Evaluation of Grid-point Atmospheric Model of IAP LASG ...meeting.iap.ac.cn/Pdf/P47.pdf ·...

1
Evaluation of Grid-point Atmospheric Model of IAP LASG version 3 (GAMIL3) and its coupled model FGOALS-g3 Lijuan Li, Li Dong, Yanli Tang, Jinbo Xie, Yongqiang Yu, Pengfei Lin, Mirong Song , Tianjun Zhou, Ye Pu, He Wang, Xiangjun Shi*, Lu Wang*, Tao Feng # , Zhenghui Xie, Bin Wang LASG, Institute of Atmospheric Physics (IAP), Chinese Academy of Sciences (CAS), Beijing; * NUIST; # Yunnan university 1. Model Description The Flexible Global Ocean-Atmosphere-Land System model, Grid-point (FGOALS-g) was developed by the LASG, Institute of Atmospheric Physics (IAP) since the early 2000s, and it is composed of four component models of the Grid- point Atmospheric Model of IAP LASG (GAMIL), LASG IAP Climate system Ocean Model (LICOM), Land Surface Model and sea ice model. Through upgrading the atmospheric, ocean, land surface components based on the FGOALS-g2 (one of CMIP5 models), the FGOALS-g3 has been developed and conducted the CMIP6 experiments. This study evaluated the performances of GAMIL3 and FGOALS-g3 based on some of CMIP experiments, such as AMIP, piControl, abrupt4xCO2, and historical runs. CICE4- LASG FGOALS-g2 (CMIP5) LICOM2 GAMIL2 CLM3 CPL6 FGOALS-g3 (CMIP6) LICOM3 GAMIL3 CAS-LSM CPL7/C-Coupler CICE4- LASG Figure 1 Framework of the FGOALS-g GAMIL2 (CMIP5) GAMIL3 (CMIP6) Dynamical Core Horizontal resolution/time step 128×60/240s 180×80/120s Parallel One-dimensional decomposition in the meridional direction Two-dimensional hybrid decomposition (Liu et al., 2014) Water vapor advection Two-step shape-preserving advection scheme (TSPAS) Bug fix about area weighting function for TSPAS Physical processes Aerosol module boundary layer scheme a first -order closure scheme (Holtslag and Boville, 1993) a 1.5-order closure scheme (Bretherton and Park, 2009) & an explicit entrainment process and cloud-top radiative cooling (Sun et al., 2016) Stratocumulus cloud fraction traditional low-tropospheric stability (Klein et al., 1993 ) Estimated Inversion Strength (EIS) based (Wood and Bretherton, 2006 & Guo and Zhou, 2014) Convection parameterization none Convective momentum transport (Wu et al., 2007) Land Land surface model CLM2 CAS-LSM (Xie et al., 2018) Coupler coupler None CPL7 (Craig et al., 2012) Forcings Forcings Recommended by CMIP5 Recommended by CMIP6 Table 1 Major differences between GAMIL2 and GAMIL3 Reference 1) Li L. J., Y. Q. Yu, Y. L. Tang, et al.., 2020, The Flexible Global Ocean–Atmosphere–Land System Model Grid-Point Version 3 (FGOALS-g3): Description and Evaluation, JAMES, Submitted 2) Li L. J., L. Dong et al., 2020, The Grid-point Atmospheric Model of IAP LASG version 3 (GAMIL3): Model Description and Evaluation, To be submitted. 3) Wang H., L. J. Li*, X. L. Chen et al., Comparison of Climate sensitivities and feedbacks between FGOALS-g3 and FGOALS- g2, to be submitted Figure 6. (Upper panels) Boreal winter November–March) lag-longitude diagram of correlation coefficients between 20–100-day filtered OLR (shaded) and U850 (contours) upon the 20–100-day filtered OLR averaged over a reference region (10°S–5°N, 75°–100°E). (Lower panels) Boreal summer (June–October) lag- latitude diagram of correlation coefficients between 20–100-day filtered OLR (shaded) and U850 (contours) upon the same OLR reference time series. Note that the OLR reference time series was reversed sign before calculating correlation. (a) and (c) are for observations while (b) and (d) are for FGOALS-g3 historical simulation. Figure 9. Net radiation at TOA against the global mean SAT changes under abrupt4xCO2 scenario relative to pi- Control run in FGOALS-g3 (a) and FGOALS-g2 (b). The black lines are presented at the fast-response stage (the first 20 years) and the slow-response stage (the other 130 years). The red lines are presented at all-response stage (all 150- year). Figure 4. Horizontal distributions of (a) annual mean SST (units: °C) and (b) bias during 1980–2014 from the ensemble mean based on six historical runs. The ERSST v5 is referenced as the observed. Figure 5. The 2-m temperature (degC) from (a) Willmott and Matsuura (2001) and (b) FGOALS-g3, and (c) the difference between and (a) and (b) during 1980–2014. 2. Model evaluation Figure 2. Multivariable Taylor diagram displaying normalized statistical comparisons of simulated (c) climatology and (d) interannual variability of different meteorological variables with ERA-interim and GPCP as observations, respectively. Compared with GAMIL2, GAMIL3 has an improvement in climatological mean and interannual variability. Figure 3. Evolution of globally averaged surface temperature from the 700-yr pre-industrial control run by FGOALS-g3 (X-axis: yr; Y-axis: °C). The 700-yr mean global average surface temperature (GAST) is 13.7°C with a standard deviation of 0.1°C, and the GAST linear trend is – 0.015°C per 100 yr (Fig. 1), which is a smaller climate drift than the CMIP5 piControl run by FGOALS-g2 (–0.039°C per 100 yr; Lin et al., 2016; Nie et al., 2019) Model stability Mean State Climate variability Figure 7. (a–b) Spatial distribution of the standard deviation of the interannual anomaly of SST, derived from the observation (left) and the FGOALS-g3 historical simulation (right). (c–d) Standard deviations of SST anomalies for each calendar month. Orange and blue bars indicate the averaged results over the Niño3 and Niño3.4 regions, respectively. Observations FGOALS-g2 FGOALS-g3 1910–1940 0.140 0.043 0.099 1940–1970 –0.026 0.025 –0.022 1970–2005 0.195 0.173 0.176 Table 2. Linear trend of global surface temperature anomalies over three periods, which were obtained from observations and the ensemble average of the historical runs from FGOALS-g2 and FGOALS-g3 (in °C/10 yr). Global warming simulation Figure 8. Time-series of the global surface temperature anomalies for the historical runs (red lines; the thin red line is the different member run and the thick red line is the ensemble mean), and four future scenario runs (SSP1-2.6 = green line; SSP2-4.5 = pink line; SSP3-7.0 = blue line; SSP5-8.5 = purple line) from FGOALS-g3 relative to the period 1960– 1990. For comparison, the observation during the historical period (black line) are also shown. Climate sensitivity

