IWRM Modelling Report

Mekong River Commission/ Information and Knowledge Management Programme DMS - Detailed Modelling Support Project Contract #001-2009, Work Package 02/...
Author: Belinda Rice
9 downloads 0 Views 3MB Size
Mekong River Commission/ Information and Knowledge Management Programme

DMS - Detailed Modelling Support Project Contract #001-2009, Work Package 02/1-2

IWRM Modelling Report

Finnish Environment Institute in association with EIA Centre of Finland Ltd.

DMS Project Document

IWRM Modelling Report

December 2010 Juha Sarkkula, Jorma Koponen, Hannu Lauri, Markku Virtanen Finnish Environment Institute Mechelininkatu 34a 00260 Helsinki Finland Tel: +358 9 403000 Fax: +358 9 40300390 www.environment.fi/syke [email protected] [email protected]

EIA Ltd. Tekniikantie 21 B 02150 Espoo Finland Tel: +358 9 70018680 Fax: +358 9 70018682 www.eia.fi [email protected]

2

Table of Contents 1

THE DMS PROJECT...................................................................................................... 8

2

NEED FOR INTEGRATED AND HOLISTIC MODELLING APPROACH .......................10

3

BACKGROUND AND SCOPE OF THE MRC DSF, WUP-FIN AND IWRM MODELS ...12

4

IWRM-TOOL ARCHITECTURE ....................................................................................14

5

MODEL TEST AREAS ..................................................................................................17 5.1 THE MAE CHAEM AREA ........................................................................................... 17 5.1.1 Description of the area ..................................................................................... 17 5.1.2 Geographical data and model grid .................................................................... 19 5.1.3 Meteorological data .......................................................................................... 20 5.1.4 Hydrological data.............................................................................................. 22 5.1.5 Model setup...................................................................................................... 24 5.1.6 Model calibration and results ............................................................................ 24 5.1.7 Hysteresis modelling ........................................................................................ 27 5.2 THE KHUWAE NOI AREA .......................................................................................... 28 5.2.1 Description of the area and model data ............................................................ 28 5.2.2 Model setup...................................................................................................... 29 5.2.3 Model calibration .............................................................................................. 30 5.2.4 Model results .................................................................................................... 31

6

IWRM MODEL IMPLEMENTATION ..............................................................................36 6.1 THE MEKONG BASIN CHARACTERISTICS ................................................................... 36 6.2 SOIL CLASSIFICATION ............................................................................................. 37 6.3 GEOGRAPHICAL DATA AND MODEL GRID ................................................................... 38 6.4 COMPARISON OF 5 KM AND 2 KM GRIDS ................................................................... 42 6.5 METEOROLOGICAL DATA ......................................................................................... 43 6.6 HYDROLOGICAL DATA ............................................................................................. 44 6.7 MODEL CALIBRATION .............................................................................................. 46 6.8 MODEL COMPUTATIONAL TIME ................................................................................. 49

7

HYDROLOGICAL BDP DEVELOPMENT SCENARIO MODELLING ...........................50 7.1 FLOW IMPACTS ....................................................................................................... 50 7.2 COMPARISON WITH THE DSF RESULTS .................................................................... 51

8

CHINA MODELLING .....................................................................................................53 8.1 METEOROLOGICAL AND HYDROLOGICAL DATA .......................................................... 53 8.2 MODEL CALIBRATION .............................................................................................. 54 8.3 RECENT DROUGHT SITUATION ................................................................................. 58 8.4 SNOW AND GLACIER MELTING .................................................................................. 60 8.5 COMPARISON WITH THE SWAT RESULTS ................................................................. 63

9

CROP, IRRIGATION AND WATER TRANSFER MODELLING ....................................67 9.1 BASIC MODEL DIVERSION STRUCTURE AND CONTROLS .............................................. 67 9.2 CROP MODELLING .................................................................................................. 70

3

10

GROUNDWATER .....................................................................................................74

REFERENCES .....................................................................................................................76

4

Illustration Index Figure 1. Comparison between the DSF (left) and IWRM+1D,2D,3D (right) modelling lines. 12 Figure 2. Components of the WUP-FIN model suite. ............................................................14 Figure 3 Graphical user interface for the IWRM model. ........................................................15 Figure 4. IWRM-model test areas. Left Mae Chaem in North-West Thailand (blue borders), right Khuwae Noi in Norhtern Thailand. ................................................................................17 Figure 5. Mae Chaem DEM .................................................................................................19 Figure 6. Location of the Mae Chaem rainfall stations (GAME-T Rainfall Measurement Data Archive of the Mae Chaem Basin) ........................................................................................21 Figure 7. Average Daily temperature at Stations P14 (observation) and W2_076 (NCEP data). ....................................................................................................................................22 Figure 8. Daily measured discharge at P14 Kaeng Ob Luang (1998-2003) .........................23 Figure 9. Average monthly measured discharge at Station P14 (1998-2003) ......................23 Figure 10. Computed and measured discharge at Station P14 (years 1998-2003) .............24 Figure 11. Computed and measured cumulative discharge at Station P14 (years 1998-2003) .............................................................................................................................................25 Figure 12. Soil water content [mm] in soil layer 1 (s1) for different locations (Ts1 and Ts2) .26 Figure 13. Soil water content [mm] in soil layer 2 (s2) for locations Ts1 and Ts2. ...............27 Figure 14. Hysteretic soil water content (θ) - pressure head (h) relation for a typical soil (Stewart, 2003). ....................................................................................................................27 Figure 15. Khuwae Noi basin. R1, R2, R3 and H1 are precipitation measurement stations and W1, W2 and W3 are automated weather stations. .........................................................29 Figure 16. Computed and measured soil moistures for the modelling period April 1996 – March 2000 in Khuwae Noi. Meas @ 40 cm is the soil moisture measurement from depth 40 cm at weather station W1. Comp S1 and Comp S2 are the computed average soil moistures of the upper and the lower soil layers. ..................................................................................31 Figure 17. Computed and measured daily average flows for the modelling period April 1996 – March 2000. ......................................................................................................................32 Figure 18. Computed and measured cumulative flows for the modelling period April 1996 – March 2000. .........................................................................................................................32 Figure 19. Computed monthly evapotranspiration and corrected monthly PET for the modelling period April 1996 – March 2000............................................................................33 Figure 20. Computed cumulative evapotranspiration and corrected cumulative PET for the modelling period April 1996 – March 2000............................................................................33 Figure 21. Comparison of computed soil moistures between two grid boxes from weather stations H1 and R2. S1 is the average soil moisture of the upper soil layer and S2 is the average soil moisture of the lower soil layer. ........................................................................34 Figure 22. Vertical outflow of water Q1 from the upper soil layer of two grid boxes at weather stations H1 and R2. ................................................................................................34 Figure 23. Vertical outflow of water Q2 from the lower soil layers of two grid boxes at weather stations H1 and R2. ................................................................................................35 5

Figure 24. Mekong River Basin. ...........................................................................................36 Figure 25. Processed soil and landuse grids for Mekong river basin model. .........................42 Figure 26. 5 km (left) and 2 km (right) land use model grids. ................................................43 Figure 27. NCEP weather data points. ................................................................................44 Figure 28. MRC flow measurement points. ...........................................................................45 Figure 29. Monthly average flows in Vientiane, Pakse and Stung Treng. .............................45 Figure 30. Observed flows (red line) compared with the modelled ones (black line) in Chiang Sae, Vientiane and Kratie. Observed flows are obtained from water level measurements and rating curves.........................................................................................................................47 Figure 31. Example of soil moisture. Blue is less porous water and red more. .....................48 Figure 32. Baseline (black line) and 20 year dam (red line) scenarios. Flow in Chiang Saen. .............................................................................................................................................50 Figure 33. Baseline (black line) and 20 year dam (red line) scenarios. Flow in Kratie. ..........51 Figure 34. WMO weather stations used in the model. .........................................................53 Figure 35. Comparison between observed flow (red line) with the simulated one. Observed flow obtained from Chiang Saen water level measurements and rating curve. .....................55 Figure 36. Comparison between observed flow (red line) with the simulated one. Period 1.5.1987 - 1.5.1990. .............................................................................................................55 Figure 37. Comparison between observed flow (red line) with the simulated one. Period 1.5.1993 - 1.5.1996. .............................................................................................................56 Figure 38. Comparison between observed flow (red line) with the simulated one. Period 1.5.1996 - 1.5.1999. .............................................................................................................56 Figure 39. Comparison between observed flow (red line) with the simulated one. Period 1.5.2003 - 1.5.2006. .............................................................................................................57 Figure 40. Comparison between observed flow (red line) with the simulated one. Period 1.5.2006 - 1.5.2009. .............................................................................................................57 Figure 41. Comparison between observed flow (red line) with the simulated one. Period 1.1.2009 - 5.3.2010. .............................................................................................................58 Figure 42. Comparison between observed flow (red and green lines) with the simulated natural flow without hydropower dams. Period 1.1.2010 - 5.3.2010. .....................................58 Figure 43. Observed (red and green) and simulated (black) yearly Chiang Saen minimum flows (weekly averages). The green line is obtained by adding 200 m3/s to the rating curve (compare to Figure 41). ........................................................................................................59 Figure 44. Snow melt expressed as discharge. Monthly (black line) and annual (red line) averages. .............................................................................................................................60 Figure 45. Glacier melt expressed as discharge. Monthly (black line) and annual (red line) averages. .............................................................................................................................61 Figure 46. Modelled Chiang Saen flow (black line) compared to weekly average snow melt (red line). ..............................................................................................................................61 Figure 47. Modelled glacier melt. ..........................................................................................62 Figure 48. Snow depth in two locations near Zadoi station in the uppermost part of the Mekong Basin. .....................................................................................................................62

