Application Of The Piecewise Modeling Approach To Evaluate The Impact Of Greenhouse Gas Emissions On The Past Climate Change

The piecewise modeling approach(PW)used in numerical sensitivity experiment can effectively reduce error accumulation caused by a continuous long-term integrations,in which the simulations are updated periodically with analysis data.In this paper,the piecewise modeling approach is first applied to study the impact of greenhouse gas emissions on the recent climate change to test the ability of PW applied in the attribution studies of climate change performed with a complex climate model.We want to know whether PW performs better than the conventional continuous modeling approach(CONT),and whether PW can get more accurate results on the simulation of greenhouse gas emissions,and whether PW can improve credibility of the sensitivity experiment.Therefore,we use the Community Earth System Model(CESM)to perform an idealized sensitivity experiment for studying the contribution of greenhouse gases emissions to climate change in resent 26 years(1979-2004).The experiment includes two simulations: The first is a real historical simulation driven with all real external forcings(referred to ALL run).In the second simulation the greenhouse gas(GHG)forcings are removed from the all forcings and the well-mixed GHG concentrations are fixed at their initial values(referred to FGHG run).The difference between the two simulations(ALL-FGHG)can be regarded as the contribution of greenhouse gas emissions.The simulated “true” value with a “perfect” model is used to compare the modeling precision of PW and CONT.Furthermore,we examine the effect of different update periods and the errors in the analysis data with PW.The main conclusions are listed as follows:(1)In our experiment,the “imperfect” model performing the sensitivity experiment is obtained by reducing the threshold value of minimum RH for high stable clouds(rhminh)in the “perfect” model,which results in increasing high cloud amount and reducing solar shortwave radiation that reaches the ground,and consequently decreasing temperature.As a result,in ALL run,the simulated surface atmospheric temperature and precipitation simulated with CONT is lower than the “true”,and there is a significant systematic error.By updating the simulations with analysis data,PW can simulate actual climate evolution process primely and achieve high accuracy,which can be found from not only annual average but also day-to-day wind and temperature on the grid points.In ALL run,PW can perform far better than CONT in the simulation of precipitation,but its simulation accuracy is slightly lower than that of wind and temperature.It indicates that updating wind and temperature can not eliminate the error in precipitation simulation completely.(2)In FGHG run,the model variables are updated partly with analysis data in PW,so we can not guarantee the high simulation accuracy as that in ALL run.Even in some areas and time,the simulation results with PW is worse than that with CONT.However,PW still improves the simulation in FGHG run to some extent.(3)Although both ALL and FGHG simulations have systematic errors in CONT run,this error can be offset to a large extent when we calculate contribution of greenhouse gas emissions by subtracting FGHG from ALL.For this reason,the error in difference field(ALL-FGHG)is less than that in each simulation,and the simulated difference fields have some credibility in assessment of the contribution of greenhouse gas emissions.However,the error of the two simulations is on the contrary in many areas and time,which lead to accumulation of the errors in calculating the difference between the two simulations,and consequently to obvious discrepancy between the simulated difference field and its “true” in spatial distribution.Benefit from the improvement of ALL run,PW significantly improves the simulation for the contribution of greenhouse gases emissions on the surface atmospheric temperature,precipitation and wind,and increases the credibility of the sensitivity experiment.The comparison to CONT shows PW has advantages over CONT,especially for the spatial distribution of the simulated fields.(4)The update period has an impact on PW.In all,the update period is shorter,the results are more accurate.There is little influence on the simulations when the update period increases from 1 day to 2 days.However,there is some influence when the update period increases to 5 days,which can make lower simulation precision in ALL and FGHF runs,and finally reduce the simulation accuracy in the contribution of greenhouse gases.The addition of analysis error can reduce the simulation effect to some degree.If analysis error is with in the acceptable range,the simulated error with PW is smaller than that with CONT.(5)The experimental results indicate that PW has good simulation ability to study the impact of human activities on climate change.PW could get more reliable results than CONT.Especially on the regional scale,PW can describe spatial distribution of human influence on climate change in detail,which implies that PW has a wide range of application in model-based sensitivity experiments.