ORIGINAL RESEARCH

published: 24 June 2021

doi: 10.3389/feart.2021.687976

Comparative Assessment and Future

Prediction Using CMIP6 and CMIP5 for
Annual Precipitation and Extreme
Precipitation Simulation
Jingjing Li†, Ran Huo†, Hua Chen *, Ying Zhao and Tianhui Zhao
State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University, Wuhan, China

This study assesses the improvement of the latest Coupled Model Intercomparison Project
Phase 6 (CMIP6) over Coupled Model Intercomparison Project Phase 5 (CMIP5) for
precipitation simulation. Precipitation simulations under different future climate scenarios
are also compared in this work. The results show that: 1) CMIP6 has no overall advantage
over CMIP5 in simulating total precipitation (PRCPTOT) and maximum consecutive dry
days (CDD). The performance of CMIP6 increases or decreases regionally in PRCPTOT
Edited by: and consecutive dry days. But it is slightly worse than CMIP5 in simulating very wet days
Alexandre M. Ramos,
University of Lisbon, Portugal (R95pTOT). 2) Comparing the trend test results of CMIP5 and CMIP6 in the future, there
Reviewed by: are more areas with significant trend based on Mann–Kendall test in CMIP6 compared with
Shoji Kusunoki, that of CMIP5. The differences in PRCPTOT are mainly found in Amazon Basin and
Meteorological Research Institute,
Japan
Western Africa. The differences between the R95pTOT trends mainly noticeable in South
Renata Goncalves Tedeschi, America and Western Africa, and the differences in CDD are mainly reflected in Central
Vale Technological Institute (ITV), Asia, Sahara Desert and central South America. 3) In Southern South America and
Brazil
Western North America, the PRCPTOT changing rate of CMIP6 in the future under
*Correspondence:
Hua Chen various scenarios is always greater than that of CMIP5; in Alaska, Western Africa,
[email protected] Southern Africa, the PRCPTOT changing rate of CMIP6 in the future under various
†
These authors have contributed scenarios is always less than that of CMIP5. In Southern South America, the R95pTOT
equally to this work
changing rate of CMIP6 in the future under various scenarios is always greater than that of
Specialty section:
CMIP5; in Alaska, East Asia, North Asia, the R95pTOT changing rate of CMIP6 in the future
This article was submitted to under various scenarios is always less than that of CMIP5. In almost half of the regions, the
Atmospheric Science, CDD changing rate of CMIP6 is less than that of CMIP5 under all scenarios, namely
a section of the journal
Frontiers in Earth Science Australia, Amazon Basin, Southern South America, Central America, Western North
Received: 30 March 2021 America, Central North America, Eastern North America, Central Asia, Tibet.
Accepted: 14 May 2021
Published: 24 June 2021 Keywords: CMIP6, precipitation, worldwide, general circulation models, regions

Citation:
Li J, Huo R, Chen H, Zhao Y and INTRODUCTION
Zhao T (2021) Comparative
Assessment and Future Prediction
To analyze the impact of future climate change, the World Climate Research Program (WCRP)
Using CMIP6 and CMIP5 for Annual
Precipitation and Extreme
initiated the Coupled Model Intercomparison Project (CMIP) in 1995 (Meehl et al., 2000). With
Precipitation Simulation. scientific development and the gradually improved understanding of climate change mechanisms,
Front. Earth Sci. 9:687976. the CMIP has evolved from CMIP Phase 1 (CMIP1), Phase 2 (CMIP2), Phase 3 (CMIP3), and Phase
doi: 10.3389/feart.2021.687976 5 (CMIP5) to the latest Phase 6 model (CMIP6). CMIP formulates climate model test standards, and

Frontiers in Earth Science | www.frontiersin.org 1 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