6

Figure 49. Observed (red squares) and simulated (black squares) yearly Chiang Saen minimum flows. ....................................................................................................................63 Figure 50. Observed (red squares) and simulated (black squares) yearly Chiang Saen minimum flows. ....................................................................................................................64 Figure 51. Simulated daily average flows of Baseline and Dam Cascade scenarios at Chiang Saen. (Räsänen 2010) .........................................................................................................65 Figure 52. Monthly average water levels of Baseline and Dam Cascade scenarios at Chiang Saen. (Räsänen 2010) .........................................................................................................65 Figure 53. Monthly average flow changes [%] caused by Lancang-Jiang cascade at five locations in the Mekong mainstream. Changes in the four downstream locations are based on simple monthly water balance calculations. (Räsänen 2010) ...........................................66 Figure 54. Definition of an irrigation and water diversion areas in the model user interface. 68 Figure 55. Irrigation, diversion and water use definitions in the IWRM model user interface. .............................................................................................................................................69 Figure 56. Irrigation water demand for dry and wet season rice as well as downstream flow in case of no irrigation (black line) and irrigation included (red line). .....................................71 Figure 57. IWRM model crop type definitions. .....................................................................71 Figure 58. IWRM model crop monthly factors. .....................................................................72 Figure 59. IWRM model rice paddy water ponding types. ....................................................72 Figure 60. IWRM model rice paddy water daily ponding depths [cm]. ..................................73 Figure 61. Definition of the IWRM model pump capacities and crop mixes. .........................73 Figure 62. Irrigation demand in case of groundwater pumping.............................................74 Figure 63. Impact of groundwater irrigation on downstream irrigation area discharge. Black line without irrigation, red line with irrigation. ........................................................................75 Figure 64. Groundwater (soil saturated water) depth in case of no irrigation (black line) and with groundwater pumping (red line) in the groundwater diversion point. ..............................75

7

1 THE DMS PROJECT

The project „Detailed modelling Support to the MRC‟ is executed under the IKMP/Component 4 (Modelling). It responds to the acute basin development issues and the extensive model services needs of the MRC Programmes and the member countries. The work is based on holistic and cross-cutting approach and team work. The general objectives of the work are: 1. Provision of comprehensive tools and data for hydrological, environmental and socio-economic impact assessment of ongoing and planned basin developments 2. Support of riparian countries and MRC programmes capacity in modelling and impact assessment 3. Promotion of cross-programme, cross-disciplinary and cross-sectoral cooperation, coordination and integration through focused group work. The project objectives are reached through a sequential process starting from (i) tools adjustment and integration, continuing in (ii) clarifying of the central Mekong processes and development impact on them and ending on (iii) socio-economic consequence analysis. The process analysis is based on the use of existing and adjusted tools forming the IWRM-tool and offering a platform for the programmes to cooperate in an integrated and coordinated way. The process and socioeconomic analysis results have been integrated in the IWRM modelling framework. The project work is organised around “Work Groups” indicating both cross-cutting themes and cooperative groups participating in the theme work. The suggested Work Groups and involved MRC programmes are: 1. IWRM-tool (EP, BDP, IKMP) 2. Sediments (EP, BDP, SHI, IKMP) 3. Forestry and agriculture productivity (AIFP, BDP, IKMP) 4. Fisheries (FP, EP, BDP) 5. Socio-economics (EP, BDP, FP (SHI, NP, FMMP, IKMP) Basin wide IWRM-tool group consists of three components – (i) IWRM watershedscale tool, (ii) IWRM basin-wide tool and (iii) IBFM integration in the model system. Sediment relates to watershed erosion, agriculture and forestry practices, sediment transport, sedimentation and erosion in the river channels, river morphology changes, lake sediment balance, sedimentation in the floodplains, sediment trapping by small and large scale dams, sources and fate of different quality sediments, sediments and primary productivity and coastal erosion. Forestry and agriculture productivity activities link AIFP activities to other MRC activities such as IWRM-tool and sediment studies. Fisheries component starts with primary productivity and habitat modelling. Primary productivity has been connected to the fisheries productivity in close cooperation with the Fisheries Programme. Socio-economics activities include inclusion of (i) monetary cost and benefits and (ii) livelihoods and well-being dependency on natural resources in the

8

modelling framework and support for BDP2 development scenario impact assessment. Climate changes studies are an integral part of the entire group. This report is result of the Work Package 2 of the DMS-project. The Work Package is divided into five sub-packages: 1. IWRM-tool 2. Indicator framework integration and socio-economics 3. Sediments 4. Productivity and fisheries 5. Management, cooperation, liaison, group work and capacity building These sub-tasks cover the five main project themes except the “Forestry and agriculture productivity.” This report presents mainly methodology and results of the two first sub-tasks, that is IWRM-tool and Indicator framework integration. The Sediments, Productivity and fisheries sub-task and Capacity building are reported separately.

9

2 NEED FOR INTEGRATED AND HOLISTIC MODELLING APPROACH

Needs for capacity, technology and research in support of sustainable water resources and environmental management are growing rapidly, given the foreseen water and environmental resources developments in the Mekong Basin. As an example, value of the planned hydropower investments in Laos over the next 10 years is estimated to be over USD 50 billion and Mekong Delta flood protection measures to combat climate change impacts amount to tens of billions, if not hundreds of billions, of dollars. Risks for the natural Mekong system are also high. For instance it is estimated that the yearly Mekong fisheries value is between 2 and 10 billion USD. The possible losses of fisheries are accentuated by the fact that especially in Cambodia the poorest part of the population is directly dependent on the natural resources. The workload and demands on water resources related authorities is expected to increase substantially within the next years, given the number of water resources projects in the pipeline. The authorities need state-of-the-art knowledge as well as impact and planning tools to successfully accomplish the tasks they are responsible for. Mathematical modelling provides information about the consequences of the planned developments. It enables understanding of complex interdependent processes, heterogeneous interlinked areas and time evolution of the basin state. Modelling provides information for forecasting, decision making, planning and impact assessment. In addition, modelling is a powerful tool for data integration and communication. The long term aim of the IKMP is to develop the capability to model all physical and environmental aspects of the Lower Mekong Basin (and areas beyond the LMB that have an influence) and link these to socio-economic issues: Geomorphology - coastal erosion, geology, deep pools, riverbed, bank erosion Climate change – changes to snow melt, glacier contribution, sea level, water resources, watershed erosion Flow effects - stream water / flooding, precipitation, groundwater, surface (floodplain) flow, drought, water balance Sediments- bed load, suspended load, erosion / deposition balance, effects on navigation Carbon - organic, FPOC, DOC, DIC, carbonate systems,CH4, carbon (natural) balance within basin Nutrients - Nitrogen, Phosphorous, Potassium, plus any other found to be significant. Assess nutrient balance, primary production and links to Fisheries and Agriculture production Water quality - biota, balance Fisheries - migration, non-migration, spawning, catch, balance, effects of nutrients, flood pulse effects

10

Agriculture - main crops, production, balance, effects of nutrients, hydrological constraints Impacts of human interventions - hydropower dams, flood protection structures, channel construction, flow regulation, bank protection structures, sand mining, river channel filling, dredging, rock blasting, irrigation, groundwater use, dikes, fertilizer, pesticides, wastewater, power plants (Oil, coal, gas, nuclear, hp, solar, others), forest clearance / mono crops, reforestation, planting/cutting of mangrove, aquaculture, mining and energy balance (renewable verses non renewable) The IWRM-tool has been developed in order to accommodate, in combination with the 2D/3D model, these factors under one comprehensive system.

