Simulating hydrologic impacts of climate change in the Dongjiang - - PowerPoint PPT Presentation

Simulating hydrologic impacts of climate change in the Dongjiang Basin, South China: Comparison of six monthly models

Yongqin David Chen

Department of Geography and Resource Management Centre of Strategic Environmental Assessment for China The Chinese University of Hong Kong, China Email: ydavidchen@cuhk.edu.hk

ICCC Pre-conference Seminars, 6 Oct 2009

Background and Motivation

• Global climate has been changing as increasingly evidenced by

temperature increase driven by anthropogenic warming. Availability and variability of water resources will be affected by global warming.

• Research on the effects of climate change on water resources includes

the use of – climate models, – downscaling techniques, and – hydrological models.

• Uncertainties exist in every step of the investigation.
• Uncertainties in GCMs and downscaling techniques have been widely

discussed in the literature.

• However, the uncertainties resulting from the use of different hydrologic

models have not been widely studied and reported in the literature.

Hydrologic Model Types and Use for Different Purposes

• Monthly water balance models – assessment of water availability under

different climatic conditions to support water resources management

• Simple statistical models – estimation of changes in the average annual

runoff for different climate change scenarios

• Conceptual rainfall-runoff models – detailed simulation of surface,

subsurface, and groundwater flow components to support a wide range of hydrologic analyses hydrologic analyses

• Physically-based distributed-parameter models – simulation of spatial

patterns of hydrologic processes in response to rainfall input for understanding the amount, pathway, and timing of flow

Study basin and data

0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Depth of water (mm) 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 Air temperature （

• C

） Runoff Rainfall Pan Evaporation Air Temperature

• Objective:

– to quantify how large the difference one would expect when using different hydrologic models to simulate the impacts of climate change as compared to their capabilities in simulating historical water balance components

Objective and Procedure

• Procedure:

– Evaluate the performance of the tested models in reproducing historical water balance components – Compare the difference of various models in simulating the hydrologic consequence of changed climate – Discuss the model structural effects on the difference in the simulation

Selection of Six Monthly Hydrologic Models

• Thornthwaite water balance model (TM, Alley, W.M., 1984)
• Vrije Universitet Brussel model (VUB, Vandewiele, G.L., Xu, C.Y. and Ni-

Lar-Win, 1992)

• The Xinanjiang model (XAJ, Chen, X., Chen Y.D. and Xu, C-Y, 2007)
• The model by Guo Shenglian (GM, Guo, S.L., 1992)
• The WatBal model (WM, Kaczmarek, Z., 1993)
• The model by Schaake (SM, Schaake, J.C., 1990)

• Alley, W.M., 1984. On the treatment of evapotranspiration, soil moisture accounting and

aquifer recharge in monthly water balance models. Water Resources Research, 20(8):1137-1149.

• Vandewiele, G.L., Xu, C.Y. and Ni-Lar-Win, 1992. Methodology and comparative study of

monthly water balance models in Belgium, China and Burma. Journal of Hydrology, 134; 315-347.

• Chen, X., Chen Y.D. and Xu, C-Y, 2007. A Distributed Monthly Hydrological Model for

Integrating Spatial Variations of Basin Topography and Rainfall. Hydrological Processes.

References for the Six Models

Integrating Spatial Variations of Basin Topography and Rainfall. Hydrological Processes. 21:242-252.

• Guo, S.L., 1992. Impact of climatic change on hydrological regimes in the Dongjiang
• Basin. Proceedings of First National Post-doctoral Conference. National Defence

Industry Press, Beijing, pp.2034-2037.

• Kaczmarek, Z., 1993. Water balance model for climate impact analysis. ACTA

Geophysica Polonica, 41(4):1-16.

• Schaake, J.C., 1990. From climate to flow. In P. E. Waggoner (Editor.), Climate change

and U.S. Water Resources. John Wiley & Sons, New York, pp. 177-206.

