasonalized to focus on synoptic-scale correlations. We find strong positive correlations of PM2.5 components with temperature in most of the US, except for nitrate in the Southeast where the correlation is negative. Relative humidity (RH) is generally positively correlated with sulfate and nitrate but negatively correlated with organic carbon. GEOS-Chem results indicate that most of the correlations of PM2.5 with temperature and RH do not arise from direct dependence but from covariation with synoptic transport. We applied principal component analysis and regression to identify the dominant meteorological modes controlling PM2.5 variability, and show that 20–40% of the observed PM2.5 day-to-day variability can be explained by a single dominant meteorological mode: cold frontal passages in the eastern US and maritime inflow in the West. These and other synoptic transport modes drive most of the overall correlations of PM2.5 with temperature and RH except in the Southeast. We show that interannual variability of PM2.5 in the US Midwest is strongly correlated with cyclone frequency as diagnosed from a spectral-autoregressive analysis of the dominant meteorological mode. An ensemble of five realizations of 1996–2050 climate change with the GISS general circulation model (GCM) using the same climate forcings shows inconsistent trends in cyclone frequency over the Midwest (including in sign), with a likely decrease in cyclone frequency implying an increase in PM2.5. Our results demonstrate the need for multiple GCM realizations (because of climate chaos) when diagnosing the effect of climate change on PM2.5, and suggest that analysis of meteorological modes of variability provides a computationally more affordable approach for this purpose than coupled GCM-CTM studies.

Download Full-text

Utility of daily vs. monthly large-scale climate data: an intercomparison of two statistical downscaling methods

Hydrology and Earth System Sciences Discussions ◽

10.5194/hessd-4-3413-2007 ◽

2007 ◽

Vol 4 (5) ◽

pp. 3413-3440 ◽

Cited By ~ 18

Author(s):

E. P. Maurer ◽

H. G. Hidalgo

Keyword(s):

Statistical Downscaling ◽

General Circulation ◽

Large Scale ◽

Impact Analysis ◽

Climate Model ◽

Daily Precipitation ◽

Circulation Model ◽

Reanalysis Data ◽

Temperature Extremes ◽

Daily Data

Abstract. Downscaling of climate model data is essential to most impact analysis. We compare two methods of statistical downscaling to produce continuous, gridded time series of precipitation and surface air temperature at a 1/8-degree (approximately 140 km² per grid cell) resolution over the western U.S. We use NCEP/NCAR Reanalysis data from 1950–1999 as a surrogate General Circulation Model (GCM). The two methods included are constructed analogues (CA) and a bias correction and spatial downscaling (BCSD), both of which have been shown to be skillful in different settings, and BCSD has been used extensively in hydrologic impact analysis. Both methods use the coarse scale Reanalysis fields of precipitation and temperature as predictors of the corresponding fine scale fields. CA downscales daily large-scale data directly and BCSD downscales monthly data, with a random resampling technique to generate daily values. The methods produce comparable skill in producing downscaled, gridded fields of precipitation and temperatures at a monthly and seasonal level. For daily precipitation, both methods exhibit some skill in reproducing both observed wet and dry extremes and the difference between the methods is not significant, reflecting the general low skill in daily precipitation variability in the reanalysis data. For low temperature extremes, the CA method produces greater downscaling skill than BCSD for fall and winter seasons. For high temperature extremes, CA demonstrates higher skill than BCSD in summer. We find that the choice of most appropriate downscaling technique depends on the variables, seasons, and regions of interest, on the availability of daily data, and whether the day to day correspondence of weather from the GCM needs to be reproduced for some applications. The ability to produce skillful downscaled daily data depends primarily on the ability of the climate model to show daily skill.

Download Full-text

Simulation of Daily Extreme Precipitation over the United States in the CMIP5 30-Yr Decadal Prediction Experiment

Journal of Applied Meteorology and Climatology ◽

10.1175/jamc-d-18-0057.1 ◽

2019 ◽

Vol 58 (4) ◽

pp. 875-886 ◽

Cited By ~ 3

Author(s):

Steve T. Stegall ◽

Kenneth E. Kunkel

Keyword(s):