11

3 BACKGROUND AND SCOPE OF THE MRC DSF, WUP-FIN AND IWRM MODELS

MRC modelling capacity has been developed over 10 years. The development started with the MRC Water Utilisation Programme (WUP) in 2001 and has continued under the MRC Information and Knowledge Management Programme (IKMP) since 2007. In the original WUP plan modelling was considered as an integrated and holistic tool covering water amounts, water quality, environment and socio-economics. However, in the WUP implementation model development and implementation was divided into two main parts: DSF (Decision Support Framework) and WUP-FIN. DSF focus has been on water volumes and basin-wide modelling and Knowledge Base. WUP-FIN has complemented DSF with detailed local flow, water quality and environmental and socio-economic modelling. WUPFIN has focused on realistic (non-schematic) description of hydrology, flow, flooding, sediments, water quality and productivity. Figure below illustrates the focal differences between the two lines of model development.

Figure 1. Comparison between the DSF (left) and IWRM+1D,2D,3D (right) modelling lines.

As DSF has been increasingly applied to local case studies and as WUP-FIN tools have been lately applied to basin-wide and regional scale, the distinction between the two lines of modelling have become less clear. At the same time the development pressures and need for comprehensive modelling have increased. It has become evident that MRC and the countries need rather a Modelling Toolbox 12

than a fixed reference model (previous DSF). The way forward since beginning of 2009 has been integration of the different models under the DSF Knowledge Base and scenario management tool. Data transfer tool is being developed to connect different modes to the KB and to each other. In parallel with the Toolbox development and DSF integration, development of an integrated modelling tool has been going on. The idea behind the development is to take different elements of the DSF and Toolbox and build a practical IWRMmodel for Integrated Water Resources Management. The IWRM-model can be used both in basin-wide and local scales. Mainly because of the basin-wide BDP process and its needs, the IWRM-model has been applied first for the whole Mekong Basin from the Tibetan Plateau to the SouthChina Sea. However, because of the limited resources available, it has not been feasible to implement full Vietnamese Delta hydrodynamic model at this stage. Construction of a 1D hydrodynamic Delta model would be quite involved, would duplicate the ISIS Delta implementation and would anyway fall short of describing the Delta conditions. It makes more sense to implement a new combined 1D/2D/3D modelling system for the Delta, but this requires also major data collection effort especially for the infrastructure (dykes and gates). Consequently in the current basin-wide IWRM-model implementation Vietnamese Delta part of the model acts to route water and sediments, but doesn't describe realistically the channel network, tidal fluctuations, floodplain flow and sedimentation and saline intrusion.

13

4 IWRM-TOOL ARCHITECTURE

The IWRM in relation to the other WUP-FIN models and platform is shown in Figure 1. Model platform consists of GIS, visualisation, data processing and analysis and graphical user interface. The graphical user interfaces are tailored separately for each model as they models need for instance different parameters. However, the underlying software libraries and many features are the same for each single user interface.

Figure 2. Components of the WUP-FIN model suite.

The platform GIS-component provides for instance creation and editing of model grids. Another platform component, GUI, is also based on the GIS-component (Figure 3). The GUI provides an interface for model control as well as input/output data processing. Visualisation system consists of time series, animation and map drawing software. The analysis tools provide large number of time series and spatial data analysis tools mostly for processing model outputs. The platform is based on shared software (VIV system software). The platform components are tailored for each separate model through an open, easy to modify, C++ scripting language, but the same service library routines are used for any tailored system. For instance model specific data input dialog windows can be modified with any text editor or a graphical dialog window editor. In addition to the functions listed above, model platform provides Windows API, multi-language (Fortran, C, C++), parallelisation, model integration and interfacing services. These form software infrastructure which any model can utilise. For instance a model can be programmed to execute in parallel in order to speed it up by employing the Windows API thread management routines and model platform calls. 14

Figure 3 Graphical user interface for the IWRM model.

The simulation models are divided into two groups: IWRM-model and 1D, 2D and 3D flow and water quality models. The IWRM-model is based on the VMOD hydrological model. It is gridded, “raster based”, watershed model. Because it is gridded it corresponds to GIS based data representation. This naturally facilitates model and GIS integration. The same applies to the 3D model. The 1D model is intended mainly for river channel modelling in basin-wide scale and for smaller channels in local applications. Water resources control modules are small models for reservoirs, irrigation, flood control etc. At the moment the 1D model can be fully coupled with the IWRM-model and the 2D/3D model. The IWRM-model provides boundary values for the 2D/3D model but the 2D/3D model has not been yet coupled with the IWRM-model. The coupling can established in a straightforward way when the grids in the IWRM and 2D/3D are the same. The reservoir, lake and flood flow can be then calculated with the 2D/3D model and hydrology with the IWRM-model. The modelling infrastructure for combining the IWRM-tool with the 2D/3D hydrodynamic/water quality model exists, although the required inter-model updating of flooding/drying-up has not yet implemented. The combination infrastructure consists of data transfer mechanisms, process/thread scheduling, mass balance counters and mixed language programming (C, C++, Fortran). 15

The elements of the IWRM-model are: 1. gridded hydrology and kinematic routing (WUP-FIN VMOD) 2. 1D hydrodynamics (WUP-FIN RNet) 3. watershed erosion, sediment transport and sediment trapping 4. water quality (WUP-FIN VMOD) 5. flooding (applicability needs to be tested further) 6. groundwater 7. crops and irrigation 8. water diversions from rivers, lakes, reservoirs and groundwater 9. reservoirs (same as in the DSF IQQM) 10. DSF KB data (through import scripts and DSF data transfer tool that is being developed) Dynamic optimisation for individual reservoirs and reservoir cascades has been recently used together with the IWRM model, but it has not yet been fully integrated in the model system. Naturally it is not necessary to apply all of the components and options in all model applications. Most importantly, in many cases full 1D hydrodynamics is not necessarily needed and the more simple kinematic approximation is used. Currently the WUP-FIN RNet 1D model is integrated in the model because it is the most accessible option for the IWRM-model developers. However, the system has been designed in a way that any 1D model with open interfacing can be used instead. IWRM-model relates to the DSF development and utilises DSF work in a number of ways: IWRM integrates the DSF model functionalities (SWAT/hydrology, IQQM/water resources management, ISIS/hydrodynamics) under one model. Knowledge Base data is utilised with data import scripts and the Data Transfer Tool. DSF scenario and model connections management benefits all of the Modelling Toolbox models including the IWRM-model. Toolbox will facilitate among other things use of output from one model as an input of another model.

16

5 MODEL TEST AREAS

Two test areas, Mae Chaem and Khuwae Noi in Thailand (Figure 4), were used for the model development and testing. The areas were selected on basis on data quality and availability. They were implemented in parallel with the basin-wide application. The applications are presented in more detail in separate reports.

Figure 4. IWRM-model test areas. Left Mae Chaem in North-West Thailand (blue borders), right Khuwae Noi in Norhtern Thailand.

5.1 5.1.1

THE MAE CHAEM AREA Description of the area Sharp relief and forest vegetation characterize the Mae Chaem. The basin has a wide range of elevation, from 282 m.a.s.l. at its lowest point (P.14 Station) to 2535 m.a.s.l. at its highest peak, Doi Inthanon (Mount Inthanon). The orographic effect induces an altitudinal increase of spatial rainfall distribution. The average annual temperature ranges from 20 to 34 °C and the rainy season is from May to October (Kuraji et al., 2001). Steep hillsides with slopes exceeding 25% are a common landscape element, resulting in rates of soil erosion that prevent advanced soil development. Thus, 17

soils are relatively shallow and have limited water-holding capacity. Dominant soil textures are sandy clay loam and clay loam (Thanapakpawin et al., 2007) The Mae Chaem basin is representative of conditions commonly found in the majority of the upper tributary watersheds in mountainous mainland Southeast Asia: Approximately 90 percent of its land area is in midland and highland zones, where more than half of its population is settled. More than 50 percent of its population is composed of mountain ethnic minority communities whose traditional forest fallow agricultural systems have never been legally recognized. About 90 percent of its area is officially classified as reserved forest, national parks and/or protected watershed forest land; and there is no official land tenure in such areas. However, overall forest is believed to have decreased during the last decade, while forest fallow cycles of traditional rotational shifting cultivation systems are believed to be rapidly decreasing, making rice deficits common. National and regional-level concerns focus on deforestation in watershed headlands and water and sediment yields flowing into major reservoirs used for irrigation and electrical generation. Local concerns in downstream communities, who increasingly blame land use practices in the mountains for floods, droughts, sedimentation of water resource infrastructure, and perceived decline of water quality (Thomas 2004). The main basin characteristics are: Area 3853 km2 Average outflow 31.1 m3/s Extent EW 98° 04‟ – 98° 34‟ E Extent NS 18° 06‟ – 19° 10‟ N Outflow Ping River Elevation range 282 – 2535 m Average precipitation 1334 mm (computed) Average evapotranspiration 1085 mm (computed)