the mechanism for sharing simulated climate data promotes CMIP5. Overall, most studies have focused on a specific region
model development and improvement. Therefore, it has (Bracegirdle et al., 2020; Davy and Outten 2020; Jiang et al., 2020)
gradually developed and become an indispensable element of rather than the world when assessing historical precipitation in
climate science. Many studies based on CMIP data provided CMIP6 (Fan et al., 2020; Ukkola et al., 2020). Compared to
support for the Intergovernmental Panel on Climate Change research that evaluated historical data in CMIP5 and CMIP6,
(IPCC) assessment reports (IPCC 2007; IPCC 2013). CMIP6 those with further analysis of future climate change in CMIP5
will provide data support for climate change research in the next and CMIP6 (Chen et al., 2020; Ukkola et al., 2020) are sparse.
five to ten years. Ukkola et al. (2020) found that drought changes in the future
CMIP6 newly proposed shared socioeconomic pathways (SSPs) (2051–2100) in CMIP6 are more considerable and more
(Moss et al., 2010), which describe different socioeconomic reference consistent than those in CMIP5. The results also showed that
assumptions. The SSPs and RCPs (representative concentration mean precipitation and variability influenced drought duration
pathways) were combined to provide an integrated scenario in and frequency. Chen et al. (2020) studied precipitation and
CMIP6 (Van Vuuren et al., 2014). In addition to the new future temperature extremes in the future after a historical data
scenario, a new generation of general circulation models (GCMs) is assessment. The work highlighted the changing rate of climate
used in CMIP6 (Veronika et al., 2016). The GCM is an indispensable indicators in the last 2 decades of this century compared with the
tool for exploring the interaction of climate systems, future climate historical record. The results suggested that extreme climate changes
change predictions, and climate change research. It has been proven are more pronounced in CMIP6 than in CMIP5. Davy and Outten
to behave well in simulating climate change and can be used to (2020) investigated the model distributions of surface air temperature
predict future precipitation under climate change (Li et al., 2019; and sea ice extent and volume in the Arctic until 2100 and compared
Khan et al., 2020). CMIP6 with CMIP5. The results showed that under SSP126, the
Climate change significantly affects the water cycle (Madsen et al., Arctic climate will stabilize by 2060. Most research only consider
2014; Zobel et al., 2018; Ortiz-Gómez et al., 2020). Obtaining high- several decades at the end of the 21st century when analyzing the
quality water resource forecast data under climate change is critical future, without a detailed analysis of the near future. Very little work
for sustainable water resource development. In addition to the total has been devoted to the analysis of future precipitation projected by
amount of water resources, the temporal and spatial distribution of CMIP6 worldwide after comparing it with CMIP5.
precipitation, especially extreme precipitation events, is worthy of More work is needed to analyze worldwide precipitation based
attention. Extreme precipitation can lead to drought and flood on CMIP6 and CMIP5 in detail. This study has two objectives: 1)
events (Ulbrich et al., 2003; Wang et al., 2016; Funk et al., 2018; to compare the historical precipitation simulation capability
Funk et al., 2019; Lal et al., 2020), which in turn will cause enormous between CMIP6 and CMIP5; and 2) analyze the differences in
societal, economic, and ecological losses (Kundzewicz et al., 2014; the temporal and spatial distributions of precipitation in the
Gao et al., 2019). To manage the hidden dangers of water security future in CMIP6 and CMIP5. In the middle of this century and at
caused by global climate change and deploy adaptive the end of this century, the impact of climate change will be very
countermeasures in advance, it is crucial to study the total different. Therefore, when analyzing future changes in
precipitation and extreme precipitation events under climate precipitation, this study divides the future into near-future and
change (Birkmann 2011; Forestieri et al., 2018; Rahmani and far-future periods. Three indicators are used in this study: total
Harrington 2019; Hosseinzadehtalaei et al., 2020; Ukkola et al., precipitation (PRCPTOT), very wet days (R95pTOT) and CDD.
2020). CMIP result data provide a meaningful way to analyze the The remainder of this paper is organized as follows. Methods and
temporal and spatial distribution of future precipitation. Materials describes the methods and materials used in this study.
With the further development of CMIP, studying the difference Results and Discussions presents the study results, focusing on four
between CMIP5 and CMIP6 has received critical attention. Several key themes. Evaluation of CMIP5 and CMIP6 Historical
studies have assessed CMIP6 by comparing the historical simulation Precipitation Simulation Capabilities presents the evaluation of
data of CMIP5 and CMIP6 with the observed data (Gusain et al., historical precipitation simulation capabilities. Then, trend
2020; Yuanhai et al., 2020; Zamani et al., 2020; Zhu et al., 2020; Zhu analysis and comparison of long series precipitation are presented
and Yang 2020). When CMIP5 and CMIP6 are compared, the multi- in Trend analysis and comparison of long series precipitation in
model mean result (Xin et al., 2020; Yuanhai et al., 2020) rather than a CMIP5 and CMIP6, followed by a comparison of the changing rates
specific GCM (Wyser et al., 2020) is used in most cases. Gusain et al. of CMIP5 and CMIP6 precipitation in the future compared with
(2020) compared CMIP6 and CMIP5 when simulating Indian history in Changing rate comparison of precipitation in the future
summer monsoon precipitation. The results showed that CMIP6 over the history. Finally, the model uncertainties in CMIP5 and
performed better than CMIP5, as the model deviation in CMIP6 was CMIP6 are compared in Comparison of Model uncertainty in CMIP5
significantly reduced. Thus, CMIP6 could better simulate the Indian and CMIP6 precipitation. Conclusions are drawn in Conclusion.
monsoon characteristics. Zhu et al. (2020) compared CMIP5 and
CMIP6 in China based on a daily observational dataset. They found
that CMIP6 behaved better in the simulation of total precipitation, METHODS AND MATERIALS
precipitation intensity, heavy precipitation, and extremely heavy rain
days, except for consecutive dry days (CDD). For precipitation, the Data
simulation of precipitation intensity was better in CMIP6, although In this study, the Global Precipitation Climatology Center
the simulation of precipitation frequency was comparable to that in (GPCC) V 2018 grid daily precipitation data were used as the

Frontiers in Earth Science | www.frontiersin.org 2 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

TABLE 1 | List of general circulation models (GCMs) in the Coupled Model Intercomparison Project phases 5 and 6 (CMIP5 and CMIP6).

CMIP6 CMIP5 Institution

Model Resolution (°) Model Resolution (° )
Lat*Lon Lat*Lon

BCC-CSM2-MR 1.12 × 1.13 BCC-CSM1.1 2.79 × 2.81 BCC

CESM2 0.94 × 1.25 CCSM4 0.94 × 1.25 NCAR
FGOALS-g3 2.0 × 2.81 FGOALS-g2 2.78 × 2.81 CAS
GFDL-CM4 1.00 × 1.25 GFDL-CM3 2.0 × 2.5 NOAA-GFDL
GFDL-ESM4 1.00 × 1.25 GFDL-ESM-2G 2.02 × 2.0 NOAA-GFDL
INM-CM4-8 1.5 × 2.0 INM-CM4 1.5 × 2.0 INM
IPSL-CM6A-LR 1.27 × 2.5 IPSL-CM5A-LR 1.89 × 3.75 IPSL
MRI-ESM2-0 1.12 × 1.13 MRI-ESM1 1.12 × 1.13 MRI

TABLE 2 | Indicators used in this study.

Indicator Acronym Definition Units

Total precipitation PRCPTOT Let Rwj be the daily precipitation amount for day w of period j. Then the total precipitation in period j mm
W
is PRCPTOTj   Rwj
w1

Very wet days R95pTOT Let Rwj be the daily precipitation amount for wet day (R > 1 mm) of period j and R95p the 95th percentile of mm
W
precipitation for wet days in the specified period. Then R95pTOT is determined as R95pTOTj   R , R > R95p
wj wj
w1

