FOCRAII, Beijing, 8 April 2013 Changes in global and regional monsoon precipitation projected by CMIP5 models Akio Kitoh Kitoh, A., H. Endo, K. Krishna Kumar, I.F.A. Cavalcanti, P. Goswami, T. Zhou, 2013: Monsoons in a changing world: a regional perspective in a global context. J. Geophys. Res., 118, doi:10.1002/jgrd.50258.

Introduction Monsoons are responsible for the majority of summer rainfall within the tropics, where billions of people depend on the monsoon rainfall. Thus, reliable future projection for monsoon rainfall, especially on a regional-scale, has been strongly demanded. Provides a latest view of global and regional monsoonal rainfall changes as projected by CMIP5 multi-models. Investigates not only mean precipitation but also some precipitation extreme indices and monsoon seasonality.

Previous studies of the global monsoon Historical records Decreasing trend in the global land monsoon precipitation over the last half of the 20th century. Wand and Ding (2006); Zhou et al. (2008) Increasing trend in the combined (oceanic and land) monsoon precipitation for recent decades (1979-2008) Wang et al. (2012) Future projections Increase of global monsoon area and precipitation intensity. CMIP3: Hsu et al. (2012) CMIP5: Lee and Wang (2012). Hsu et al. (2013)

CMIP5 model simulations Analyzed experiments 20th century historical simulations 21st century projections under RCP4.5/RCP8.5 scenarios Analyzed models Monthly output: 29 models Daily output: 21 models

CMIP5 climate model

Global monsoon Present-day: 1986-2005 Future: 2080-2099

Global monsoon domain Model mean generally reproduces the observed domain Some biases over eastern Asia and the tropical Pacific ・最初に、モンスーン領域の再現性について示します。 ・本研究では、雨季と乾季の降水量の差が2.5mm/day以上の地域をモンスーン領域と定義しました。 ・実線が観測のＧＰＣＰ，陰影が２９モデル平均降水量をもとに定義したモンスーン領域です。 ・モデル平均は観測の分布を良く再現しています。 ・しかし、東アジアや熱帯太平洋で少し目立つバイアスがある。 : GPCP : 29 model mean Monsoon domain: Annual range >= 2.5 mm day-1 Annual range: Difference between MJJAS and NDJFM 　 (Wang et al., 2011)

Global monsoon domain Expansion over the central to eastern tropical Pacific, the southern Indian Ocean, and eastern Asia. ・次に、モンスーン領域の将来変化を示します。 （アニメーション） ・将来気候は、RCP8.5シナリオの2080～2099年です。 ・将来気候で新たにモンスーン域となる領域を濃い青色、モンスーン領域から外れる領域を橙色で示します。 ・将来気候では、モンスーン領域は全般的には拡大傾向にある。 ・特に、東アジア、熱帯太平洋中部・東部、南インド洋などで拡大する。 Reduction Expansion Period: 2080 to 2099 Monsoon domain: Annual range >= 2.5 mm day-1 Annual range: Difference between MJJAS and NDJFM 　 (Wang et al., 2011)

Global monsoon precipitation Multi-model ensemble matches the observations Monsoon-related precipitation will remarkably increase → consistent with other studies analyzing CMIP3/CMIP5 models by Hsu et al. (2012, 2013); Lee and Wang (2012) RCP4.5　 RCP8.5 Change ratio [%] GMI Historical RCP4.5 RCP8.5 ×: GPCP/CMAP GMA GMI GMP GMA (ratio to the globe) GMA: Global monsoon (GM) area GMP: GM summer total precipitation GMI: GM intensity (= GMP/GMA) Large marks: Mean of all models

Time series of global land monsoon Decrease from 1950s to 1980s, and increase in the 21st C The simulated trend is consistent with the observations but with less amplitude RCP4.5 RCP8.5 Thick line：29 model mean Shading：29 model spread (10th-90th percentile) Historical-run Observations Observational data CRU-TS3.1 GPCC-v6 GPCC-VASClimO CMAP GPCP Change ratio to 1986-2005 av. [%] 20-year running mean ・次に、陸上モンスーン域の降水量の過去から将来の変化について示します。 ・過去の期間については、いくつかの観測データをプロットしています。 ・1986～2005年に対する比を２０年移動平均で示しています。 ・太線はモデル平均値、陰影はモデル間のばらつきを示しています。 ・観測データでは、1950年～1980年代にかけて降水が減少し、その後は増加傾向にあります。 ・モデルはこのような傾向を定性的には再現しています。 ・過去の減少トレンドは、エアロゾル増加の影響が考えられます。今後詳しく調べる予定です。