18

5.1.2

Geographical data and model grid

Figure 5. Mae Chaem DEM The model grid for the Mae Chaem basin was constructed using the following information: Digital elevation model (SRTM 90m) (Figure 5) Land use (Global Land Cover 2000, 25m gridded data) Soil data (FAO soil map of the world, 1km resolution) Mae Chaem catchment boundary All geographical data was first transformed to 1km resolution and UTM 48N coordinate system. For DEM average elevation value from the raster values within the model grid box was used for grid box elevation. For soil and land use data the most common class within the model grid box was selected to the grid box type. After the grid was ready the Mae Chaem basin area was extracted using catchment boundary data. The river data in the model grid was created from the above data by following steps: 19

elevation data was lowered by 10 m in grid boxes containing any part of the Mae Chaem river main channel for the flow network computation in order to guarantee right connection of the tributaries flow network was computed using the modified DEM and lowest neighbour principle the original unmodified DEM was used for the model grid river widths in grid boxes were set using average leaching of 50 l/s. The land use types were simplified for model computation from the previously established classes by Global Land Cover 2000 Project (GLC2000) in Table 1, thus reducing the number of land use classes in the Mae Chaem basin. The reclassification is shown in Table 2 with the percentage of area in each class. Table 1. GLC land use classification

Table 2. Land use classes in Mae Chaem basin

5.1.3

Meteorological data Precipitation data were obtained from GAME-T Rainfall Measurement Data Archive of the Mae Chaem Basin, available online. Rainfall information from 15 stations in the basin from years 1998 to 2003 were transferred to a suitable format in order to be used in the model. The locations of the weather stations are shown in Figure 6.

20

Figure 6. Location of the Mae Chaem rainfall stations (GAME-T Rainfall Measurement Data Archive of the Mae Chaem Basin) The model also used NCEP Reanalysis II meteorological data for one grid point located at latitude 18.09499931 (N) and longitude 99.375 (E). The NCEP Reanalysis II dataset is a global dataset with 2.5 degree resolution. For the modelling, Gaussian Grid 6h interval surface data was used. The data was downloaded from NCEP website for the South East Asia area, and converted for model use. Figure 7 shows comparison between observed and NCEP temperature data. For each point following data was used: (GAME-T) precipitation (6h) (NCEP) minimum temperature (6h) (NCEP)maximum temperature (6h) (NCEP) wind speed (10m level) (NCEP) relative humidity (NCEP) incoming solar radiation (NCEP) pan evaporation (6h)

21

Figure 7. Average Daily temperature at Stations P14 (observation) and W2_076 (NCEP data).

5.1.4

Hydrological data Flow and water level measurements were available for the outflow point at station Kaeng Ob Luang. A flood event, the largest in living memory for the lower Mae Chaem River (Kidson, 2003), occurred in August 2001 and can be seen in the discharge graph (Figure 8).

22

Figure 8. Daily measured discharge at P14 Kaeng Ob Luang (1998-2003) The following statistics were calculated in the IWRM-model based on the measured discharge for the years 1998 to 2003 at the station P14 Kaeng Ob Luang: average flow 31.3 m3/s, minimum flow 3.1 m3/s, maximum flow 530 m3/s. The average monthly discharges are presented in the Figure 9.

Figure 9. Average monthly measured discharge at Station P14 (1998-2003)

23

5.1.5

Model setup The IWRM model was setup using model grid and weather data presented in the previous chapters. Initial model parameterization was obtained from previous model used in the area, soil characteristics and land use characteristics. For precipitation and temperature elevation correction factors were used, that corrected the model data using the difference of elevation between the model grid box elevation and precipitation observation elevation. The correction for precipitation was +0.002 mm/meter of elevation, for temperature the correction factor was -0.006 C/meter of elevation. For evaporation computation the pan evaporation method was used. Additionally the evaporation was multiplied by 1.43 to obtain a reasonable water balance fit with the observed flows. Soil characteristics for Acrisols were obtained from Lecture Notes on the Major Soils of the World (Driessen 2001). Soil parameterization such as saturated water content, field capacity and residual water content were derived from this source using Soil Hydraulic Properties Calculator (New Mexico Climate Center). Land use characteristics and parameterization were taken from the MRC WUP-FIN Nam Songkhram model except leaf area index (LAI). LAI for various land surfaces have been defined by Hageman (2002) and these values were used in the model with some adaptations.

5.1.6

Model calibration and results The model was calibrated using the observed flows from the period 1998-2003. The main parameters considered for this process were those related with the soil model component, particularly the soil layer depth, dz. This parameter had been identified as the most sensitive for calibration in the Nam Songkhram case study (Sarkkula 2006). Infiltration parameter, infkz, was also used for calibration. Figure 10 shows the computed flow compared to measured flow at Station P14, the coefficient of determination R2 between calibrated and measured daily flow was 0.8.

Figure 10. Computed and measured discharge at Station P14 (years 1998-2003) 24

Figure 11. Computed and measured cumulative discharge at Station P14 (years 1998-2003) The cumulative flow (Figure 11) for the whole calibration period was 3.2 % smaller than measured flow values. Calibration results were reasonably accurate, in terms of behaviour of the flow. Though there were considerable differences between measured and computed – maximum and minimum flows, the average measured and computed flows were essentially the same. Model results show no great variation between the behaviour of water content in the upper soil layers at locations Ts1 and Ts2 (Figure 12). Also, it can be seen that the soil dries faster during the dry season, as expected, keeping higher moisture contents during the rainy season (where the flow [m3/s] in the graph is greatly increased).

25

Figure 12. Soil water content [mm] in soil layer 1 (s1) for different locations (Ts1 and Ts2) The location of Ts1 is found at an elevation of 469.4 a.s.m.l. and the land use class for that particular grid cell is agriculture whereas the Ts2 grid cell is at an elevation of 691.8 a.s.m.l. and it is part of the deciduous forest land use type. During the wetter periods, a higher value of soil moisture in the second layer (s2 in the model) is observed for Ts1 (Figure 13), perhaps due to the fact that in the case of Ts2, the canopy plays a major role in the interception of rainfall, thus showing smaller water content in the deeper soil.

26

Figure 13. Soil water content [mm] in soil layer 2 (s2) for locations Ts1 and Ts2.

5.1.7

Hysteresis modelling

Figure 14. Hysteretic soil water content (θ) - pressure head (h) relation for a typical soil (Stewart, 2003). The relationship between soil matric potential (φm) and volumetric moisture content (Θ) is the basis of the Soil Water Characteristic Curve (SWCC), also known as Water Retention Curve (WRC). Such curve is not a unique function since its behaviour depends on whether the soil is wetting (intake or adsorption) or drying 27

(withdrawal or desorption). Thus, soil moisture hysteresis (Figure 14) arises from differences in wetting behaviour causing dry medium to re-wet. In other words, it depends on the saturation history of the porous medium (Flynn et. al. 2005). After the Hysteresis Model was implemented in IWRM model, the time required to perform each simulation was doubled. In the case of the Mae Chaem basin this might not be a problem since the area is relatively small, but if the Mekong River basin is to be modelled, using the Hysteresis Model in the IWRM it could take a considerably longer time. Due to the fact that the results obtained after the testing of the Hysteresis Model were not significantly better than the previous simulation in the catchment without the model, it can be concluded that, in the case of the Mae Chaem basin, it is not necessary to include the hysteresis effect in soil moisture. Though it has been proved by previous studies that the phenomena associated with hysteresis has an impact on the behaviour of soil moisture (also seen in this work with the Hydrus 1D simulation), it seems that in the case of distributed hydrological modelling, the overall effect is minimal. Nevertheless, it must be said that the addition of the Hysteresis Model improved the estimated values of minimum and maximum flows in some cases, when compared to the calibration simulation case without hysteresis. It is also worth mentioning that the Hysteresis Model was able to reproduce statistical outflow values that were closer to the measured ones, compared to the outflow statistics of the simulation without hysteresis. As final conclusion, further testing of the Hysteresis Model in different catchments is required to establish whether it is indispensable to perform simulations that take into account the soil moisture hysteresis phenomena.