Maximum consecutive dry CDD Let Rij be the daily precipitation amount for day i of period j. Then count the largest number of consecutive days where day
days Rij˂1 mm

observation data (1982–2005) (Fuchs et al., 2009), with a In this study, the root mean square error (RMSE) was used as
resolution of 0.5° × 0.5°. Eight GCMs of CMIP6 and eight the criterion for evaluating modeled precipitation.
corresponding GCMs in CMIP5 were adopted for the modeled 













data. The specific GCM information is presented in Table 1. We 1 n
RMSE   (Xi − Yi )2 (1)
selected paired GCMs in CMIP5 and CMIP6. For example, the n i1
FGOALS-g3 in CMIP6 and FGOALS-g2 in CMIP5. It can reflect
the difference between CMIP6 and CMIP5. where X represents the GCM precipitation, Y represents the
Three corresponding future scenarios from CMIP6 and CMIP5 observation value, and n represents the series length, i
were used to explore how forcing level and socioeconomic represents the number of the variable.
development affect the temporal and spatial distribution of
precipitation. Those scenarios are SSP126 (RCP26), SSP245 The Mann—Kendall (MK) Trend Test
(RCP45), and SSP585 (RCP85), where radiative forcing stabilizes
at approximately 2.6 W/m2, 4.5 W/m2, and 8.5 W/m2 in 2100,
Method
MK trend test method is used to analyze hydrological series trends.
respectively. Due to historical simulation data limitations, the
The MK trend test method is widely used because of its simplicity
historical period was from 1982 to 2005 (24 years). The future is
(Mann 1945; Kendall 1990; Hamed 2008). When the standard normal
divided into near-future period (2037–2060, 24 years), and far-future
statistics Z is beyond the threshold of [−2.32, 2.32], the hydrological
period (2077–2100, 24 years) in this study.
series has a significance changing rate at the 99% confidence level. The
Because the resolution of each GCM is different, all GCM and
standard normal statistics Z is calculated as follow:
observation data were interpolated into 1° × 1° grid data sets using







the bilinear interpolation method.
⎨ (S − 1) Var(S), S > 0
⎪
⎧
Z⎪ 0,






S  0 (2)
⎩
Indicators (S + 1) Var(S), S < 0
Three precipitation indices were selected to describe the where S is a statistic,
precipitation characteristics from the 27 climate change indicators
n−1 n
recommended by the World Meteorological Organization (WMO)
S    sgn x j − x(i) (3)
and the Expert Team on Climate Change Detection, Monitoring i  1 j i+1
Indices (ETCCDMI) (Zhang et al., 2011): PRCPTOT, R95pTOT,
and CDD (Table 2). It helps to understand the total amount of water where n represents the series length, i, j represents the number of
resources and the risks of drought and flooding. the variable, sgn is calculated as:

Frontiers in Earth Science | www.frontiersin.org 3 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 1 | Regions used in this study.

⎪
⎧
⎨ 1, x(i) > x j TABLE 3 | List of regions used in this study. Only land grid points are used in the
analysis.
sgn x(i) − x j   ⎪ 0, x(i)  x j (4)
⎩
−1, x(i) < x j Name Acronym Latitude (°) Longitude (°)

Australia AUS 45 ± 11 S 110 E ± 155 E

Amazon Basin AMZ 20 S ± 12 N 82 ± 34 W
Sen’s Slope Estimator Test Southern South America SSA 56 ± 20 S 76 ± 40 W
We use Sen’s slope estimator (Swain 2014; Verma et al., 2016) to Central America CAM 10 ± 30 N 116 ± 83 W
detect the linear trend of precipitation in this study. The size of Western North America WNA 30 ± 60 N 130 ± 103 W
the trend is given as: Central North America CNA 30 ± 50 N 103 ± 85 W
Eastern North America ENA 25 ± 50 N 85 ± 60 W
xi − xj Alaska ALA 60 ± 72 N 170 ± 103 W
β  Median , ∀j < i (5) 50 ± 85 N 103 ± 10 W
i−j Greenland GRL
Mediterranean Basin MED 30 ± 48 N 10 W ± 40 E
48 ± 75 N 10 W ± 40 E
where β is the size of the trend of data series. Median means the Northern Europe NEU
Western Africa WAF 12 S ± 18 N 20 W ± 22 E
median of the data series. Eastern Africa EAF 12 S ± 18 N 22 ± 52 E
Southern Africa SAF 35 ± 12 S 10 W ± 52 E
Regions Sahara SAH 18 ± 30 N 20 W ± 65 E
There are considerable differences in the climate and Southeast Asia SEA 11 S ± 20 N 95 ± 155 E
East Asia EAS 20 ± 50 N 100 ± 145 E
geographical characteristics across the world. Therefore,
South Asia SAS 5 ± 30 N 65 ± 100 E
regional assessments are more meaningful for practical Central Asia CAS 30 ± 50 N 40 ± 75 E
guidance. The following factors should be considered Tibet TIB 30 ± 50 N 75 ± 100 E
when choosing regions (Giorgi and Francisco 2000): 1) the North Asia NAS 50 ± 70 N 40 ± 180 E
shape of the regions should be simple; 2) within a region, the
climate and geographical characteristics should be similar;
and 3) the regions should cover all the land except RESULTS AND DISCUSSIONS
Antarctica. Based on the above standards, the world can
be divided into 21 regions (Giorgi and Francisco 2000) Evaluation of CMIP5 and CMIP6 Historical
(Figure 1). The 21 regions include Australia (AUS), Precipitation Simulation Capabilities
Amazon Basin (AMZ), Southern South America (SSA), PRCPTOT, R95pTOT, and CDD were selected to evaluate the
Central America (CAM), Western North America (WNA), agreement between the simulated GCMs and the observed data
Central North America (CAN), Eastern North America in the historical period. The RMSE was used to characterize the
(ENA), Alaska (ALA), Greenland (GRL), Mediterranean simulation capability; the lower the RMSE, the better the simulation
Basin (MED), Northern Europe (NEU), Western Africa capability. The global distribution maps of the multi-model mean
(WAF), Eastern Africa (EAF), Southern Africa (SAF), RMSE of the three indicators in the historical period for CMIP5 and
Sahara (SAH), Southeast Asia (SEA), East Asia (EAS), CMIP6 are shown in Supplementary Figures S1–S3. Global
South Asia (SAS), Central Asia (CAS), Tibet (TIB), and distribution maps of the RMSE difference between CMIP5 and
North Asia (NAS). Table 3 provides detailed information CMIP6 are shown in Figure 2, calculated by the RMSE of CMIP5
on the 21 regions. minus the RMSE of CMIP6.

Frontiers in Earth Science | www.frontiersin.org 4 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 2 | Global distribution maps of root-mean-square error (RMSE) difference between CMIP5 and CMIP6. The multi-model mean RMSE of PCRPTOT,
R95pTOT, CDD is calculated based on the data in the historical period (1982–2005). When the RMSE of CMIP5 is bigger than that of CMIP6 (green color), it means that
CMIP6 has better simulation performance than CMIP5. When the RMSE of CMIP5 is small than that of CMIP6 (red color), it means that CMIP6 has worse simulation
performance than CMIP5.