Transcript of Evaluation of Grid-point Atmospheric Model of IAP LASG ...meeting.iap.ac.cn/Pdf/P47.pdf ·...

Page 1: Evaluation of Grid-point Atmospheric Model of IAP LASG ...meeting.iap.ac.cn/Pdf/P47.pdf · Institute of Atmospheric Physics (IAP) since the early 2000s, and it is composed of four

Evaluation of Grid-point Atmospheric Model of IAP LASG version 3 (GAMIL3) and its coupled model FGOALS-g3

Lijuan Li, Li Dong, Yanli Tang, Jinbo Xie, Yongqiang Yu, Pengfei Lin, Mirong Song , Tianjun Zhou, Ye Pu, He Wang, Xiangjun Shi*, Lu Wang*, Tao Feng#, Zhenghui Xie, Bin Wang

LASG, Institute of Atmospheric Physics (IAP), Chinese Academy of Sciences (CAS), Beijing; * NUIST; # Yunnan university

1. Model DescriptionThe Flexible Global Ocean-Atmosphere-Land System model, Grid-point (FGOALS-g) was developed by the LASG,

Institute of Atmospheric Physics (IAP) since the early 2000s, and it is composed of four component models of the Grid-

point Atmospheric Model of IAP LASG (GAMIL), LASG IAP Climate system Ocean Model (LICOM), Land Surface Model and

sea ice model. Through upgrading the atmospheric, ocean, land surface components based on the FGOALS-g2 (one of

CMIP5 models), the FGOALS-g3 has been developed and conducted the CMIP6 experiments. This study evaluated the

performances of GAMIL3 and FGOALS-g3 based on some of CMIP experiments, such as AMIP, piControl, abrupt4xCO2, and

historical runs.

CICE4-LASG

FGOALS-g2 (CMIP5)

LICOM2

GAMIL2

CLM3CPL6

FGOALS-g3 (CMIP6)

LICOM3

GAMIL3

CAS-LSMCPL7/C-CouplerCICE4-

LASG

Figure 1 Framework of the FGOALS-g

GAMIL2 (CMIP5) GAMIL3 (CMIP6)

Dynamical

Core

Horizontal resolution/time

step

128×60/240s 180×80/120s

Parallel One-dimensional decomposition in

the meridional direction

Two-dimensional hybrid

decomposition (Liu et al., 2014)

Water vapor advection Two-step shape-preserving advection

scheme (TSPAS)

Bug fix about area weighting

function for TSPAS

Physical

processes

Aerosol module

boundary layer scheme a first-order closure scheme (Holtslag

and Boville, 1993)

a 1.5-order closure scheme

(Bretherton and Park, 2009) & an

explicit entrainment process and

cloud-top radiative cooling (Sun

et al., 2016)