5.2 5.2.1

THE KHUWAE NOI AREA Description of the area and model data The Khuwae Noi river basin (Figure 15) resides in Northern Thailand and covers 5083 km2. The terrain is mostly flat in south and mountainous in north and east. The elevation varies between 50-2100 m. About 77 percent of the basin is occupied by forest and the rest is cropland and pasture. The climate of the area is dominated by monsoon rains that divide the year to wet season from April to October, and dry season from November to March. The data used in the model setup included digital elevation model, land use data, soil data, catchment boundary data, meteorological data, flow data and soil moisture data. Parameterisation of the model was done according to previous model application in the region and with the knowledge of soil and land use properties in the basin. The calibration of the model was done against measured daily average flow and soil moisture measurements. The modelling period was April 1996 – March 2000.

28

Figure 15. Khuwae Noi basin. R1, R2, R3 and H1 are precipitation measurement stations and W1, W2 and W3 are automated weather stations.

5.2.2

Model setup The IWRM model was setup using model grid presented in Chapter 3, and weather data presented in Chapter 4. Initial model parameterisation was obtained from previous MRC WUP-FI Nam Songkhram model used in the area (Sarkkula et al. 2006). Soil characteristics for Khuwae Noi river basin were obtained from Lecture Notes on the Major Soils of the World (Driesen 2001). Soil parameterization such as saturated water content, field capacity and residual water content were derived from this source using Soil Hydraulic Properties Calculator (New Mexico Climate Center) Land use characteristics and parameterisation were taken from the MRC WUP-FIN Nam Songkhram model except leaf area index (LAI). LAI for various land surfaces have been defined by Hageman (2002) and these values were used in the model with some adaptations. For precipitation and temperature elevation correction factors were used, that corrected the model data using the difference of elevation between the model grid box elevation and precipitation observation elevation. The correction for precipitation was +0.002 mm/meter of elevation, for temperature the correction factor was -0.006 C/meter of elevation. Potential evapotranspiration (PET) was calculated with Penman-Monteith method from relative humidity, air temperature, wind speed and net solar radiation measurements for three locations W1, W2 and W3. 29

The modelling or calibration period was April 1996 – March 2000. The initial condition for soil water content of the model was setup by using the end state of a model run which already had a good fit against the measured river flow.

5.2.3

Model calibration The model was calibrated against observed water flow at measuring station H1 and soil moisture at weather station W1. The observed flows used in the calibration are from the period April 1996 – March 2000 and soil moistures from the period January 1998 – March 2000. Calibrated parameters which were found to be important were vertical and horizontal hydraulic conductivities of the soil layers, infiltration, river friction, slope of the model grid and PET correction. Initial PET used in the model resulted in too low levels of evapotranspiration, therefore, calculated PET was multiplied by 1.4. Corrected PET agreed better with other evaporation measurements in Thailand (see eg. Vudhivanich 1996, Tebakari et. al. 2005). The calibration of the model produced good results. The overall dynamics, such as annual variation and major flow periods were in line with the measurements. The cumulative flow for the whole calibration period was 3.9 % smaller than measured and the coefficient of determination R2 between calibrated and measured daily flow was 0.87. However, minor differences can be noticed between calibration and measurements. This can be seen as model overestimation during the largest flow peaks. Results for the first year of modelling period may also contain some uncertainties since the initial condition of the model is not necessarily correct. Errors due to initial condition will be overcome as the modelling proceeds to the second year. Measured and modelled soil moistures are presented in the Figure 16. The measured and modelled soil moistures are not directly comparable with each other. Meas @ 40 cm represents a point measurement from single depth of 40 cm while Comp S1 represents average soil moisture of upper soil layer which was of 10 cm deep. Comp S2 represents average soil moisture of lower soil layer which was of 40 cm deep. In any case, comparison of measured and computed soil moisture show that the model was able to follow the dynamics of the measured soil moisture very well. Also, soil layers closer to surface dry faster during the dry period than the deeper soil layers. Measured soil moisture was used as a rough guidance in calibration.

30

50.00 45.00 40.00 35.00

[%]

30.00 25.00 20.00 15.00 Meas @ 40cm

10.00

Comp S1 Comp S2

5.00 0.00 1.1.98

1.4.98

1.7.98

1.10.98

1.1.99

1.4.99

1.7.99

1.10.99

1.1.00

Figure 16. Computed and measured soil moistures for the modelling period April 1996 – March 2000 in Khuwae Noi. Meas @ 40 cm is the soil moisture measurement from depth 40 cm at weather station W1. Comp S1 and Comp S2 are the computed average soil moistures of the upper and the lower soil layers.

During the calibration potential evapotranspiration (PET) had to be corrected. The initial PET was found to be too low and it was corrected 40 % upwards. The corrected PET resulted in good model fit with the measurements. Corrected PET also agreed better with other evaporation measurements in Thailand (see eg. Vudhivanich 1996, Tebakari et. al. 2005). The overall conclusion of this calibration is that IWRM model is applicable in modelling of tropical river basins in Thailand. The measured river flow was reproduced by the model very well and the modelled soil moisture behaviour confirmed the goodness of the model. However, further work is needed to improve the parameterization of different land uses. This would increase the models applicability in identifying hydrological changes caused by changes in the land use. 5.2.4

Model results Figure 17 shows the computed flow compared to measured flow at point H1. The computed cumulative flow was 3.9 % smaller than measured flow for the modelling period April 1996 – March 2000. The coefficient of determination R2 for the computed and measured daily flow was 0.87 for the whole modelling period. The results of the first year of the calibration are sensitive to the initial conditions set in the model. Therefore, R2 was calculated also for the period of January 1997 March 2000 and it had value of 0.89. Cumulative flow comparison is shown in Figure 18. Table 3 shows also some statistics of the computed and measured flow.

31

1200 Measured Computed

1000

[m3/s]

800

600

400

200

0 1.1.96

1.5.96

1.9.96

1.1.97

1.5.97

1.9.97

1.1.98

1.5.98

1.9.98

1.1.99

1.5.99

1.9.99

Figure 17. Computed and measured daily average flows for the modelling period April 1996 – March 2000.

9000 Computed

8000

Measured

7000

[Mm3]

6000 5000 4000 3000 2000 1000 0 8.1.96

8.5.96

8.9.96

8.1.97

8.5.97

8.9.97

8.1.98

8.5.98

8.9.98

8.1.99

8.5.99

8.9.99

Figure 18. Computed and measured cumulative flows for the modelling period April 1996 – March 2000. Table 3. Flow statistics for computed and measured flows for the modelling period April 1996 – March 2000.

Location H1

Average flo w

Maximum flow

Minimum flow

Cumulative flow

(m3/s)

(m3/s)

(Mm3)

3

(m /s) Measured

64,3

811,3

1,2

8107,0

Computed

62,0

1053

1,3

7786,8

32

Figure 19 shows the comparison of monthly values of computed evapotranspiration and corrected PET. Figure 20 shows comparison of computed cumulative evapotranspiration and corrected cumulative PET. 200 Corrected PET W2

180

Computed

160 140

[mm]

120 100 80 60 40 20 0 1.4.96

1.9.96

1.2.97

1.7.97

1.12.97

1.5.98

1.10.98

1.3.99

1.8.99

1.1.00

Figure 19. Computed monthly evapotranspiration and corrected monthly PET for the modelling period April 1996 – March 2000.

7000

Corrected PET W2

6000

Computed

[mm]

5000 4000 3000 2000 1000 0 1.4.96

1.9.96

1.2.97

1.7.97

1.12.97

1.5.98

1.10.98

1.3.99

1.8.99

1.1.00

Figure 20. Computed cumulative evapotranspiration and corrected cumulative PET for the modelling period April 1996 – March 2000. Figure 21 shows comparison of computed soil moistures between two grid boxes from weather stations H1 and R2. H1 is located close to the river basin outflow at an altitude of 59 m and R2 is located at the upper reaches of the basin at an altitude of 460 m.

33

60 50

[%]

40 30 20

H1 S1 R2 S1

10

H1 S2 R2 S2

0 1.4.96

1.10.96

1.4.97

1.10.97

1.4.98

1.10.98

1.4.99

1.10.99

Figure 21. Comparison of computed soil moistures between two grid boxes from weather stations H1 and R2. S1 is the average soil moisture of the upper soil layer and S2 is the average soil moisture of the lower soil layer. Figure 22 shows the vertical outflow of water from the upper soil layer of two grid boxes at weather stations H1 and R2 and Figure 23 the vertical outflow of water from the lower soil layer of the same grid boxes.