Time series of NH/SH land monsoon Large differences in the amplitude between the NH and SH Model spread over the SH in the future projections is larger than that over NH Change ratio to 1986-2005 av. [%] N. H. land monsoon S. H. land monsoon

Attribution of the decrease in the 20th C Neither experiments with GHG forcing nor natural forcing reproduce the decreased trend simulated by the all forcing → aerosol effect is essential ? Thick line: 14 model mean Shading: Inter-model spread (S.D.) HistoricalGHG： GHG forcing only HistoricalNat： Natural forcing only Historical： All forcing Change rate to 1861-1880 av. [%] 20-year running mean 14model = “CNRM-CM5 CSIRO-Mk3-6-0 CanESM2 GFDL-CM3 GFDL-ESM2M GISS-E2-H GISS-E2-R HadGEM2-ES IPSL-CM5A-LR MIROC-ESM-CHEM MIROC-ESM MRI-CGCM3 NorESM1-M bcc-csm1-1"

Attribution of the decrease in the 20th C Land monsoon precipitation in the NH shows larger response both to the GHG forcing and the all forcing N. H. land monsoon S. H. land monsoon Change rate to 1861-1880 av. [%] HistoricalGHG： GHG forcing only HistoricalNat： Natural forcing only Historical： All forcing

Regional monsoon Present-day: 1986-2005 Future: 2080-2099

Definition of indices (1) Mean/extreme precipitation indices Pav: Average precipitation [mm day-1] SDII: Simple precipitation daily intensity index [mm day-1] Total precipitation is divided by the wet days R5d: Seasonal maximum 5-day precipitation total [mm] CDD: Seasonal maximum consecutive dry days [days] Calculated for the summer season. (MJJAS in NH and NDJFM in SH) Calculated over the monsoon domain at the present-day by each model.

Definition of indices (2) Monsoon onset/retreat date (Wang and LinHo, 2002) Smooth regionally averaged climatological daily precipitation using its first 12 harmonics. Remove the dry month mean precipitation (January in NH and July in SH) from it.  Relative climatological precipitation (RCP) [mm/day] ● ● RCP DUR 5 mm/dy JAN. av. ONS RET

Validation of the indices Mean/extreme precipitation are generally reproduced well Monsoon seasonality is more difficult to simulate Pav: Average precipitation SDII: Precipitation intensity　 R5d: Seasonal maximum of 5-day precipitation CDD: Seasonal maximum of consecutive dry days ONS: Onset date RET: Retreat date DUR: Monsoon duration × TRMM-3B42 △ GPCP-1DD (2.0 deg. lat/lon)

Future change of the precipitation indices Largest increases over the Asian monsoon domains Large increases in extremes over America and Africa [%] East Asia (EAS) RCP4.5　RCP8.5 Pav: Average precipitation SDII: Precipitation intensity　 R5d: Seasonal maximum of 5-day precipitation CDD: Seasonal maximum of consecutive dry days N. America (NAM) N. Africa (NAF) S. Asia (SAS) ・次に、地域別の降水の将来変化について示します。 ・縦軸は変化率で、左上の図で示した７地域における、平均降水量、平均降水強度、季節最大５日間降水量、無降水日数を示します。 ・これら降水指数は概ね増加傾向にありますが、地域間の違いは大きいことが分かります。 （クリック） ・地域別では、アジアモンスーン地域における増加が顕著です。 ・北アメリカや北アフリカのモンスーン地域では、平均降水量はほとんど変化しませんが、極端な降水は大幅に増加することが予測されています。 S. America（SAM） S. Africa（SAF） Australia （AUS）

Future change of the precipitation indices Change rate of indices in the 50th percentile