Thornthwaite water balance model (TM)

PET P Soil Moisture Storage (S) Ea ∆Q ∆Q Water Surplus (Q) Simulated Streamflow R

• No. of soil

zones

• No. of storages

(deficit) Types of storage (deficit) Runoff components 1 2 Soil moisture storage, Water surplus Runoff

Vrije Universitet Brussel model (VUB)

PET P Available Water Storage (W) Ea Rf Rs (W) Simulated Streamflow

• No. of soil

zones

• No. of storages

(deficit) Types of storage (deficit) Runoff components 1 1 Soil moisture storage Fast runoff, Slow runoff

The Xinanjiang model (XAJ)

PET P Upper Layer Soil Moisture Storage (WU) Low Layer Soil Moisture Storage (WL) Deep Layer Soil Moisture Storage (WD) EU EL ED Free Water Storage (S) Ea R Rf

• No. of soil
• No. of

Types of storage Runoff

Simulated Streamflow Groundwater Storage (GS) ΔGS Rs

• No. of soil

zones

• No. of

storages (deficit) Types of storage (deficit) Runoff components 3 5 Upper layer tension storage, Lower layer tension storage, Deep layer tension storage, Free water storage, Groundwater storage Fast runoff, Slow runoff

The WatBal model (WM)

P Relative Soil Moisture Storage (Z) Ea T PET Peff Rd Rs Rss Rb Simulated Streamflow Rb

• No. of soil

zones

• No. of storages

(deficit) Types of storage (deficit) Runoff components 1 1 Relative soil moisture storage Direct runoff, Surface flow, Sub-surface flow, Baseflow

The model by Guo Shenglian (GM)

PET P Soil Moisture Storage (S) Ea WS Groundwater storage (GS) Rg Rs Ri (GS)

Simulated Streamflow

• No. of soil

zones

• No. of storages

(deficit) Types of storage (deficit) Runoff components 1 2 Soil moisture storage, Groundwater storage Surface runoff, Interflow, Groundwater flow

The model by Schaake (SM)

P Relative Soil Moisture Ea T PET Peff Rd Rs Relative Soil Moisture Storage (Z) Simulated Streamflow Rss Rb

• No. of

soil zones

• No. of storages

(deficit) Types of storage (deficit) Runoff components 1 1 Relative soil moisture storage Direct runoff, Surface flow, Sub-surface flow, Baseflow

Monthly Water Balance Models

• The common and basic water balance equation at the monthly time scale:
• These models differ in how E and Q are conceptually considered and

mathematically represented.

• Actual E is estimated from potential E using soil moisture extraction function
• r coefficient of evapotranspiration.
• Large differences exist in the treatment of soil moisture accounting. Except

VUB, all models adopt a threshold value of soil moisture storage capacity.

• River flow routing is not considered and all the flow components run off

directly at the basin outlet on a monthly basis.

Model calibration and validation results for Dongjiang Basin (1960-1988)

• All six models produced good results for the calibration and

validation periods.

• The XAJ model performed the best as indicated by the highest E

value and the lowest RMSE value, followed by GM, VUB, TM, SM, and WM.

Comparison of model ability in reproducing the mean monthly runoff (1960-1988)

All six models simulate quite well the mean monthly runoff except the WM model for June and July, and the SM model for October and November.

Comparison of model ability in reproducing the mean monthly actual ET (1960-1988)

There is a good agreement in the mean monthly evapotranspiration simulated by the six models, except that the WM and SM models yield smaller values for winter and spring months.

Comparison of model ability in reproducing the mean monthly soil moisture (1960-1988)

Large differences exist because the models conceptualize and estimate soil moisture dynamics quite differently. In particular, the soil depth is usually not explicitly defined in all monthly

• models. Similar results have been reported in the literature.

Comparison of model results in predicting changes of mean annual runoff in response to changed climate (DT=1,2,4˚C; DP=0,± ± ± ±10,± ± ± ±20% )

• The differences between models

increase as P change increases in both directions.