Figure 2 shows the following: 1) for PRCPTOT, the simulation CMIP6 are mainly concentrated at low latitudes. Same as
ability of CMIP6 improved in CAM, WNA, SAH, TIB, the east PRCPTOT, CMIP6 has no advantage over CMIP5 in CDD
coast of AMZ, the coastal areas of SAF and SSA. However, the simulation.
simulation effect of CMIP6 in SAS the west coast of AMZ and The land-average RMSE values of the sixteen GCMs are listed
WAF is slightly worse than in CMIP5. Overall, CMIP6 has no in Supplementary Table S1 and drawn in Figure 3. Comparing
advantage over CMIP5. 2) For R95pTOT, the overall simulation the land-average RMSE of the GCM, which participates in both
effect of CMIP6 decreases, especially in WAF. 3) For CDD, the CMIP5 and CMIP6, the precipitation indicator simulation
simulation ability of CMIP6 improved in eastern SAH, SAF, and capability of a single GCM in CMIP6 is not enhanced
central South America. However, the simulation ability of CMIP6 compared to that of CMIP5.
is weakened in Northeastern Brazil, central Africa, and central In general, CMIP6 has no overall advantage over CMIP5 in the
SAH. The areas with large differences between CMIP5 and simulation of PRCPTOT and CDD. The performance of CMIP6

Frontiers in Earth Science | www.frontiersin.org 5 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 3 | The land-average RMSE of the PCRPTOT, R95pTOT, CDD in the historical period (1982–2005) of each General Circulation Model (GCM) in CMIP5 and
CMIP6.

increases or decreases regionally in PRCPTOT and CDD. But it is values of the long series precipitation (1982–2100) under
slightly worse than CMIP5 in the simulation of R95pTOT. different future scenarios in CMIP5 and CMIP6. And the
positive β values mean the increasing trend of precipitation
indices and vice versa. In addition, the points in Figures 4–6
Trend Analysis and Comparison of Long represent non-significant trend at the 99% confidence level based
Series Precipitation in CMIP5 and CMIP6 on the MK test. Besides, the global distribution maps of β without
The Mann—Kendall (MK) trend test method is used to analyze showing the significant areas are shown in Supplementary
precipitation series trends. The spatial distributions of the Sen’s Figures S4–S6. And global distribution maps of β difference
slope estimator test (β) for multi-model mean PRCPTOT, between CMIP5 and CMIP6 are also shown in Supplementary
R95pTOT, and CDD are shown in Figures 4–6, which show β Figures S7–S9.

Frontiers in Earth Science | www.frontiersin.org 6 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 4 | The spatial distributions of the Sen’s slope estimator test (β) for multi-model mean PRCPTOT. It shows β values of the long series precipitation
(1982–2100) under different future scenarios in CMIP5 (A–C) and CMIP6 (D–F). The points represent non-significant trend at the 99% confidence level based on the
MK test.

Figure 4 shows that: 1) PRCPTOT shows an increasing Figure 5 shows that: 1) all of the CMIP5 and CMIP6 GCMs
trend from the historical period to the future in many parts of indicate that R95pTOT will most likely increase across most
the world under three future scenarios, especially for SSP585 of World from the historical period to the future under three
(RCP85) scenarios with significant increase for most areas. future scenarios. That is to say, in general R95pTOT increases
Specifically, for SSP245 (RCP45) and SSP585 (RCP85) over wider and decreases over smaller areas than PRCPTOT.
scenarios, there is a significant increasing trend for For example, under the scenarios of SSP585 (RCP85), except
PRCPTOT in most regions of ALA, GRL, WNA, ENA, for a few regions in WAF and SAF, the R95pTOT for almost
NEU, EAF, and the whole Asia. And the results of MK test all land areas (with black dots) shows a significant increasing
show a significant decreasing trend in most parts of CAM and trend. This may lead to more extreme flooding in the world
SAF; 2) Comparing the trend from the historical period to the under global warming scenarios based on the CMIP5 and
future of CMIP5 and CMIP6, the β values in CMIP5 and CMIP6 simulations; 2) for the average extreme precipitation
CMIP6 show roughly similar spatial distributions in most indices (R95pTOT) shown in Figure 4, we can also see that
land areas. There are, however, some exceptions, such as the the trends from the historical period to the future from all
fact that there is an opposite trend in some areas of AMZ and CMIP5 and CMIP6 GCMs display same directions (increase
WAF for CMIP5 and CMIP6. In addition, the differences also or decrease) compared with PRCPTOT for many regions,
exist in the degree of increasing/decreasing trend and which is helpful to discern uniform trend for a particular
coverage area; 3) We can also see more regions with region. However, in some regions of AMZ and SSA where
significant trend based on MK test in CMIP6 compared PRCPTOT decreases, there is a statistically signi?cant
with that of CMIP5. And as the level of radiative forcing increase in R95pTOT for all CMIP5 and CMIP6 future
increases (RCP26/SSP126---RCP45/SSP245----RCP85/ scenarios.
SSP585), there are more regions showing significant change Figure 6 shows that: 1) more regions in the world for CDD
from the historical period to the future. show a significant decreasing trend. Specifically, CDD shows a

Frontiers in Earth Science | www.frontiersin.org 7 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 5 | The spatial distributions of the Sen’s slope estimator test (β) for multi-model mean R95pTOT. It shows β values of the long series precipitation
(1982–2100) under different future scenarios in CMIP5 (A–C) and CMIP6 (D–F). The points represent non-significant trend at the 99% confidence level based on the
MK test.

