reveal.js

All slides are available at [www.pik-potsdam.de/~menz/CapTainRain/Webinar](http://www.pik-potsdam.de/~menz/CapTainRain/Webinar)

Introduction into regional climate scenarios and sensitivities of heavy rainfall indicators

Christoph Menz

RDII

Potsdam Institut for Climate Impact Research

### Table of Content - Introduction - Climate Change - Climate Change from Global Climate Models Persepctive - Climate Change from Regional Climate Models Persepctive - Brief Introduction to Bias Adjustment - Projected Precipitation Change in Jordan

# Introduction - Climate Change

## Assessment Reports ### WG1: Physical Science Basis

1990

1995

2001

2007

2013

2021

414P

588P

893P

1007P

1552P

3175P

### Temperature

### Precipitation

Global mean annual changes 1971-2100 vs. 1971-2000 IPCC AR5: CMIP5 IPCC AR6: CMIP6 RCP2.6/SSP1-2.6: Sustainable development RCP8.5/SSP5-8.5: Fossil-fueled development

### Temperature

### Precipitation

Global mean annual changes 2071-2100 vs. 1971-2000 IPCC AR6: CMIP6 SSP5-8.5: Fossil-fueled development

# Climate Change from Global Climate Models Perspective

\begin{equation} \nonumber \rho \frac{d \vec{\mathbf{v}}}{d t} = - \vec \nabla p + \rho g -2\vec{\mathbf{\Omega}} \times (\rho \vec{\mathbf{v}}) - \vec \nabla \overleftrightarrow{\mathbf{t}} \end{equation} \begin{equation} \nonumber \frac{d \rho}{d t} = - \rho \vec \nabla \vec{\mathbf{v}} \end{equation} \begin{equation} \nonumber \rho \frac{d q^x}{dt} = - \vec \nabla \vec{\mathbf{J}}^x + I^x \end{equation} \begin{equation} \nonumber \rho \frac{d e}{d t}=-p \vec \nabla \vec{\mathbf{v}} - \vec \nabla (\vec{\mathbf{J}_e} + \vec{\mathbf{R}})+\epsilon \end{equation}

## Development of Global Models

#### RCP: Representative Concentration Pathways #### SSP: Shared Socioeconomic Pathways - Scenarios representing greenhouse gas concentrations in the atmosphere - Narratives for future socio-economic evolution - No adaptation/mitigation to climate change - Similar in CMIP5 and CMIP6 - van Vuuren et al. (2011) and Riahi et al. (2017)

Historical

Observations

RCP2.6

SSP1-2.6

Peak around 2040 (490ppm)

400ppm in 2100

RCP4.5

SSP2-4.5

moderate increase

650ppm in 2100

RCP8.5

SSP5-8.5

strongest increase

1370ppm in 2100

## Temperature vs Precipitation 2071-2100 vs. 1971-2000 - annual mean

Global

Amman

## Distribution of Precipitation

# Climate Change from Regional Climate Models Persepctive

## Evolution of daily mean precipitation

## GCM

## RCM

### CORDEX - Coordinated Regional Downscaling Experiment - Description for 14 Regions accross globe - Most Regions in different resolution (~50km, ~25km and ~12.5km)

#### Boundary Problem - RCMs are limited area models driven at boundaries by external forcing (usually Reanalysis or GCM data) - Discrepancy in spatial and temporal resolution of RCM and driver - Leads to boundary effects - Boundary zone coupling: \begin{equation} \nonumber \psi^{n+1} = \psi^* - \alpha_b (\psi^* - \psi^{n+1}_b) \end{equation} Davis (1976, 1983)

