Amoussou e 20150708_1730_upmc_jussieu_-_room_207

24
Our Common Future Under Climate Change International Scientific Conference 7-10 July 2015 Paris, France. Analysis hydrometeorological events of floods in the watershed Mono shared Benin-Togo (West Africa) / Analyse hydrométéorologique des crues dans le bassin-versant du Mono en Afrique de l’Ouest avec un modèle conceptuel pluie-débit Amoussou Ernest, Tramblay Yves, Totin V. S. Henri, Mahé Gil, Paturel Jean-Emmanuel, Ribstein Pierre, Descroix Luc, Houndénou Constant & Boko Michel Paris, 7-10 july 2015

Transcript of Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Page 1: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Our Common Future Under Climate Change

International Scientific Conference

7-10 July 2015 Paris, France.

Analysis hydrometeorological events of floods in the

watershed Mono shared Benin-Togo (West Africa) /

Analyse hydrométéorologique des crues dans le

bassin-versant du Mono en Afrique de l’Ouest avec

un modèle conceptuel pluie-débit

Amoussou Ernest, Tramblay Yves, Totin V. S. Henri,

Mahé Gil, Paturel Jean-Emmanuel, Ribstein Pierre,

Descroix Luc, Houndénou Constant & Boko Michel

Paris, 7-10 july 2015

Page 2: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Context

The change / climate variability resulting in a plan

amendment and rate of precipitation (frequency and

intensity).

Rainfall instability observed in the hydrological basin

variations with a recurrence of floods in recent years.

1 & 2: Climate bimodal; 3 & 4: unimodal climate

Source: Amoussou & al., 2012

Page 3: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

The recurrence of extreme precipitation

anomalies resulting in flooding or drought is a

normal component of natural climate variability.

The adverse effects of floods in recent years

have strong socio-economic and ecological

impacts loss of life and property damage.

Page 4: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Vulnerability to natural hazards is high in

West Africa and the Mono Basin in

particular, where populations tend to occupy

more and more the most exposed areas.

The goal is to analyze the

dynamics of the basin

Hydrometeorological and

distribution of incoming flood

dam Nangbéto through the

hydrological model GR4J in

daily time to assess the risk of

flooding in the lower valley of

the river.

Page 5: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Between 06 ° 16 'and 09 ° 20'N and between 0 ° 42' and 2 ° 25'E

Area: 27,870 km²

Two rivers in basin: the Mono and Couffo

Where is the study area?

Page 6: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

The sub-basin of Nangbéto:

9952 km²

The launch in September

1987 of dam

Page 7: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

TOOLS

GR4J Hydrological

modeling of floods

Calculation of ETP (Oudin et al., 2005)

Daily climate data (rain, ETP); 14 rainfall stations were selected

Daily flow data (flow) at the inlet and outlet of the dam

Correlation between annual

floods and hydro-climatic

indicators

Period:1988- 2010

Data and methods

Ordinary kriging block by

precipitation (climatological

variogram)

Detecting trends

Mann-Kendall and non-

stationary GEV

Page 8: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Fonctionnement GR4J

Source : Perrin & al., 2003

Two production parameters (X1:

Production capacity of the tank and F

(X2): parameter exchange (gain or

loss);

Two transfer parameters (X3: the

routing capacity tank and X4: time

base (maximum) of the hydrograph.

Model parameters are calibrated

automatically with the Simplex

algorithm with multi-objective

function :

BiasRRMSENashf peak 11

Page 9: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

RESULTS

Page 10: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

1. Dynamic flood and Trends

Page 11: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Climate trends in the basin

A slight non-significant decrease in rainfall during the late

2010s and a slight increase in water evaporated blades

decline in flood flows?

Page 12: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Interannual variability of the maximum 24-hour rainfall from

1988 to 2010 in the Mono Basin Nangbéto

A net increase in the annual maximum daily

water slides precipitated significant at the

95% level.