35 H1 Q1

30

R2 Q1

[m3/s]

25 20 15 10 5 0 1.4.96

1.10.96

1.4.97

1.10.97

1.4.98

1.10.98

1.4.99

1.10.99

Figure 22. Vertical outflow of water Q1 from the upper soil layer of two grid boxes at weather stations H1 and R2.

34

3.5 H1 Q2 3

R2 Q2

[m3/s]

2.5 2 1.5 1 0.5 0 1.4.96

1.10.96

1.4.97

1.10.97

1.4.98

1.10.98

1.4.99

1.10.99

Figure 23. Vertical outflow of water Q2 from the lower soil layers of two grid boxes at weather stations H1 and R2.

35

6 IWRM MODEL IMPLEMENTATION

6.1

THE MEKONG BASIN CHARACTERISTICS The Mekong river is one of the major rivers in the South-East Asia. The Mekong river basin is 795‟000 km2 large and resides in six countries, China, Burma, Thailand, Laos, Cambodia and Vietnam (Figure 24). The average daily flow is about 15000 m3/s. The climate of the area is dominated by monsoon rains that divide the year to wet season from May to October, and dry season from November to April. Some of the Mekong Basin characteristics are summarised in Table 4.

0

200

400

600 km

Figure 24. Mekong River Basin.

36

Table 4. Basin characteristics

6.2

Basin

Mekong river basin

Area

795‟000 km2

Average outflow

15‟000 m3/s

Lakes (% of area)

0.8% (dry season)

Extent EW

1560 km

Extent NS

2800 km

Outflow

South China Sea

Elevation range

0 – 6200 m

Average yearly precipitation

1600 mm (model value)

Average evaporation

940 mm (model value)

Highest point

6740m (Kawagebo peak)

SOIL CLASSIFICATION From the soil types present in the Mekong River Basin (Table 5) a new classification was obtained by analyzing the hydrological behaviour of each soil class and associating those that had common characteristics. The new classification was based on the document Lecture Notes on the Major Soils of the World (Driessen 2001) and its associated CD-ROM containing sample soil profiles for each of the 30 World Reference Base (WRB) soil types (Figure 18). Table 6 shows the proposed soil classification for the Mekong River Basin. Table 5. Soil types in the Mekong basin

37

Table 6. Soil reclassification

Properties such as sand, silt and clay percentages were obtained from the soil profiles mentioned earlier; for the case where two soil classes were merged into a new one (based on their similarities), their properties were averaged. Once the textural classes percentages were identified, parameters for the IWRM model such as thr (soil residual water content), thf (field capacity) and ths (maximum water content/saturation) were estimated by using the Soil Water Characteristics - Hydraulic Properties Calculator found in Working Paper No.5 (Sarkkula 2006) and developed by Saxton and Rawls (2006).

6.3

GEOGRAPHICAL DATA AND MODEL GRID The model grid for the Mekong river basin was constructed using the following information: Digital elevation model (SRTM 90 m) Landuse (Global Land Cover 2000, 1 km resolution) Soil data (FAO soil map of the world, 1 km resolution) Mekong river catchment boundary (Mekong River Comission, 1:50000 vector data) River bed and lake shore data (Mekong River Comission, 1:50000 vector data). The reason for using global datasets was twofold. First the applicability of these datasets was tested and second the application was more fast with them. In the future the global datasets need to be replaced with more accurate MRC data, if available. The work is planned to be implemented in connection with the modelling capacity building programme. All geographical data was first transformed to 1 km resolution and UTM 48N coordinate system. For DEM average elevation value from the raster values within the model grid box was used for grid box elevation. For soil and landuse data the most common class within the model grid box was selected to the grid box type. To construct a 5 km resolution grid the 1 km data was combined to 5 km grid boxes using minimum elevation for DEM, and most common type for soil and landuse data. After the 5 km grid was ready the Mekong river basin area was extracted using catchment boundary data. The river data in the model grid was created from the above data by following steps: 38

In river channel generation, elevation data was lowered by 10 m in grid boxes containing any part of the Mekong river main channel in order to guarantee ; this Flow network was computed using the modified DEM and lowest neighbour principle River widths in grid boxes were set using average leaching of 100 l/s. The FAO soil data was simplified for model computation, by reducing the number of soil classes in the FAO data. The reclassification is shown in Table 7. Similar reclassification is presented in Table 8. Resulting land use and soil rasters are shown in Figure 25.

39

Table 7. Landuse classes Class number

Title

1

Water

2

Decidious forest

3

Evergreen forest

4

Shrub and grassland

5

Irrigated agriculture

6

Agriculture

7

Floodplain

8

Urban

9

Glacier

GLC2000

Explanation

Reclassified as

1

Tree Cover, broadleaved, evergreen

3

2

Tree Cover, broadleaved, deciduous, closed

2

3

Tree Cover, broadleaved, deciduous, open

2

4

Tree Cover, needle-leaved, evergreen

3

5

Tree Cover, needle-leaved, deciduous

2

6

Tree Cover, mixed leaf type

2

7

Tree Cover, regularly flooded, fresh water (& brackish)

7

8

Tree Cover, regularly flooded, saline water

7

9

Mosaic: Tree cover / Other natural vegetation

2

10

Tree Cover, burnt

4

11

Shrub Cover, closed-open, evergreen

4

12

Shrub Cover, closed-open, deciduous

4

13

Herbaceous Cover, closed-open

4

14

Sparse Herbaceous or sparse Shrub Cover

4

15

Regularly flooded Shrub and/or Herbaceous Cover

7

16

Cultivated and managed areas

6

17

Mosaic:

6

18

Mosaic:

6

19

Bare Areas

4

20

Water Bodies (natural & artificial)

1

21

Snow and Ice (natural & artificial)

9

22

Artificial surfaces and associated areas

8

40

Table 8. Soil classes Model class

Title

Explanation

1

Water

Permanent water body

2

Acrisols

Subsurface accumulation of clays, low base saturation

3

Histosols

Organic material

4

Argic

Argic/Ochric horizon, sand on top, clay below

5

Ferrasols

Deep strongly weathered soils

6

Alluvial

Permanent or temporary wetness

7

Lithosols

Limited soil development

8

Cracking

Hard when dry, plastic when wet

FAO class

Explanation

Reclassified as

1

Ferric Acrisols

1

2

Gleyic Acrisols

1

3

Orthic Acrisols

1

4

Ferrasols

4

5

Gleysols

5

6

Lithosols

6

7

Fluvisols

5

8

Luvisols

3

9

Nitosols

7

10

Histosols

2

11

Vertisols

7

12

Planosols

3

41

Figure 25. Processed soil and landuse grids for Mekong river basin model.

6.4

COMPARISON OF 5 KM AND 2 KM GRIDS In addition to the 5 km resolution model, a 2 km grid was generated for testing purposes. Figure 26 shows how the increased accuracy impacts the generated river network and how land use is represented with different resolutions. In local scale the resolution has naturally great impact, but in basin-wide scale the resolution plays smaller role. The 2 km resolution model contains over 200‟000 active grid cells. This size of grid creates memory allocation problems. It may necessary to compile the model into a 64-bit application and run under a 64-bit operation system. This needs to be done with over 1‟000‟000 grid cell 3D model applications which result over 2 Gb processes. These can‟t be accommodated by a 32-bit operation system without tweaking.

42

Figure 26. 5 km (left) and 2 km (right) land use model grids.

6.5

METEOROLOGICAL DATA In addition to the MRC precipitation observation data, the model uses NCEP Reanalysis II meteorological data to drive the model. The NCEP Reanalysis II dataset is a global dataset with 2,5 degree resolution. For the modelling, Gaussian Grid 6h interval surface data was used. The data was downloaded from NCAR www-site for the South East Asia area, and converted for model use. Figure 27 shows the data points used in the Mekong basin wide model. The NCEP data was available from years 1970 to 2008. Of these years the period from 1990 onwards was used. For each point following data was used: precipitation (6h) minimum temperature (6h) maximum temperature (6h) wind speed (10 m level) relative humidity incoming solar radiation

43

Figure 27. NCEP weather data points.

Comparison of the NCEP data with MRC ground measurements shows large discrepancies with the NCEP and ground precipitation. The NASA TRMM (Tropical Rainfall Measuring Mission) data is more promising and is intended to be used in the future. The University of Washington uses this data in their VIC Mekong hydrological model.

6.6

HYDROLOGICAL DATA Flow and water level measurements were available from the lower basin only. The main river locations where flow measurements were available are shown in Figure 28.

44

Figure 28. MRC flow measurement points.

