Praga

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Modelling water availability from a snow influenced basin in

central Chile under past scenarios of climate change

A. OSSES & X. VARGAS


División Recursos Hídricos y Medio Ambiente, Departamento de Ingeniería Civil, Facultad de Ciencias Físicas y
Matemáticas, Universidad de Chile. Casilla 228/3, Santiago, Chile
xvargas@ing.uchile.cl

Abstract The aim of this study is to compare water availability obtained using a
hydrological simulation model considering averaged and individual meteorological input
series of SRES A2 and B2 downscaled scenarios from years 1980 to 2005, generated from
HadCM3 general circulation model for a mountainous catchment located in central Chile.
If averaged series of mean daily temperature and daily evaporation are used as inputs of
the hydrologic model, results are not statistically different from those obtained with
individual series of these variables. By other hand, if the average precipitation series is used
as input of the hydrological model, the resulting flows are significantly different from
those obtained with individual series. In conclusion, the flow generated with the averaged
series of the generated weather variables show an overestimation of the discharge and a
poor statistical relationship with the observed values (1980-2005), being a poor approach
to assess the water availability. Better results can be obtained using individual series of
downscaled precipitation. Neither A2 nor B2 scenarios of climate change seem to have
occurred during the censored period.
Key words climate change; water availability; modelling;

INTRODUCTION

Assessing water availability at a local scale in a reliable way is nowadays challenge,


especially when different future scenarios of climate change are possible. Possible
scenarios of climate change can be obtained from simulations made in centres
especially dedicated to the study of these changes, being one of these the Hadley
Centre for Climate Prediction and Research from UK. In this study we use climate
predictions for two scenarios generated by the Hadley Centre through the HadCM3
atmosphere-ocean global circulation model (GCM) to obtain forcing meteorological
variables for hydrological models. To transfer this data to the local scale is possible to
use different techniques. This study explores a statistical downscaling technique and
intends to verify the occurrence of these scenarios during the censored period. Also
results of modelled discharge using average or individual downscaled series of
precipitation are analyzed for the snow influenced basin, El Yeso.
“El Yeso” reservoir is located in the Andes Mountains (2450 amsl) at “Piuquenes”
valley 100 km from Santiago, Chile’s capital; The reservoir was built all through 1953
and 1967 as one of the sources of drinking water supply for the city. The climate in
this region is characterized by relative warm summers (14.4ºC, mean) and cold winters
(2.9ºC mean), with temperature under 0°C low as -9ºC. The precipitation is
concentrated in winter and falls mostly as snow.
The water availability assessment is achieved through downscaling of the weather-
impacted variables that are used subsequently as inputs of the hydrological model to
simulate the basin discharge. The hydrological model selected for the study was
Sacramento model coupled with a snow model based in degree-day approach. This
model was calibrated and validated using observed available daily data from years 1980
up to 2005. We assessed the climate change scenarios performance in the registration
period by comparing observed data (1980-2005) and simulated flows using climate
change scenarios with averaged and single downscaled series

MODEL CALIBRATION AND VALIDATION

The implemented hydrological Sacramento-Snow17 model was calibrated using the


available daily data of temperature, evaporation and precipitation, from year 1980 to
1986 and validated from year 1986 to 1989 and 2000 to 2005. It is important to note
that, because the basin area has only one weather station, the model considers the
uncertainty of the observed lapse rate to represent the spatial and temporal variability
of the weather input variables (Muñoz & Vargas, 2006).
As it is seen in figure 1, simulated monthly values are close to observed values in
the calibration period (Nash-Sutcliffe coefficient, NS, is 0.83) but during validation
period there is a low adjustment especially at the end of the period (NS ranges from
0.86 to 0.25). This was attributed to a change in the measurement procedure of
meteorological variables and a new calibration of the parameters associated to those
input variables was performed; the final result is also presented in figure 1 and shows a
better representation of data with an increase of efficiency criteria (NS=0,59).