• Decrease of P results in larger

differences than increase of P by the same amount. same amount.

• TM, XAJ and GM models behave

similarly and produce larger changes in runoff for a given climate change scenario, while the SM model produces the smallest changes.

• Runoff changes are more sensitive to

precipitation than to temperature.

Comparison of model results in predicting changes of mean annual evapotranspiration in response to changed climate (DT=1,2,4˚C; DP=0,± ± ± ±10,± ± ± ±20% )

• Even at the annual level there are large

differences in ET responses simulated by the six models driven by the perturbed climate scenarios.

• Models can be divided into two groups:

TM, XAJ and GM respond almost TM, XAJ and GM respond almost identically, and the other models show different responses.

• The effect of P changes on ET is smaller

than on runoff.

• Contrary to runoff, ET increases as

temperature increases.

Comparison of model results in predicting changes of mean annual soil moisture in response to changed climate (DT=1,2,4˚C; DP=0,± ± ± ±10,± ± ± ±20% )

• The differences in soil moisture changes

simulated by VUB are larger than the results of other models, probably because VUB is the only model which does not have an upper threshold limit for soil moisture. moisture.

• The differences in soil moisture changes

under alternative climates are smaller than the runoff changes predicted by the five models excluding VUB.

Comparison of simulated mean monthly changes in runoff for the three climate change scenarios (a) DT=2˚C and DP=-20%; (b) DT=2˚C and DP=0%; (c) DT=2˚C and DP=20%

• Different models produce very different mean

monthly runoff under the same climatic forcing.

• The larger differences in percent changes in

runoff for winter months may be caused by smaller absolute runoff values in dry season.

• The TM, XAJ and GM models respond

similarly while the other three models show a similar pattern of seasonal variation of predicted runoff.

• On average, when temperature increases by

2oC the mean monthly runoff changes by -30 -

• 50%, -5 - -10% and 10-30%, respectively for P

changes of -20%, 0% and 20%, depending on the model.

Comparison of simulated mean monthly changes in actual ET for the three climate change scenarios (a) DT=2˚C and DP=-20%; (b) DT=2˚C and DP=0%; (c) DT=2˚C and DP=20%

• Different models produce very different mean

monthly actual ET under the same climatic forcing.

• The TM, XAJ and GM models respond very

similarly while the differences among the other three models are considerable. three models are considerable.

• In the summer rainy months the model-

predicted actual ET does not change significantly with the change in precipitation. A two degree temperature increase causes about 10% increase in ET in all three cases, indicating the controlling effect of energy, instead of water availability, on ET in the rainy season in the humid area.

Comparison of simulated mean monthly changes in soil moisture for the three climate change scenarios (a) DT=2˚C and DP=-20%; (b) DT=2˚C and DP=0%; (c) DT=2˚C and DP=20%

• Different models produce very different mean

monthly soil moisture under the same climatic forcing.

• The TM, XAJ and GM models respond very

similarly while differences among the other three models are remarkable, especially VUB three models are remarkable, especially VUB behaves very differently from the other models.

• All the six tested models can reproduce almost equally well the historical runoff

data series, while large differences exist in the model simulated soil moisture.

• Using alternative climates as input to the tested models, large differences exist in

model predicted runoff, actual ET and soil moisture. The differences depend on the climate scenarios, season, and water balance variables under examination.

• Storage capacity is a critical parameter controlling the simulated soil moisture
• dynamics. A model without a threshold in soil moisture simulation can generate

much greater changes in model predicted soil moisture with respect to changed

Summary and Conclusions

much greater changes in model predicted soil moisture with respect to changed climate.

• Hydrologic models are developed for simulating hydrologic variables in stationary

conditions, correct reproduction of historical hydrologic variables provides no guarantee to correctly simulate hydrologic response of changed climate.

• Future water resources scenarios predicted by any particular hydrologic model

represent only the results of the model selected and used. More studies using different hydrologic models on different catchments need to be carried out in order to provide more general conclusions.