Flow characteristics of some of the measurement points are shown in Figure 29 and Table 9.

StungTreng Pakse Vientiane

50000

m3/s

40000 30000 20000 10000 0 1992

1994

1996

1998

2000

Figure 29. Monthly average flows in Vientiane, Pakse and Stung Treng.

45

Table 9. Flow statistics in Vientiane, Pakse and Stung Treng. Location

Average flow (m3/s)

Minimum flow (m3/s)

Maximum flow (m3/s)

13‟326

1‟740

41‟600

Pakse

9‟750

1‟750

30‟210

Vientiane

4‟108

1‟030

11‟340

Stung Treng

6.7

MODEL CALIBRATION The model initial parameterisation was obtained from previous models used in the area. The model was calibrated using the observed flows from the period 19902001. For precipitation and temperature elevation correction factors were used, that corrected the model data using the difference of elevation between the model grid box elevation and precipitation observation elevation. The correction for precipitation was +0.002 mm/meter of elevation, for temperature the correction factor was -0.006 C/meter of elevation. For evaporation computation a penman evaporation formulation was used. The evaporation was multiplied by 1.2 to obtain a reasonable water balance fit with the observed flows. In the Upper Mekong Basin in China the NCEP precipitation data was calibrated. The precipitation coefficient varied from 0.7 in the North-Eastern Lao PDR to 1.6 in the Northernmost NCEP point. Table 10 shows computed flow compared to flow observations in four measurement points, Chiang Saen, Vientiane, Pakse and Stung Treng. Table shows average flow and statistical measure of fit, R2. If R2=1 the fit is perfect. Even the lowest value 0.68 in Chiang Saen is statistically significant. The decreasing fit towards North is obviously caused by the data: in the Lower Mekong Basin most of the rainfall data is observations whereas in the Upper Basin precipitation is obtained from meteorological model. Table 10. Flow statistics for computed and measured flows Location

Average flow measured/ model (m3/s)

R2

13‟326/ 13‟555

0.92

Pakse

9‟750/ 10‟369

0.91

Vientiane

4‟108/ 4‟731

0.77

Chiang Saen

2‟650/ 2‟657

0.68

Stung Treng

Figure 30 shows time series of observed flows (red lines) compared with the modelled ones (black lines) in Chiang Sae, Vientiane and Kratie. Observed flows are obtained from water level measurements and rating curves and don‟t necessarily show real flows. Especially high flow peaks may not be well represented by the rating curves. Observe how modelling improves in Chiang Saen after year 2000 when improved rainfall data from China has been used.

46

Figure 30. Observed flows (red line) compared with the modelled ones (black line) in Chiang Sae, Vientiane and Kratie. Observed flows are obtained from water level measurements and rating curves.

47

The Khuwae Noi calibration results for soil moisture have been utilised in the whole Mekong model. Example of the basin-wide soil moisture simulation is in Figure 31.

Figure 31. Example of soil moisture. Blue is less porous water and red more.

48

6.8

MODEL COMPUTATIONAL TIME The basin-wide IWRM model application grid size is 5 km, or 25 km2. There are more that 32‟600 active grid cells in the model. This can be compared to the basinwide DSF application which has only 800 hydrological sub-areas. Despite the large number of grid-cells the model is quite fast. Computation time with a I7 4 GHz processor is less than 2 minutes for a year. It would be feasible to use even 1 km grid size because this would increase the computation time to a tolerable time of 50 min/year.

49

7 HYDROLOGICAL BDP DEVELOPMENT SCENARIO MODELLING

The IWRM model was preliminarily applied for the BDP scenarios. Only two scenarios were simulated: baseline and 20 year dams. The 20 year dams includes most probable hydropower future hydropower developments. China dams and 8 downstream mainstream dams are included in the 20 year dam scenario.

7.1

FLOW IMPACTS The impacts of the dams on flow regimes are illustrated in Figure 32 and Figure 33. The impacts of dams are more pronounced for the upper basin because the storage volume of the China dams compared to the flows is much larger than in the lower basin. The early flood flows are reduced and dry season flows are increased. In China the dry season flows can even double. Especially the China dam operations are not in good agreement with the modelled river flow. This is because DSF operations have been used directly in the IWRM although the hydrological modelling and reservoir inflows may differ significantly in the China part of the DSF and the IWRM models.

Figure 32. Baseline (black line) and 20 year dam (red line) scenarios. Flow in Chiang Saen.

50

Figure 33. Baseline (black line) and 20 year dam (red line) scenarios. Flow in Kratie.

7.2

COMPARISON WITH THE DSF RESULTS The results of the basin-wide IWRM model implementation can be compared with the DSF. The objectives of the comparison work are: finding possible differences in results and predictions obtaining insight in reliability of predictions (if results are same this increases confidence, if they differ they need to presented with more caution and reservations) pinpointing areas needing improvement giving guidance for future model development. It is useful to develop model comparison further into a multi-model ensemble approach. This approach is widely used in meteorology, for instance hurricane path forecasts are often presented with results from many models. Another example in which the Consultant has participated, is application of model ensemble for the Baltic Sea "as a tool for better quantitative description of Baltic Sea eutrophication necessary to increase confidence in scenario simulations of the expensive nutrient load reduction measures." 1 The approach provides a possibility to utilise strengths of different models and build confidence in results. The model results have been preliminarily compared. The impacts of the dams are similar in both the DSF and the IWRM-model. But for a proper comparison, it is necessary to include irrigation, diversions and domestic and industrial water use in the IWRM model. They are for the time being missing from it. Also operation rules for the dams need to be reviewed especially in China where the DSF and IWRM

1

http://www.fimr.fi/en/tutkimus/muu_tutkimus/eutrophication_maps/en_GB/eutrophication_maps/ 51

hydrological modelling may differ significantly and consequently the reservoir inflows in the two models can be quite different. Initially the DSF rating curves for the dams were used. Their use is problematic for two reasons: the rating curves are derived using the DSF modelled flow which doesn‟t necessarily correspond to the IWRM model and the methodology for deriving the rating curves may not be very accurate. The problem is illustrated below in a figure where IWRM model results with dynamic optimisation modelling of the China cascade are compared with Adamson and DSF results. In some months the difference is very large between the IWRM and DSF. Because of the problem only China dam results are presented below.

52

8 CHINA MODELLING

8.1

METEOROLOGICAL AND HYDROLOGICAL DATA In the previous chapter NCEP reanalysis weather data was used for the whole Mekong model. This was supplemented with MRC observed precipitation data. For the China application presented in this report MRC modelling team retrieved WMO meteorological data including precipitation, minimum and maximum temperature, wind speed and solar radiation for the period 1.1.1985 - 6.3.2010. A script was devised to read and process the data from csv-format (comma separated values text file) files and place the stations in the GIS-interface (Figure 27). Only precipitation and temperature data were used. Obviously some of the data was gap filled, for instance Nangqen station in upper part of the model area doesn't have original data after 1997, but data is filled in. A thorough presentation of the data and data gap filling can be found in Kittipong 2007. Chiang Saen station discharge was used for the model calibration. Data was obtained from the MRC Modelling Team.

Figure 34. WMO weather stations used in the model.

53

8.2

MODEL CALIBRATION The model initial parameterisation was obtained from previous models used in the Mekong region. The whole Mekong model was calibrated using the observed flows from the period 1990-2001 and NCEP weather data supplemented with the MRC precipitation data for the lower Mekong basin. For precipitation and temperature elevation correction factors were used, that corrected the model data using the difference of elevation between the model grid box elevation and precipitation observation elevation. The correction for precipitation was +0.0004 mm/meter of elevation, for temperature the correction factor was -0.006 C/meter of elevation. For evaporation computation a penman evaporation formulation was used. The China application was recalibrated because much better quality weather data was available than before. The main parameters that were recalibrated were: penman evaporation correction coefficient (0.7) rainfall correction coefficient (1.3) snow evaporation coefficient (0.05) vertical conductivity between layers (2 for all layers) horizontal conductivity in soil layer 1 (0.25) exponent (shape parameter) for the layer 2 exponential flow (0.1). In addition the scaling function for the layer 2 horizontal flow was changed from h2/hmax to min(h2,1.) in order to model better dry season flow. h2 is the layer 2 water level and hmax the layer 2 thickness. All the simulations have been conducted without reservoirs, irrigation or other diversions. The observed and simulated flow are compared in Figure 35. The selected comparison period is 1.1.2001 - 1.1.2009. R2 is 0.82 for this period which is statistically strong correlation. Measured average flow for this period is 2583 m3/s and computed one 2618 m3/s. In other words the simulated and measured water volumes differ only 1%. The whole simulation period R2 is 0.73 because of two quite anomalous years - 1992 and 1999 which may have erroneous data.