Change of the monsoon seasonality Onset: 　 Earlier or not changed Retreat: Delayed ⇒ longer summer monsoon season [days] East Asia (EAS) Delayed Earlier N. America (NAM) N. Africa (NAF) S. Asia (SAS) RCP4.5　RCP8.5 ONS: Onset date RET: Retreat date DUR: Monsoon duration ・次に、地域別の雨季の変化について示します。 ・縦軸は変化日数で、左から、雨季入り日、明け日、雨季の期間を示します。 ・ここで、雨季の入りと明けを、気候平均日降水量を用いて定義しました。 ・雨季の入り時期は、一部の地域で早くなる傾向にあります。 ・雨季の明け時期は、多くの地域で遅れる傾向にあります。 ・結果として、多くの地域で雨季の期間が長くなる傾向にあります。 S. America（SAM） S. Africa（SAF） Australia （AUS）

Attributions of changes in monsoon rainfall Enhanced moisture flux conv. due to increased moisture ⇒ increased monsoon rainfall However, large differences among the monsoon regions Changes with consensus of more than 75% of models Future change (RCP8.5) mm day-1 ∆ 𝑞𝑽 & −∆ 𝛻∙ 𝑞𝑽 land monsoon [mm day-1] Anomaly −∆ 𝛻∙ 𝑞𝑽 𝐿𝑇 land monsoon [10-3 kg m-2 s-1] Anomaly −∆ 𝛻∙𝑽 𝐿𝑇 < > LT: Integration in 1000-500hPa

Summary Global monsoon Regional monsoon Decreased monsoon rainfall from the 1950s to 1980s both in observations and simulations. Remarkable increase in monsoon precipitation in the 21st century Regional monsoon Asian monsoon : largest increase both in mean and heavy precipitation. American and African monsoon : large increase in heavy precipitation, despite modest change in mean precipitation. Longer summer monsoon season in many regions.

Thank you ! “Harerun” mascot of JMA Reference: Kitoh, A., H. Endo, K. K. Kumar, I. F. A. Cavalcanti, P. Goswami, and T. Zhou, 2013, Monsoons in a changing world: a regional perspective in a global context. J. Geophys. Res., doi: 10.1002/jgrd.50258.

Attribution of the decreased rainfall in the 20th C land NH land Thick line: 14 model mean Shading: Inter-model spread (S.D.) HistoricalGHG HistoricalNat Historical ocean SH land

Change of the monsoon seasonality [days] Delayed Earlier For calculation of the global mean, the 7 monsoon domain statistics are averaged with weighting based on their area in the present-day.

Attribution of changes in monsoon rainfall ∆𝑃=−∆ 𝛻∙ 𝑞𝑽 +∆𝐸 [mm/day] [mm/day] −∆ 𝛻∙ 𝑞𝑽 𝐿𝑇 [mm/day] [10-3kg m-2 s-1] −∆ 𝛻∙𝑽 𝐿𝑇

−∆ 𝛻∙ 𝑞𝑽 𝐿𝑇 −∆ 𝛻∙ 𝑞𝑽 land monsoon −∆ 𝛻∙ 𝑞𝑽 𝐿𝑇 < > LT: Vertical integration in 1000-500hPa land monsoon −∆ 𝛻∙ 𝑞𝑽 < > : Vertical integration in 1000-100hPa

∆𝑃=−∆ 𝛻∙ 𝑞𝑽 +∆𝐸 MJJAS NDJFM Future change (RCP8.5) ∆𝑃 −∆ 𝛻∙ 𝑞𝑽 ∆𝐸 ∆𝑃=−∆ 𝛻∙ 𝑞𝑽 +∆𝐸 MJJAS NDJFM ∆𝑃 −∆ 𝛻∙ 𝑞𝑽 ∆𝐸 < > : Vertical integration in the troposphere (1000hPa-100hPa) Only changes with consensus of more than 75% of models are shown

∆𝑃=−∆ 𝛻∙ 𝑞𝑽 +∆𝐸 MJJAS NDJFM Future change (RCP8.5) ∆𝑃 −∆ 𝛻∙ 𝑞𝑽 𝐿𝑇 ∆𝑃=−∆ 𝛻∙ 𝑞𝑽 +∆𝐸 MJJAS NDJFM ∆𝑃 −∆ 𝛻∙ 𝑞𝑽 𝐿𝑇 −∆ 𝛻∙𝑽 𝐿𝑇 < > LT: Vertical integration in the lower troposphere (1000hPa-500hPa) Only changes with consensus of more than 75% of models are shown

Global Land monsoon domain