Page 13: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

A marked seasonal rhythm with almost all flow rates recorded during

the summer between June and October small role but significant dam

(Payan, 2007 and Amoussou, 2010)

Seasonal dynamics of flow (m3 / s) flows at the dam Nangbéto

Page 14: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Dam Influence Nangbéto on floods

Empirical distributions of annual

maximum flows measured

upstream and downstream of the

dam Nangbéto

Annual maximum flows out of the

dam and Nangbéto rate corresponding

incoming

Good correlation between

incoming and outgoing flows of

the same day R² = 0.77 the dam

does not protect against floods

and waits almost all daily

maximum flood, especially for

high flow rates

Page 15: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Correlations between annual floods to incoming Nangbéto

and hydro-climatic indices

R2 = 0,7483

0

200

400

600

800

1000

1200

1400

1600

0 20 40 60 80 100 120 140 160 180 200

Q moyen annuel

Q m

ax an

nu

el

Correlation mean annual flow and annual

maximum flow

Correlation between maximum rates and

annual rainfall over several days max

Good correlation (r = 0.86) with the annual flood mean annual flow

The best correlation coefficient (r = 0.67) is with the annual rainfall

averaged over 19 days max

Greater influence of saturated soils that

heavy rainfall on floods

Page 16: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Frequency analysis of the return period flood

Paramètres Valeur Bornes intervalle de confiance 95%

Nangbéto

amont

k -0.10 -0.33 0.13

sigma 329.33 242.98 446.38

mu 861.62 716.57 1006.66

Nangbéto

aval

k 0.44 -0.16 1.04

sigma 258.21 159.46 418.11

mu 372.69 238.35 507.03

Adjustment to GEV inflows

and outbound Nangbéto

Settings GEV distributions

Quantiles de débit

Période de retour (années) Nangbéto amont Nangbéto aval

2 961.21 526.68

5 1330.89 914.53

10 1577.45 1210.55

20 1815.30 1527.97

50 2125.15 1993.82

100 2358.82 2388.69

Return periods of annual maximum flows

Type Frechet

Type Weibull

Page 17: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

2. Modeling with GR4J

Page 18: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Modeling of flood flows in Nangbéto: Calibration GR4J

Daily discharges observed and

simulated from 1988 to 2010

Nangbéto

Average flows observed and

simulated Nangbéto 1988-2010

Model the distribution of incoming flood dam Nangbéto through the

hydrological model GR4J the daily time over the period 1988-2010.

The model reproduces well the

seasonal dynamics of flows and daily

values

Page 19: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

The flow rates extracts raw data observed and simulated data by the calibrated

model the two distributions are similar and not statistically different

(Kolmogorov-smirnov)

Empirical distributions

of incoming annual flood

flows to Nangbéto,

observed and simulated

by GR4J

Page 20: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Validation Test 1: Split-sample "classic" into 2 periods

Calibration

(1988-1999)

Validation

(2000-2010)

NASH 0,80 0,71

RRMSE 0,28 0,57

BIAS 0,002 0,16

Statistique KS 0,27 0,40

p-value 0,74 0,31

x1 x2 x3 x4

Calibration 226,49 0,63 77,34 1,60

Validation 262,51 0,25 118,24 1,97

Parameter values of the calibration model and

validation samples

statistical model of highly degraded between the period and the timing of

validation

A bias on volumes of 2% to 16% and especially a RRMSE on the tips of

flood 28% to 57%.

Quantile-quantile distribution of annual

maximum flows to Nangbéto

Page 21: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Validation Test 2: Calibration of each for each year to analyze

the variability of the model parameters x1

0

100

200

300

400

500

600

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

x2

0

0,2

0,4

0,6

0,8

1

1,2

1,4

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

Parameter x1 (capacity

of the production)

Parameter x2 (coefficient of underground exchanges)

The parameters x1 and x2

are not stationary and have

an upward trend for x1 and

x2 for downward over time.

Page 22: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Validation Test 3: Split-sample of 3 periods

Calibration Validation1 Validation2

NASH 0,80 0,78 0,62

RRMSE 0,31 0,28 0,70

BIAS 0,00 0,04 0,28

Statistique KS 0,25 0,25 0,57

pvalue 0,93 0,93 0,13

The Pettitt test indicates a break in the ETP since 1995. Three periods of 8 years from 1988

to 1995 (calibration), 1996-2003 (validation1) and 2004-2010 (validation 2).

The further away from the calibration period 1988-1995, plus we get bad

results, including the 2004-2010 period.

Calculating the ETP formula

Oudin et al., 2005

Rupture

Page 23: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Conclusion 1.No trends in flows and rainfall from 1988 to 2010, but

upward trend temperatures.

2.The difference between the input and output rates of

borderline significance dam showing the low

retention capacity of the reservoir dam at high water.

3. Flooding induced annual rates maxima and well

correlated with mean annual flows and soil saturation

conditions.

4.Good performance in calibration GR4J but degraded

performance validation and temporal instability of the

parameters

Page 24: Amoussou e 20150708_1730_upmc_jussieu_-_room_207

Continuos reflection !!!!

Thank you!