UCL DEPARTMENT OF GEOGRAPHY Climate change impacts on the Mekong River Daniel Kingston, Richard Taylor, Julian Thompson, Martin Todd Department of Geography, University College London Geoff Kite Hydro-Logic Solutions
The Mekong basin 795, km 2 42km long From Tibetan plateau (>5m) to Vietnam and South China Sea China, Burma, Thailand, Laos, Cambodia & Vietnam ~ 5 million people Socio-economic importance: Fish: 7, tons, 3 species p.a. (1992) Fish are 5-8% total protein intake Agriculture Hydropower
Climatology Mid-May to early October: southwesterly circulation, rainy. 9% annual precip between May-Oct October-March: northeasterly circulation, dry Snow storage and release in Tibet vs monsoon rains in lower basin Mean annual rainfall ranges from 1mm in northeast Thailand to >32mm in mountainous regions of Laos. Mean annual precip (mm) (IWMI Atlas)
Hydrology Mean total annual discharge = 475bn m 3 6 th largest in world Mekong at Chiang Saen Mekong at Mukdahan Streamflow, m3/s 3 25 2 15 1 5 Mekong Basin Streamflow Chi at Yasothon 1 2 3 4 5 6 7 8 9 1 11 12 Month Mekong at Pakse, 1981-199 Mekong at Chiang Sen, 196-1987 Mekong at Mukdahan, 1924-1987 Chi at Yasothon, 1953-1987 Mun at Ubon Rachathani, 1955-1987 Mun at Ubon Mekong at Pakxe
Hydrological model SLURP (Semi-distributed Land-Use Runoff Process) model (Kite, 1995) Semi-distributed, physically based Previously applied to the Mekong Kite, G. (21) Journal of Hydrology, 253 pp1-13. Model period 1994-1998 Percolation Transpiration Irrigation Interception Sublimation Withdrawals Precipitation Snowmelt Infiltration Groundwaterflow Evapotranspiration Canopy Storage Snow Storage Runoff Fast Storage Interflow Slow Storage
Initial model set-up 13 sub-basins derived from DEM USGS GTOPO-3 Sub-basins further divided, based on land-use (9 categories) USGS data FAO world soil map
Re-calibration of SLURP for QUEST-GSI Change from sparse network of daily station climate data to.5 degree gridded monthly data set Change of calibration period from 1994-1998 to QUEST 1961-9 period With 1991-1998 used for model validation Namngum Chi Lancang Namou Mekong1 Modelled to Pakxe only 57,km3, ~7% of basin Mun Chi-mun Mekong2
Re-calibration (2) Substitution of CRU TS 3 precipitation data with University of Delaware data set Change of PET algorithm From Penman-Monteith to Linacre method Data reliability issues mean daily discharge (m -3 s -1 ) 35 3 25 2 15 1 5 1 2 3 4 5 6 7 8 9 1 11 12 obs obs-15% obs+15% model Manual parameter adjustment Final calibration Nash-Sutcliffe =.94 Spearman coefficient =.95 1991-1998 validation Consistent with calibration period mean daily discharge (m -3 s -1 ) 5 4 3 2 1 2 4 6 8 1 % exceedence obs model
Scenarios 1-6 C prescribed warming on HadCM3 2 C prescribed warming on all 7 GCMs (HadCM3, HadGEM1, CCCMA, CSIRO, IPSL, MPI, NCAR) All 4 SRES scenarios on HadCM3 (24-69) A1b, A2, B1, B2 SRES A1b on all 7 GCMs (24-69)
Prescribed warming on HadCM3 Near-linear trend in annual runoff with increased mean global temperature Decreased peak season runoff Increased early season runoff anomaly (%) 5-5 Annual runoff anomaly from 1deg 2deg 3deg 4deg 5deg 6deg mean daily discharge (cumecs) 3 25 2 15 1 5 1deg 2deg 3deg 4deg 5deg 6deg
Prescribed warming on HadCM3: temperature vs precipitation Temperature climate change signal Temperature: decreasing peak season flow Precipitation: increasing flow from May-December mean daily discharge (m - 3 s -1 ) 3 25 2 15 1 5 1deg 2deg 3deg 4deg 5deg 6deg Precipitation climate change signal mean daily discharge (m - 3 s -1 ) 35 3 25 2 15 1 5 1deg 2deg 3deg 4deg 5deg 6deg
2deg prescribed warming on all GCMs Annual runoff anomaly from No consistent signal 1 Either on an annual or seasonal basis anomaly (%) 5-5 hadcm3 cccma csiro ipsl mpi ncar hadgem No outlier GCM -1-15 mean daily discharge (m -3 s -1 ) 35 3 25 2 15 1 5 hadcm3 cccma csiro ipsl mpi ncar hadgem1
2 degree prescribed warming: temperature vs precipitation Temperature climate change signal Temperature climate change signal very similar between GCMs Little consistency in precipitation climate change signal between GCMs mean daily discharge (m-3s-1) 3 25 2 15 1 5 Precipitation climate change signal hadcm3 cccma csiro ipsl mpi ncar hadgem mean daily discharge (m-3s-1) 35 3 25 2 15 1 5 hadcm3 cccma csiro ipsl mpi ncar hadgem
SRES scenarios on HadCM3 (24-69) Little difference between A1b, A2, B1 and B2 All show very small (<1%) changes in mean annual runoff Decreased peak season flow; slight increase in early season flow mean daily discharge (m -3 s -1 ) 3 25 2 15 1 5 anomaly (%) 5 3 1-1 -3-5 Annual runoff anomaly from a1b a2 b1 b2 a1b a2 b1 b2
SRES A1b on all 7 GCMs (24-69) Follows pattern at 2 C prescribed warming: No consistent signal Either on an annual or seasonal basis No outlier GCM mean daily discharge (m -3 s -1 ) 3 25 2 15 1 5 % anomaly 1 5-5 -1-15 annual runoff anomaly from hadcm3 cccma csiro ipsl mpi ncar hadgem 1 2 3 4 5 6 7 8 9 1 11 12 hadcm3 cccma csiro ipsl mpi ncar hadgem
Summary: Uncertainty envelopes 1-6 C prescribed warming on HadCM3 2 C prescribed warming across all 7 GCMs mean daily discharge (m -3 s -1 ) 3 25 2 15 1 5 mean daily discharge (m -3 s -1 ) 35 3 25 2 15 1 5 HadCM3 SRES scenarios (24-269) SRES A1b across all 7 GCMs (24-69) mean daily discharge (m -3 s -1 ) 3 25 2 15 1 5 mean daily discharge (m -3 s -1 ) 3 25 2 15 1 5 Solid line=; dotted lines indicate upper and lower bounds of climate change signal
Summary GCM uncertainty > climate sensitivity and emissions uncertainty Common themes Emissions uncertainty relatively small for 24-69 Uncertainty from GCMs primarily from precipitation, not temperature Little change in low flow season
Further work SRES A1b (24-269) GLOBAL MODEL Model uncertainty Parameterisation Model structure (comparison with global model) runoff (mm) 2 15 1 5 J F M A M J J A S O N D cccma ipsl mpi ncar hadcm3 Land use change CATCHMENT MODEL Abstractions? runoff (mm) 2 15 1 5 cccma ipsl mpi ncar hadcm3 J F M A M J J A S O N D