Spatial: ~250km → ~25km Temporal: ~6h → ~0.1h

#### Which should be used for Jordan?

RCP 2.6

RCP 4.5

RCP 8.5

CORDEX-EUR44
CORDEX-EUR11

~50km
~12km

11
18

21
17

26
37

CORDEX-MNA44
CORDEX-MNA22

~50km
~25km

1
-

4
1

4
2

CORDEX-MED11

~12km

# Model BIAS

### Temperature Bias - Annual - Ensemble Median - 1981-2010

### CORDEX-MNA44

### CORDEX-MNA22

### CORDEX-EUR11

### Precipitation Bias - Annul - Ensemble Median - 1981-2010

### CORDEX-MNA44

### CORDEX-MNA22

### CORDEX-EUR11

### Temperature and Precipitation Bias Seasonality

1971-2005 Observation: E-OBS v19.0e 37 CORDEX-EUR11 Models

- Underestimation of wet season (NDJFM) temperature and precipitation - Overestimation of dry season (JJAS) temperature and precipitation - Precipitation bias up to 1/5 in wet season and twice pre and post dry season observation

# Brief Introduction to Bias Adjustment

## Example of Model Bias

JJA Temperatures, Bangladesh

## Example of Model Bias

JJA 1971-2000, Amman

## Example of Model Bias

2021-2050
2071-2100

Boberg and Christensen (2012)

## Bias Adjustment - Workflow

#### Chosen Bias Adjustment Method - ISIMIP3BASD method used for bias adjustment - Parametric quantile mapping - Trend preserving

1981-2010

Annual - Ensemble Mean

### Temperature Bias

### CORDEX-MNA44

### CORDEX-MNA22

### CORDEX-EUR11

#### non-adjusted

#### adjusted

1981-2010

Annual - Ensemble Mean

### Precipitation Bias

### CORDEX-MNA44

### CORDEX-MNA22

### CORDEX-EUR11

#### non-adjusted

#### adjusted

1971-2005 Observation: E-OBS v19.0e 37 CORDEX-EUR11 Models

MPI-ESM1-2-HR

CCLM 4.8.17

- 1971-2000

# Projected Precipitation Change

- Global climate model projections of - CMIP 5: up to 42 simulations - CMIP 6: up to 41 simulations - Regional climate model projections of - CORDEX-MNA44: up to 4 simulations - CORDEX-MNA22: up to 2 simulations - CORDEX-MNA44: up to 37 simulations - Bias adjustment using ISIMIP3BASD - Projections until 2100 under - RCP 8.5 - Fossil-fueled development - Focus on Amman and surrounding area - Focus on ETCCDI indices

### Climate Change Signal - NDJFM 2071 - 2100 vs. 1981 - 2010

CMIP5 RCP 8.5

CMIP6 SSP5-8.5

CORDEX-MNA44 RCP 8.5

CORDEX-MNA22 RCP 8.5

CORDEX-EUR11 RCP 8.5

Temperature

Precipitation

#### Precipitation Events --- number of rainy days (RR1) ≥1mm --- number of vey wet days (R20mm) ≥20mm --- NDJFM --- Projection period 2071-2100 Reference period 1981-2010 --- 90% significance level

CORDEX-MNA44

CORDEX-MNA22

CORDEX-EUR11

(≥1mm)
# rainy days

days (≥20mm)
# very wet days

#### Precipitation Events --- Seasonal maximum precipitation (RX1day) --- Reference period 1981-2010

NDJFM

JJAS

#### Generalized extreme value distribution --- Annual maximum precipitation Empirical distribution fitted to GEV probability density --- Projection period 2071-2100 Reference period 1981-2010 ---

Standardized pdf

\begin{equation} f(s;\xi) = \begin{cases} \exp(-s) \exp\Bigl(-\exp(-s)\Bigr) \& ~~ \text{ for } ~~ \xi = 0 \newline {} \newline \Bigl(1+\xi s\Bigr)^{-(1+1/\xi)} \exp\Bigl(-(1+\xi s)^{-1/\xi}\Bigr) \& ~~ \text{ for } ~~ \xi \neq 0 ~~ \text{ and } ~~ \xi \, s > -1 \newline {} \newline 0 & ~~ \text{ otherwise. } \end{cases} \end{equation}