significant decreasing trend mainly in several regions of NAS and plots for R95pTOT are shown in Figures 9, 10. Box plots for
EAS, WAF and EAF, and SAH. Under the three future scenarios, CDD are shown in Figures 11, 12.
the area with the decreasing trend is larger than the area with an Figures 7, 8 shows that for PRCPTOT, the agreement
increasing trend, indicating that the probability of drought shows between CMIP5 and CMIP6 is good because there are few
a decreasing trend in the future. 2) Comparing the results of regions where the changing rates simulated by CMIP5 and
CMIP5 and CMIP6 from the historical period to and the future, CMIP6 diverge. Under SSP126 (RCP26) and SSP245
the CDD of the two projects is different in some regions. For (RCP45), the far future changes slightly compared with the
example, some areas in AMZ and SSA show an increasing trend near future. However, for SSP585 (RCP85), the far future
for CMIP5 and a decreasing trend in CMIP6; 3) similar spatial changes significantly compared to the near future. PRCPTOT
patterns are shown for CMIP5/CMIP6 future scenarios. In will inevitably increase in the future under the three scenarios,
addition, there are larger areas showing stronger decrease especially in ALA, GRL, NEU, EAF, SAH, EAS, TIB, NAS.
trend for CDD from SSP126 (RCP26) scenario to SSP585 However, not all regions of the world will show an increase in
(RCP85) scenario. PRCPTOT. Compared with CMIP5, CMIP6 changing rate does
not show an overall increase or decrease trend. In some regions,
the changing rate of CMIP6 is greater; in some regions, the
Changing Rate Comparison of Precipitation changing rate of CMIP6 is less. In the SSA and WNA regions,
in the Future Over the History the changing rate of CMIP6 in the future under various
Calculates the multi-model changing rates in each 1° × 1° land scenarios is always greater than that of CMIP5; in the ALA,
grid in the near-future (2028–2060) and far-future periods WAF, SAF regions, the changing rate of CMIP6 in the future
(2068–2100) with respect to the historical period (1982–2014) under various scenarios is always less than that of CMIP5. The
in CMIP5 and CMIP6. Box plots for PRCPTOT multi-model difference between land-grids is relatively large in WAF and
mean changing rates over 21 regions (Table 3) in the near-future SAH; the difference between land-grids is also large in GRL
and far-future periods are shown in Figures 7, 8, respectively. Box under SSP585(RCP85). In EAS, the difference between land-

Frontiers in Earth Science | www.frontiersin.org 8 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 6 | The spatial distributions of the Sen’s slope estimator test (β) for multi-model mean CDD. It shows β values of the long series precipitation (1982–2100)
under different future scenarios in CMIP5 (A–C) and CMIP6 (D–F). The points represent non-significant trend at the 99% confidence level based on the MK test.

grids of CMIP6 is always significantly smaller than that of show a future increase in CDD. Under SSP126 (RCP26) and
CMIP5 under various scenarios. SSP245 (RCP45), the far future changes slightly compared with
Figures 9, 10 shows that for R95pTOT, the trends of CMIP6 the near future. However, under SSP585 (RCP85), the far
and CMIP5 are similar. As the radiative forcing level increases future changes significantly compared to the near future. In
(from RCP26/SSP126, and RCP45/SSP245 to RCP85/SSP585), almost half of the regions, the changing rate of CMIP6 is less
the overall changing rate of R95pTOT indicators increases, and than that of CMIP5 under all scenarios, namely AUS, AMZ,
the number of regions with a changing rate that is greater than 0 SSA, CAM, WNA, CNA, ENA, CAS, TIB. The difference
increases for both projects. Under the SSP585 (RCP85) scenario, between land-grids is relatively large in the AMZ and SEA
R95pTOT in all regions increase. For SSP126 (RCP26) and regions. In the WNA, EAF, SAH, EAS, TIB regions, the
SSP245 (RCP45), the far future changes slightly compared difference between land-grids of CMIP6 is always
with the near future. However, for SSP585 (RCP85), the far significantly smaller than that of CMIP5 under various
future changes significantly compared to the near future. In scenarios. In the GRL, NAS regions, the difference between
the SSA region, the changing rate of CMIP6 in the future land-grids of CMIP6 is always significantly bigger than that of
under various scenarios is always greater than that of CMIP5; CMIP5 under various scenarios.
in the region ALA, EAS, NAS regions, the changing rate of
CMIP6 in the future under various scenarios is always less
than that of CMIP5. The difference between land-grids is Comparison of Model Uncertainty in CMIP5
relatively large in WAF, EAF, and SAH. In the AMZ, WNA, and CMIP6 Precipitation
ALA, WAF, SAH regions, the difference between land-grids of Figures 13–15 show the data series of land-average PRCPTOT,
CMIP6 is always bigger than that of CMIP5 under various R95pTOT, and CDD from 1982 to 2100, respectively. The orange
scenarios. and purple solid lines represent the multi-model mean results of
Figures 11, 12 shows that for CDD, the changing rates CMIP5 and CMIP6, respectively. The upper and lower limits of
between the 21 regions differ significantly. Nearly half of the the orange and purple uncertainty intervals represent the 75 and
regions experience a future decline in CDD, and other regions 25% quantiles, respectively.

Frontiers in Earth Science | www.frontiersin.org 9 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 7 | Box plots for PRCPTOT multi-model mean changing rates in 21 regions in near-future period (2028–2060) over the historical period (1982–2005) in
CMIP5 and CMIP6. CMIP5 is presented in red and CMIP6 is presented in blue.

Frontiers in Earth Science | www.frontiersin.org 10 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 8 | Box plots for PRCPTOT multi-model mean changing rates in 21 regions in far-future period (2068–2100) over the historical period (1982–2005) in
CMIP5 and CMIP6. CMIP5 is presented in red and CMIP6 is presented in blue.

Frontiers in Earth Science | www.frontiersin.org 11 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 9 | Box plots for R95pTOT multi-model mean changing rates in 21 regions in near-future period (2028–2060) over the historical period (1982–2005) in
CMIP5 and CMIP6. CMIP5 is presented in red and CMIP6 is presented in blue.

Frontiers in Earth Science | www.frontiersin.org 12 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 10 | Box plots for R95pTOT multi-model mean changing rates in 21 regions in far-future period (2068–2100) over the historical period (1982–2005) in
CMIP5 and CMIP6. CMIP5 is presented in red and CMIP6 is presented in blue.

Frontiers in Earth Science | www.frontiersin.org 13 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 11 | Box plots for CDD multi-model mean changing rates in 21 regions in near-future period (2028–2060) over the historical period (1982–2005) in CMIP5
and CMIP6. CMIP5 is presented in red and CMIP6 is presented in blue.