United States ◽

Extreme Precipitation ◽

Daily Precipitation ◽

The United States ◽

Natural Variability ◽

Return Level ◽

Historical Period ◽

Regional Variability ◽

Anthropogenic Forcing ◽

Near Term

AbstractThe CMIP5 decadal hindcast (“Hindcast”) and prediction (“Predict”) experiment simulations from 11 models were analyzed for the United States with respect to two metrics of extreme precipitation: the 10-yr return level of daily precipitation, derived from the annual maximum series of daily precipitation, and the total precipitation exceeding the 99.5th percentile of daily precipitation. Both Hindcast simulations and observations generally show increases for the 1981–2010 historical period. The multimodel-mean Hindcast trends are statistically significant for all regions while the observed trends are statistically significant for the Northeast, Southeast, and Midwest regions. An analysis of CMIP5 simulations driven by historical natural (“HistoricalNat”) forcings shows that the Hindcast trends are generally within the 5th–95th-percentile range of HistoricalNat trends, but those outside that range are heavily skewed toward exceedances of the 95th-percentile threshold. Future projections for 2006–35 indicate increases in all regions with respect to 1981–2010. While there is good qualitative agreement between the observations and Hindcast simulations regarding the direction of recent trends, the multimodel-mean trends are similar for all regions, while there is considerable regional variability in observed trends. Furthermore, the HistoricalNat simulations suggest that observed historical trends are a combination of natural variability and anthropogenic forcing. Thus, the influence of anthropogenic forcing on the magnitude of near-term future changes could be temporarily masked by natural variability. However, continued observed increases in extreme precipitation in the first decade (2006–15) of the “future” period partially confirm the Predict results, suggesting that incorporation of increases in planning would appear prudent.

Download Full-text

Bias Correction of Zero-Inflated RCM Precipitation Fields: A Copula-Based Scheme for Both Mean and Extreme Conditions

Journal of Hydrometeorology ◽

10.1175/jhm-d-18-0126.1 ◽

2019 ◽

Vol 20 (4) ◽

pp. 595-611 ◽

Cited By ~ 5

Author(s):

Rajib Maity ◽

Mayank Suman ◽

Patrick Laux ◽

Harald Kunstmann

Keyword(s):

Climate Change ◽

Bias Correction ◽

Extreme Precipitation ◽

Joint Distribution ◽

Daily Precipitation ◽

Extreme Values ◽

Entire Study ◽

Daily Precipitation Data ◽

Zero Values ◽

The Mean

Abstract Changes in extreme precipitation due to climate change often require the application of methods to bias correct simulated atmospheric fields, including extremes. Most existing bias correction techniques (i) only focus on the bias in the mean value or on the extreme values separately, and (ii) exclude zero values from analysis, even though their presence is significant in daily precipitation. We developed a copula-based bias correction scheme that is suitable for zero-inflated daily precipitation data to correct the bias in mean as well as in extreme precipitation at any specific statistical quantile. In considering the whole of Germany as a test bed, the proposed scheme is found to work well across the entire study area, including the German Alpine regions. The joint distribution between observed and regional climate model (RCM)-derived precipitation is developed through copulas. In particular, the joint distribution is modified to make it discrete at zero in order to account for zero values. The benefit of considering zero precipitation values is revealed through the improved performance of bias correction both in the mean and extreme values. Second, the quantile that best captures the bias (whether in the mean or any extreme value) is determined for a specific location and varies spatially and seasonally. This relaxation in selecting the location-specific optimal quantile renders the proposed methodology spatially transferable. By acknowledging possible changes in extreme precipitation due to climate change, the proposed scheme is expected to be suitable for climate change impact assessments for extreme events worldwide.

Download Full-text

Multimodel Estimates of Atmospheric Response to Modes of SST Variability and Implications for Droughts

Journal of Climate ◽

10.1175/2010jcli3295.1 ◽

2010 ◽

Vol 23 (16) ◽

pp. 4327-4341 ◽

Cited By ~ 11

Author(s):

Philip J. Pegion ◽

Arun Kumar

Keyword(s):

United States ◽

Climate Variability ◽

General Circulation ◽

Dominant Role ◽

Circulation Model ◽

The United States ◽

Precipitation Variability ◽

Climate Extreme ◽

Central Pacific ◽

Sst Variability

Abstract A set of idealized global model experiments was performed by several modeling centers as part of the Drought Working Group of the U.S. Climate Variability and Predictability component of the World Climate Research Programme (CLIVAR). The purpose of the experiments was to assess the role of the leading modes of sea surface temperature (SST) variability on the climate over the continents, with particular emphasis on the influence of SSTs on surface climate variability and droughts over the United States. An analysis based on several models gives more creditability to the results since it relies on the assessment of impacts that are robust across different models. Coordinated atmospheric general circulation model (AGCM) simulations forced with three modes of SST variability were analyzed. The results show that the SST-forced precipitation variability over the central United States is dominated by the SST mode with maximum loading in the central Pacific Ocean. The SST mode with loading in the Atlantic Ocean, and a mode that is dominated by trends in SSTs, lead to a smaller response. Based on the response to the idealized SSTs, the precipitation response for the twentieth century was also reconstructed. A comparison with the Atmospheric Model Intercomparison Project (AMIP) simulations forced with the observed SSTs illustrates that the reconstructed precipitation variability was similar to the one in the AMIP simulations, further supporting the conclusion that the SST modes identified in the present analysis play a dominant role in the precipitation variability over the United States. One notable exception is the Dust Bowl of the 1930s, and further analysis regarding this major climate extreme is discussed.

Download Full-text