Stratocumulus cloud

fraction

traditional low-tropospheric stability (Klein et al., 1993 )

Estimated Inversion Strength (EIS)

based (Wood and Bretherton,

2006 & Guo and Zhou, 2014)

Convection

parameterization

none Convective momentum transport

(Wu et al., 2007)

Land Land surface model CLM2 CAS-LSM (Xie et al., 2018)

Coupler coupler None CPL7 (Craig et al., 2012)

Forcings Forcings Recommended by CMIP5 Recommended by CMIP6

Table 1 Major differences between GAMIL2 and GAMIL3

Reference

1) Li L. J., Y. Q. Yu, Y. L. Tang, et al.., 2020, The Flexible Global Ocean–Atmosphere–Land System Model Grid-Point Version 3 (FGOALS-g3): Description and Evaluation, JAMES, Submitted

2) Li L. J., L. Dong et al., 2020, The Grid-point Atmospheric Model of IAP LASG version 3 (GAMIL3): Model Description and Evaluation, To be submitted.

3) Wang H., L. J. Li*, X. L. Chen et al., Comparison of Climate sensitivities and feedbacks between FGOALS-g3 and FGOALS-

g2, to be submitted

Figure 6. (Upper panels) Boreal winter November–March) lag-longitude diagram of correlation coefficients between 20–100-day filtered OLR (shaded) and U850 (contours) upon the 20–100-day filtered OLR averaged over a reference region (10°S–5°N, 75°–100°E). (Lower panels) Boreal summer (June–October) lag-latitude diagram of correlation coefficients between 20–100-day filtered OLR (shaded) and U850 (contours) upon the same OLR reference time series. Note that the OLR reference time series was reversed sign before calculating correlation. (a) and (c) are for observations while (b) and (d) are for FGOALS-g3 historical simulation.

Figure 9. Net radiation at TOA against the global mean SAT changes under abrupt4xCO2 scenario relative to pi-Control run in FGOALS-g3 (a) and FGOALS-g2 (b). The black lines are presented at the fast-response stage (the first 20 years) and the slow-response stage (the other 130 years). The red lines are presented at all-response stage (all 150-year).

Figure 4. Horizontal distributions of (a) annual meanSST (units: °C) and (b) bias during 1980–2014 fromthe ensemble mean based on six historical runs. TheERSST v5 is referenced as the observed.

Figure 5. The 2-m temperature (degC) from (a)Willmott and Matsuura (2001) and (b) FGOALS-g3,and (c) the difference between and (a) and (b) during1980–2014.

2. Model evaluation

Figure 2. Multivariable Taylor diagram displaying normalized statistical comparisons of simulated (c) climatology and (d) interannual variability of different meteorological variables with ERA-interim and GPCP as observations, respectively.

Compared with GAMIL2,GAMIL3 has an improvement inclimatological mean andinterannual variability.

Figure 3. Evolution of globally averaged surfacetemperature from the 700-yr pre-industrial controlrun by FGOALS-g3 (X-axis: yr; Y-axis: °C).The 700-yr mean global average surfacetemperature (GAST) is 13.7°C with a standarddeviation of 0.1°C, and the GAST linear trend is –0.015°C per 100 yr (Fig. 1), which is a smallerclimate drift than the CMIP5 piControl run byFGOALS-g2 (–0.039°C per 100 yr; Lin et al., 2016;Nie et al., 2019)

Model stability

Mean State

Climate variability

Figure 7. (a–b) Spatial distribution of the standard deviation of the interannual anomaly of SST, derived from the observation (left) and the FGOALS-g3 historical simulation (right). (c–d) Standard deviations of SST anomalies for each calendar month. Orange and blue bars indicate the averaged results over the Niño3 and Niño3.4 regions, respectively.

Observations FGOALS-g2 FGOALS-g31910–1940 0.140 0.043 0.099

1940–1970 –0.026 0.025 –0.022

1970–2005 0.195 0.173 0.176

Table 2. Linear trend of global surface temperature anomalies over three periods, which were obtained from observations and the ensemble average of the historical runs from FGOALS-g2 and FGOALS-g3 (in °C/10 yr).

Global warming simulation

Figure 8. Time-series of the global surface temperature anomalies for the historical runs (red lines; the thin red line isthe different member run and the thick red line is the ensemble mean), and four future scenario runs (SSP1-2.6 = greenline; SSP2-4.5 = pink line; SSP3-7.0 = blue line; SSP5-8.5 = purple line) from FGOALS-g3 relative to the period 1960–1990. For comparison, the observation during the historical period (black line) are also shown.

Climate sensitivity