Frontiers in Earth Science | www.frontiersin.org 14 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 12 | Box plots for CDD multi-model mean changing rates in 21 regions in far-future period (2068–2100) over the historical period (1982–2005) in CMIP5
and CMIP6. CMIP5 is presented in red and CMIP6 is presented in blue.

Frontiers in Earth Science | www.frontiersin.org 15 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 13 | Global land-average multi-model mean PRCPTOT during 1982–2100 in CMIP5 and CMIP6. The shaded region shows the uncertainty range
(between the 25 quantiles and the 75 quantiles).

For PRCPTOT and R95pTOT, both CMIP5 and CMIP6 show a However, CDD shows a slight decreasing trend from the
significant increasing trend from the historical period to the end of historical period to the end of the 21st century in both
the 21st century. There are similar fluctuations and trends under the CMIP5 and CMIP6. The multi-mode mean CDD of
corresponding future scenarios (RCP26 and SSP126; RCP45 and CMIP6 is slightly higher than that of CMIP5. The model
SSP245; RCP85 and SSP585) for CMIP5 and CMIP6. As the level of uncertainty range of CDD in CMIP6 is smaller than that in
radiative forcing increases (RCP26/SSP126---RCP45/SSP245---- CMIP5 (Figure 15).
RCP85/SSP585), the increasing trend becomes more apparent.
The increase in PRCPTOT is greater than that in R95pTOT. The
multi-mode mean PRCPTOT and R95pTOT of CMIP6 are CONCLUSION
significantly higher than those of CMIP5. The model uncertainty
range of PRCPTOT and R95pTOT in CMIP6 is larger than that in (1) CMIP6 has no overall advantage over CMIP5 in the
CMIP5, with a small overlap area (Figures 13, 14). simulation of PRCPTOT. The performance of CMIP6 of

Frontiers in Earth Science | www.frontiersin.org 16 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 14 | Global land-average multi-model mean R95pTOT during 1982–2100 in CMIP5 and CMIP6. The shaded region shows the uncertainty range (between
the 25 quantiles and the 75 quantiles).

PRCPTOT increases or decreases regionally. For the fact that there is an opposite trend in some areas of AMZ and
simulation of R95pTOT, the overall effect of CMIP6 is WAF for CMIP5 and CMIP6. The differences between the
weaker than that of CMIP5, especially in WAF. Same as R95pTOT trends mainly noticeable in South America and
PRCPTOT, CMIP6 has no advantage over CMIP5 in CDD WAF. The CDD trend of the two projects is different in some
simulation. The areas with large differences of CDD between regions. For example, some areas in AMZ and SSA show an
CMIP5 and CMIP6 are mainly concentrated at low latitudes. increasing trend for CMIP5 and a decreasing trend in CMIP6.
(2) The MK trend test method and Sen’s slope estimator test is used (3) Analyzing the changing rates of the 21 regions in the near-future
to analyze the trends of long series precipitation. Overall, there are period and far-future period over the historical period, it can be
more areas with significant trend based on MK test in CMIP6 found that in most regions, the simulated changing rates of
compared with that of CMIP5. The β values of PRCPTOT in CMIP6 have no obvious change direction compared with
CMIP5 and CMIP6 show roughly similar spatial distributions in CMIP5. However, in some regions, CMIP6 has a steady
most land areas. There are, however, some exceptions, such as the upward or decreasing trend compared to CMIP5. In the SSA

Frontiers in Earth Science | www.frontiersin.org 17 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

FIGURE 15 | Global land-average multi-model mean CDD during 1982–2100 in CMIP5 and CMIP6. The shaded region shows the uncertainty range (between the
25 quantiles and the 75 quantiles).

and WNA regions, the PRCPTOT changing rate of CMIP6 in (4) For PRCPTOT and R95pTOT, the uncertainty of CMIP6 is
the future under various scenarios is always greater than that of larger than that of CMIP5; for CDD, the uncertainty of
CMIP5; in the ALA, WAF, SAF regions, the PRCPTOT CMIP6 is smaller than that of CMIP5.
changing rate of CMIP6 in the future under various
scenarios is always less than that of CMIP5. In the SSA
region, the R95pTOT changing rate of CMIP6 in the future DATA AVAILABILITY STATEMENT
under various scenarios is always greater than that of CMIP5; in
the ALA, EAS, NAS regions, the R95pTOT changing rate of Publicly available datasets were analyzed in this study. This data can be
CMIP6 in the future under various scenarios is always less than found here GPCC: https://climatedataguide.ucar.edu/climate-data/
that of CMIP5. In almost half of the regions, the changing rate of gpcc-global-precipitation-climatology-centre CMIP5: https://esgf-
CMIP6 is less than that of CMIP5 under all scenarios, namely node.llnl.gov/search/cmip5/ CMIP6: https://esgf-node.llnl.gov/search/
AUS, AMZ, SSA, CAM, WNA, CAN, ENA, CAS, TIB. cmip6.

Frontiers in Earth Science | www.frontiersin.org 18 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

AUTHOR CONTRIBUTIONS FUNDING

JL, RH, and HC carried out the study design. YZ and TZ This work was supported by the National Key Research and
download the CMIP5 CMIP6, and GPCC data. JL, RH, TZ, Development Program (No. 2017YFA0603702)
and YZ carried out the calculation. HC was responsible for
supervision, project administration and funding acquisition. JL SUPPLEMENTARY MATERIAL
and RH checked the data processing progress. RH and HC
reviewed and edited the manuscript after JL wrote the original The Supplementary Material for this article can be found online at:
draft. All authors contributed to the article and approved the final https://www.frontiersin.org/articles/10.3389/feart.2021.687976/
manuscript. full#supplementary-material

Report of the Intergovernmental Panel on Climate Change. Cambridge:

REFERENCES Cambridge University Press.
Jiang, D., Hu, D., Tian, Z., and lang, X. (2020). Differences between CMIP6 and
Birkmann, J. (2011). “Regulation and Coupling of Society and Nature in the CMIP5 Models in Simulating Climate over China and the East Asian Monsoon.
Context of Natural Hazards,” in Coping with Global Environmental Change, Adv. Atmos. Sci. 37 (10), 1102–1118. doi:10.1007/s00376-020-2034-y
Disasters and Security Berlin, Heidelberg: Springer, 1103–1127. doi:10.1007/ Kendall, M. G. (1990). Rank Correlation Methods. Br. J. Psychol. 25 (1), 86–91.
978-3-642-17776-7_68 doi:10.1111/j.2044-8295.1934.tb00727.x
Bracegirdle, T. J., Holmes, C. R., Hosking, J. S., Marshall, G. J., Osman, M., Khan, A., Koch, M., and Tahir, A. (2020). Impacts of Climate Change on the Water
Patterson, M., et al. (2020). Improvements in Circumpolar Southern Availability, Seasonality and Extremes in the Upper Indus Basin (UIB).
Hemisphere Extratropical Atmospheric Circulation in CMIP6 Compared to Sustainability 12 (4), 1283. doi:10.3390/su12041283
CMIP5. Earth Space Sci. 7 (6), 1065. doi:10.1029/2019EA001065 Kundzewicz, Z. W., Kanae, S., Seneviratne, S. I., Handmer, J., Nicholls, N., Peduzzi,
Chen, H., Sun, J., Lin, W., and Xu, H. (2020). Comparison of CMIP6 and CMIP5 P., et al. (2014). Flood Risk and Climate Change: Global and Regional
Models in Simulating Climate Extremes. Sci. Bull. 65 (17), 1415–1418. doi:10. Perspectives. Hydrological Sci. J. 59 (1), 1–28. doi:10.1080/02626667.2013.
1016/j.scib.2020.05.015 857411
Davy, R., and Outten, S. (2020). The Arctic Surface Climate in CMIP6: Status and Lal, P., Prakash, A., Kumar, A., Srivastava, P. K., Saikia, P., Pandey, A. C., et al.
Developments since CMIP5. J. Clim. 33 (18), 8047–8068. doi:10.1175/JCLI-D- (2020). Evaluating the 2018 Extreme Flood hazard Events in Kerala, India.
19-0990.1 Remote Sensing Lett. 11 (5), 436–445. doi:10.1080/2150704X.2020.1730468
Fan, X., Miao, C., Duan, Q., Shen, C., and Wu, Y. (2020). The Performance of Li, L., Shen, M., Hou, Y., Xu, C.-Y., Lutz, A. F., Chen, J., et al. (2019). Twenty-first-
CMIP6 versus CMIP5 in Simulating Temperature Extremes over the Global century Glacio-Hydrological Changes in the Himalayan Headwater Beas River
Land Surface. J. Geophys. Res. Atmos. 125 (18), 33031. doi:10.1029/ basin. Hydrol. Earth Syst. Sci. 23 (3), 1483–1503. doi:10.5194/hess-23-1483-
2020JD033031 2019
Forestieri, A., Arnone, E., Blenkinsop, S., Candela, A., Fowler, H., and Noto, L. V. Madsen, H., Lawrence, D., Lang, M., Martinkova, M., and Kjeldsen, T. R. (2014).
(2018). The Impact of Climate Change on Extreme Precipitation in Sicily, Italy. Review of Trend Analysis and Climate Change Projections of Extreme
Hydrological Process. 32 (3), 332–348. doi:10.1002/hyp.11421 Precipitation and Floods in Europe. J. Hydrol. 519, 3634–3650. doi:10.1016/
Fuchs, T., Schneider, U., and Rudolf, B. (2009). The Global Precipitation j.jhydrol.2014.11.003
Climatology Centre (GPCC)-In Situ Observation Based Precipitation Mann, H. B. (1945). Nonparametric Tests Against Trend. Econometrica 13,
Climatology on Regional and Global Scale. EGUGA 11, 10519. 245–259. doi:10.2307/1907187
Funk, C., Davenport, F., Eilerts, G., Nourey, N., and Galu, G. (2018). Contrasting Meehl, G. A., Boer, G. J., Covey, C., Latif, M., and Stouffer, R. J. (2000). The
Kenyan Resilience to Drought: 2011 and 2017. USAID Special Rep. Coupled Model Intercomparison Project (CMIP). Bull. Amer. Meteorol. Soc. 81
Funk, C., Shukla, S., Thiaw, W. M., Rowland, J., Hoell, A., McNally, A., et al. (2019). (2), 313–318. doi:10.1175/1520-0477(2000)081<0313:tcmipc>2.3.co;2
Recognizing the Famine Early Warning Systems Network: Over 30 Years of Moss, R. H., Edmonds, J. A., Hibbard, K. A., Manning, M. R., Rose, S. K., Van
Drought Early Warning Science Advances and Partnerships Promoting Global Vuuren, D. P., et al. (2010). The Next Generation of Scenarios for Climate
Food Security. Bull. Am. Meteorol. Soc. 100 (6), 1011–1027. doi:10.1175/BAMS- Change Research and Assessment. Nature 463 (7282), 747–756. doi:10.1038/
D-17-0233.1 nature08823
Gao, L., Tao, B., Miao, Y., Zhang, L., Song, X., Ren, W., et al. (2019). A Global Ortiz-Gómez, R., Muro-Hernández, L. J., and Flowers-Cano, R. S. (2020).
Dataset for Economic Losses of Extreme Hydrological Events during 1960- Assessment of Extreme Precipitation through Climate Change Indices in
2014. Water Resour. Res. 55 (6), 5165–5175. doi:10.1029/2019WR025135 Zacatecas, Mexico. Theor. Appl. Climatol 141 (3), 1541–1557. doi:10.1007/
Giorgi, F., and Francisco, R. (2000). Uncertainties in Regional Climate Change s00704-020-03293-2
Prediction: A Regional Analysis of Ensemble Simulations with the HADCM2 Rahmani, V., and Harrington, J. (2019). Assessment of Climate Change for
Coupled AOGCM. Clim. Dyn. 16 (2), 169–182. doi:10.1007/PL00013733 Extreme Precipitation Indices: A Case Study from the central United States.
Gusain, A., Ghosh, S., and Karmakar, S. (2020). Added Value of CMIP6 over Int. J. Climatol 39 (2), 1013–1025. doi:10.1002/joc.5858
CMIP5 Models in Simulating Indian Summer Monsoon Rainfall. Atmos. Res. Swain, S. (2014). Impact of Climate Variability over Mahanadi River Basin. Int.
232, 104680. doi:10.1016/j.atmosres.2019.104680 J. Eng. Res. Tech. 3 (7), 938–943. doi:10.1109/INSS.2009.5409957
Hamed, K. H. (2008). Trend Detection in Hydrologic Data: The Mann-Kendall Ukkola, A. M., De Kauwe, M. G., Roderick, M. L., Abramowitz, G., and Pitman, A.
Trend Test under the Scaling Hypothesis. J. Hydrol. 349 (3-4), 350–363. doi:10. J. (2020). Robust Future Changes in Meteorological Drought in CMIP6
1016/j.jhydrol.2007.11.009 Projections Despite Uncertainty in Precipitation. Geophys. Res. Lett. 47 (11),
Hosseinzadehtalaei, P., Tabari, H., and Willems, P. (2020). Climate Change 820. doi:10.1029/2020GL087820
Impact on Short-Duration Extreme Precipitation and Intensity-Duration- Ulbrich, U., Brücher, T., Fink, A. H., Leckebusch, G. C., Krüger, A., and Pinto, J. G.
Frequency Curves over Europe. J. Hydrol. 590, 125249. doi:10.1016/j.jhydrol. (2003). The central European Floods of August 2002: Part 1 - Rainfall Periods
2020.125249 and Flood Development. Weather 58 (10), 371–377. doi:10.1256/wea.61.03A
IPCC (2007). IPCC Fourth Assessment Report: Working Group II Report "Impacts, Van Vuuren, D. P., Kriegler, E., O’Neill, B. C., Ebi, K. L., Riahi, K., Carter, T. R.,
Adaptation and Vulnerability. Cambridge: Cambridge University Press. et al. (2014). A New Scenario Framework for Climate Change Research:
IPCC (2013). Summary for Policymakers Climate Change 2013: The Physical Scenario Matrix Architecture. Climatic Change 122 (3), 373–386. doi:10.
Science Basis. Contribution of Working Group I to the Fifth Assessment 1007/s10584-013-0906-1