Fig. 1 Observed (Obs) mean monthly inflow to El Yeso reservoir and modelled mean monthly flow
calibrated in period 1980-1986 (Sim) and recalibrated in period 2000-2005 (Simc)

CALCULATION OF DISCHARGE FOR RECENT CLIMATE CHANGE


SCENARIOS A2 AND B2

Through statistical downscaling, using the software SDSM 3.1 (Wilby et al, 2004), we
evaluated the effect of regional changes in three climate variables (temperature,
precipitation and evaporation). The downscaling was done using the results from
HadCm3 global circulation model and the historical data series of the climate variables
from 1980 to 2000. The analysis of downscaled weather variables for the censored
period (Osses & Vargas, 2007) showed good agreement for the monthly mean
temperature and evaporation values observed at the weather gage station; however,
monthly precipitation, especially high values, are not detected. For scenarios A2 and
B2, Osses & Vargas (2007) propose the use of any set of the single series of daily
precipitation, evaporation or mean temperature generated by the program SDSM 3.1
but they also show that the average series of mean daily temperature or daily
evaporation are statistically indistinguishable with the single series; therefore these
average series can also be used in modelling. Finally, they indicate that if there is a
climate change scenario, B2 scenario is more similar to observed meteorological
variables.
The comparison between basin discharge observed data (1980-2005) and
simulated discharge under climate change scenarios using average and single series of
precipitation were used to assess the climate change scenarios performance in the
censored period. Figures 2 and 3 illustrate the results for SRES A2 and B2,
respectively.

Fig. 2 Observed (Obs) mean monthly inflow to El Yeso reservoir and modelled under B2 scenario using
average (SimPm) precipitation series and average (SimQm) of modelled mean monthly flow using
single precipitation series as inputs

These figures indicate that for scenarios A2 and B2 the discharge is greater than
observed values if the averaged precipitation series are used as input but lower if the
single precipitation series are used. For the first cases the probability of storm events in
the basin is near 55 percent, much greater than the 12 percent observed value. The
probability of snowfall is near 10 percent, value that is close to the observed value of
6.3 percent. By other hand, when single series of precipitation are used, the probability
of storm events is close to 7.5 percent and the probability of snowfall is near to 1.5
percent, being both values a little higher for A2 scenario.

Fig. 3 Probability of Observed (Obs) mean monthly inflow to El Yeso reservoir and modelled under B2
and A2 scenarios using mean (QPm) precipitation series and average (Qm) of modelled mean monthly
flows using single precipitation series as inputs.

CONCLUSIONS

As expected, results obtained indicated that when the average precipitation series is
used as input of the hydrological model, the resulting flows are significantly different
from those obtained with the single series.
Results obtained for both, SRES A2 and B2, scenarios are comparable and
significantly different from the observed discharge. Consequently, neither of these
two scenarios has been occurring in this basin.
To state the former conclusion, regional atmospheric models should be used as
an alternative procedure for obtaining downscaled meteorological variables.

Acknowledgements The assistance and support of Rodrigo Herrera, civil engineering


student and the data support by Aguas Andinas are acknowledged.

REFERENCES

Osses, A. & X. Vargas. (2007) Análisis de Escenarios de Cambio Climático en el Período 1980-2005 en la Cuenca del
Embalse El Yeso (Analysis of Climate Change Scenarios during 1980-2005 for El Yeso Reservoir Basin). XVII
Chilean Congress of Hydraulics, Santiago, Chile.
Muñoz, R. & X. Vargas. (2006) Escalamiento Espacial con error aleatorio en variables de entrada al modelo de simulación
hidrológica Sacramento (Spatial distribution with random error of Sacramento hydrological model input variables)
XXII Latin-American Congress of Hydraulics, IAHR, Venezuela.
Wilby, R. L. & C. W., Dawson. (2004) Using SDSM Version 3.1: A Decision Support Tool for the Assessment of
Regional Climate Change Impacts, User Manual. Available at https://co-public.lboro.ac.uk/cocwd/SDSM/.

You might also like