## Climate Change Summary - Amman

Variable

Unit

JJAS

NDJFM

1981-2010

2011-2040

2041-2070

2071-2100

1981-2010

2011-2040

2041-2070

2071-2100

Temperature

${}^\circ\,\mathrm{C}$

$\mathbf{26.6}$

$\color{red}{\mathbf{1.3}}$

$\color{red}{\mathbf{2.9}}$

$\color{red}{\mathbf{4.8}}$

$\mathbf{12.5}$

$\color{red}{\mathbf{1.0}}$

$\color{red}{\mathbf{2.3}}$

$\color{red}{\mathbf{3.9}}$

Precipitation

$\mathrm{mm}/\mathrm{d}$

$\mathbf{0.01}$

$\color{blue}{ 0.025}$

$\color{blue}{ 0.039}$

$\color{blue}{ 0.032}$

$\mathbf{2.02}$

$\color{red}{-0.089}$

$\color{red}{-0.265}$

$\color{red}{\mathbf{-0.508}}$

Rainy Days

$\#$

$\mathbf{0.1}$

$\color{blue}{ 0.06}$

$\color{blue}{ 0.04}$

$\color{blue}{ 0.02}$

$\mathbf{33.7}$

$\color{red}{-0.43}$

$\color{red}{\mathbf{-0.98}}$

$\color{red}{\mathbf{-1.61}}$

R10mm

$\#$

$\mathbf{0.03}$

$\color{blue}{0.08}$

$\color{blue}{0.07}$

$\mathbf{10.3}$

$\color{red}{-0.6}$

$\color{red}{-1.6}$

$\color{red}{\mathbf{-2.9}}$

R20mm

$\#$

$\mathbf{0.03}$

$\color{blue}{ 0.013}$

$\color{blue}{ 0.017}$

$\color{blue}{ 0.015}$

$\mathbf{2.92}$

$\color{blue}{ 0.001}$

$\color{red}{-0.062}$

$\color{red}{-0.167}$

RX1day

$\mathrm{mm}/\mathrm{d}$

$\mathbf{1.4}$

$\color{blue}{1.7}$

$\color{blue}{2.3}$

$\color{blue}{1.8}$

$\mathbf{30.1}$

$ \color{blue}{\hphantom{-}1.2}$

$ \color{blue}{\hphantom{-}0.5}$

$\color{red}{-1.5}$

RX5day

$\mathrm{mm}/5\mathrm{d}$

$\mathbf{1.4}$

$\color{blue}{2.5}$

$\color{blue}{3.6}$

$\color{blue}{2.8}$

$\mathbf{63.8}$

$\color{blue}{\hphantom{-}0.4}$

$\color{red}{-3.2}$

$\color{red}{-8.8}$

# The End

# Backup Slides

#### Temperature Climatology

1971-2000 Observation: E-OBS v24.0e

#### Precipitation Climatology

1971-2000 Observation: E-OBS v24.0e

## Observation Data

## Observation vs. Reference dataset

**Tarim Basin**

## Multivariate bias adjustment - Physical/Statistical dependencies between variables ignored so far (only univariate distributions adjusted) - How can we adjust multivariate distributions: - Conditional quantile mapping for single bins (Piani and Haerter, 2012) - Random rotations of variable-vector combined with univariate quantile mapping (Cannon, 2017)

**Simulation**

**Observation**

## Variance Inflation - Bias adjustment $\neq$ downscaling (Maraun, 2013) - Bias adjustment can lead to **variance inflation** in case of large scale gaps - Former local extremes are transfered to every station within a grid cell

Maraun (2013)

## Further issues - Statistical adjustment **NO** physical reasoning - We assume that $g(x_\mathrm{sim})$ does **not change in time** - Seperate bias adjustement for different season, month or day of year might be necessary - Bias adjustment can change temporal structure of timeseries on different scales - Naive QM can distort climate trend - Bias adjustment using conditional resampling of huge ensemble (Sippel et al., 2017)

## How to choose a suitable bias adjustment? - Which biases and how large (data exploration)? - What is important in your impact assessment (goal exploration)?