Frontiers in Earth Science | www.frontiersin.org 19 June 2021 | Volume 9 | Article 687976

Li et al. Comparation of CMIP6 and CMIP5

Verma, M. K., Verma, M. K., and Swain, S. (2016). Statistical Analysis of Zhang, X., Alexander, L., Hegerl, G. C., Jones, P., Tank, A. K., Peterson, T. C., et al.
Precipitation over Seonath River Basin, Chhattisgarh, India. Int. J. Appl. (2011). Indices for Monitoring Changes in Extremes Based on Daily Temperature
Eng. Res. 11 (4), 2417–2423. and Precipitation Data. Wires Clim. Change 2 (6), 851–870. doi:10.1002/wcc.147
Veronika, E., Sandrine, B., Meehl, G. A., Senior, C. A., Bjorn, S., Stouffer, R. J., et al. Zhu, H., Jiang, Z., Li, J., Li, W., Sun, C., and Li, L. (2020). Does CMIP6 Inspire More
(2016). Overview of the Coupled Model Intercomparison Project Phase 6 Confidence in Simulating Climate Extremes over China?. Adv. Atmos. Sci. 37
(CMIP6) Experimental Design and Organization. Geoentific Model. Dev. 9 (10), 1119–1132. doi:10.1007/s00376-020-9289-1
(5), 1937–1958. doi:10.5194/gmd-9-1937-2016 Zhu, Y.-Y., and Yang, S. (2020). Evaluation of CMIP6 for Historical
Wang, Y., Xu, Y., Lei, C., Li, G., Han, L., Song, S., et al. (2016). Spatio-temporal Temperature and Precipitation over the Tibetan Plateau and its
Characteristics of Precipitation and Dryness/wetness in Yangtze River Delta, Comparison with CMIP5. Adv. Clim. Change Res. 11 (3), 239–251.
Eastern China, during 1960-2012. Atmos. Res. 172–173, 196–205. doi:10.1016/j. doi:10.1016/j.accre.2020.08.001
atmosres.2016.01.008 Zobel, Z., Wang, J., Wuebbles, D. J., and Kotamarthi, V. R. (2018). Analyses for
Wyser, K., van Noije, T., Yang, S., von Hardenberg, J., O’Donnell, D., and Döscher, High-Resolution Projections Through the End of the 21st Century for
R. (2020). On the Increased Climate Sensitivity in the EC-Earth Model from Precipitation Extremes over the United States. Earth’s Future 6 (10),
CMIP5 to CMIP6. Geosci. Model. Dev. 13 (8), 3465–3474. doi:10.5194/gmd-13- 1471–1490. doi:10.1029/2018EF000956
3465-2020
Xin, X., Wu, T., Zhang, J., Yao, J., and Fang, Y. (2020). Comparison of CMIP6 and Conflict of Interest: The authors declare that the research was conducted in the
CMIP5 Simulations of Precipitation in China and the East Asian Summer absence of any commercial or financial relationships that could be construed as a
Monsoon. Int. J. Climatol 40, 6423–6440. doi:10.1002/joc.6590 potential conflict of interest.
Yuanhai, F. U., Lin, Z., and Guo, D. (2020). Improvement of the Simulation of the
Summer East Asian westerly Jet from CMIP5 to CMIP6. Atmos. Oceanic Ence Copyright © 2021 Li, Huo, Chen, Zhao and Zhao. This is an open-access article
Lett. 13 (6), 1–9. doi:10.1080/16742834.2020.1746175 distributed under the terms of the Creative Commons Attribution License (CC BY).
Zamani, Y., Hashemi Monfared, S. A., Azhdari moghaddam, M., and The use, distribution or reproduction in other forums is permitted, provided the
Hamidianpour, M. (2020). A Comparison of CMIP6 and CMIP5 original author(s) and the copyright owner(s) are credited and that the original
Projections for Precipitation to Observational Data: the Case of publication in this journal is cited, in accordance with accepted academic practice.
Northeastern Iran. Theor. Appl. Climatol 142 (3–4), 1613–1623. doi:10.1007/ No use, distribution or reproduction is permitted which does not comply with
s00704-020-03406-x these terms.

Frontiers in Earth Science | www.frontiersin.org 20 June 2021 | Volume 9 | Article 687976