54

Figure 35. Comparison between observed flow (red line) with the simulated one. Observed flow obtained from Chiang Saen water level measurements and rating curve.

More detailed comparison in different periods are presented in Figure 30 - Figure 40. The differences between the observed and simulated results are from three possible sources: model input data including weather, soil, land use and dem is not accurate or representative model calibration can be improved model doesn't capture all hydrological processes or simplifies them too much. Despite the differences model results can be considered very good, especially taking into consideration that the model is a preliminary pilot version that needs to be updated with more accurate MRC data.

Figure 36. Comparison between observed flow (red line) with the simulated one. Period 1.5.1987 - 1.5.1990. 55

Figure 37. Comparison between observed flow (red line) with the simulated one. Period 1.5.1993 - 1.5.1996.

Figure 38. Comparison between observed flow (red line) with the simulated one. Period 1.5.1996 - 1.5.1999.

56

Figure 39. Comparison between observed flow (red line) with the simulated one. Period 1.5.2003 - 1.5.2006.

Figure 40. Comparison between observed flow (red line) with the simulated one. Period 1.5.2006 - 1.5.2009.

57

8.3

RECENT DROUGHT SITUATION

Figure 41. Comparison between observed flow (red line) with the simulated one. Period 1.1.2009 - 5.3.2010.

Figure 42. Comparison between observed flow (red and green lines) with the simulated natural flow without hydropower dams. Period 1.1.2010 - 5.3.2010.

58

Figure 41 presents the Chiang Saen observed and simulated flow for the period 1.1.2009 - 5.3.2010 and Figure 42 is a close-up to the year 2010. The green line is an updated rating curve based on 2010 discharge measurements and the red one is the previous rating curve. The simulated flow is calculated without any reservoirs and the difference between the observed and simulated flow may show impact of the Chinese hydropower operations, assuming there is no data error. With this assumption it can be concluded: the observed flow is clearly lower than the natural simulated flow for the period 15.6.2009 - 25.8.2009 and may signify filling up of Chinese dams in general observed flow is lower than the simulated natural flow for the period 9.9.2009 - 27.11.2009 but the difference is less striking than for the previous period before beginning of February 2010 observed flow is somewhat higher than simulated natural flow and after lower the observed minimum flow in 2010 is about 400 - 600 m3/s which is, although very low, is not exceptional compared to other years (see Figure 43) the updated rating curve highlights the importantce of direct flow measurements and the uncertainty in the low flow rating curves. Figure 43 shows modelled yearly minimum flows in comparison with the rating curve based flow estimates. The red line is the previous rating curve and the green line is obtained by adding 200 m3/s to the rating curve. The added value is based on the typical flow increase in using the updated rating curve for the year 2010.

Figure 43. Observed (red and green) and simulated (black) yearly Chiang Saen minimum flows (weekly averages). The green line is obtained by adding 200 m3/s to the rating curve (compare to Figure 41).

As a summary it seems rather probably that Chinese dam operations have had large impact on upper Mekong flow in 1990. The minimum flow observed at Chiang Saen is not exceptional and current record low flow down from the Chinese 59

border seems to be caused mostly by natural drought conditions. However, the Chinese dam operations may have periodically contributed to the low flow.

8.4

SNOW AND GLACIER MELTING The IWRM model was used to calculate the snow and glacier melt. Modelled average monthly and annual melting for snow and glaciers (ice) is given in Figure 44 and Figure 45 respectively. Melting is expressed in m3/s. In the model the area of glaciers is based on the FAO land use classification and is 525 km2. Ice melt degree-day factor is set to 1.2 mm/C°/d which is small compared to general observed values ranging between 5 – 8 (NIH 1998-1999). The value gives average annual melting of ice 1.14 m. In the Karakoram average ice melting (ablation) is about 3 m, but the average over the whole glacier is smaller because melting ice is fed by ice flow from higher altitudes (Hewitt, 2009). Even if the degree-day factor would be 5 times higher in the model the annual average ice melt would correspond only to 100 m3/s discharge. The annual snow melt is 500 m3/s but can reach near or over 1000 m3/s in April to July. The April – July snow melt can be important factor for the upper basin early flood flow as can be seen from figure Figure 46. The highest ice melt is in between April and October (Figure 47), but this depends on how well the upper altitude temperatures are represented in the model. The simulated annual average snow melt 500 m3/s can be compared to an earlier estimate that glacier (+snow?) melting contributes 6.6% of Mekong mean discharge, that is 730 m3/s (source?).

Figure 44. Snow melt expressed as discharge. Monthly (black line) and annual (red line) averages.

60

Figure 45. Glacier melt expressed as discharge. Monthly (black line) and annual (red line) averages.

Figure 46. Modelled Chiang Saen flow (black line) compared to weekly average snow melt (red line).

61

Figure 47. Modelled glacier melt.

Figure 48 presents simulated snow depth time series near Zadoi station in far northern part of the Mekong basin. The interesting thing is that year 2008 - 2009 there is exceptional accumulation of snow in parts of the high plateau which may have contributed to some extent to the downstream drought conditions.

Figure 48. Snow depth in two locations near Zadoi station in the uppermost part of the Mekong Basin.

62

8.5

COMPARISON WITH THE SWAT RESULTS SWAT and IWRM model results are compared in Figure 49 and Figure 50. The main differences of the SWAT results compared to the IWRM model and observations are: flow goes too much down in the late dry season beginning of flood season flow is delayed/ the early flood flow is too small the falling flood is delayed/ the falling flood flow is too small the dry season flow is first too large and then too small flow is too even. SWAT storage should be faster to release water in the beginning of the falling flood but slower in the late dry season.

Figure 49. Observed (red squares) and simulated (black squares) yearly Chiang Saen minimum flows.

63

Figure 50. Observed (red squares) and simulated (black squares) yearly Chiang Saen minimum flows.

The following three figures show impact of the China cascade on Chiang Saen flow and water levels as well as the impact on downstream flow in various locations.

64

Figure 51. Simulated daily average flows of Baseline and Dam Cascade scenarios at Chiang Saen. (Räsänen 2010)

Figure 52. Monthly average water levels of Baseline and Dam Cascade scenarios at Chiang Saen. (Räsänen 2010)

65

Figure 53. Monthly average flow changes [%] caused by Lancang-Jiang cascade at five locations in the Mekong mainstream. Changes in the four downstream locations are based on simple monthly water balance calculations. (Räsänen 2010)

66

9 CROP, IRRIGATION AND WATER TRANSFER MODELLING

9.1

BASIC MODEL DIVERSION STRUCTURE AND CONTROLS Irrigation, industrial and municipal water consumption and inter or intra basin water transfer can be defined in the user interface. Inter-basin transfer and other water uses where water is not returned to the modelled area can be defined with a river discharge control. With it water can be added or subtracted from any grid point or river flow specified. The addition, subtraction or set discharge can be specified to be either constant or read from a time series. The second option specifies water diversion into irrigation or other use with impact on basin hydrology and flow. User selects first the irrigation or other area where water is diverted to either by giving the grid cell coordinates numerically or specifying the area on map with mouse (Figure 54). The coordinates can be changed in the dialog window in the “Irrigation area” block (Figure 55). Diversion point is specified by giving either the map or grid coordinates in the “Diversion point” dialog. Water extraction can be defined from river, groundwater, lake or reservoir. Diversion amount can be defined with 3 options: crop demand, constant diversion or time series based diversion. Constant and time series based diversions are distributed to the specified irrigation or other water use area. When crop demand is selected crop mix can be defined for the selected irrigation area. When crop demand is selected only one grid cell can be selected for the irrigation area because otherwise the distribution of the crop areas within multiple basic grid cells would not be well defined. The most important feature of the crop model is that the crop areas are calculated as any other grid cell with full hydrology including infiltration, soil moisture, lateral surface and soil water flow etc. The crop areas enable practical modelling not only of many different crop types but also in general different land use types. The crop cells are sub-divisions of the basic grid cells and can be considered to increase the model grid resolution.

67

Figure 54. Definition of an irrigation and water diversion areas in the model user interface.

68

Figure 55. Irrigation, diversion and water use definitions in the IWRM model user interface. Crop modelling is based on the FAO56 (Allen et. al. 2000) and DSF/IQQM (Beecham 2003) approach. It includes the following factors: How much area and what crop type is planted each year. The area and crop mix may also be specified as a time series file that changes each year. How much water each crop type needs for evapotranspiration and ponding. How much water needs to be diverted from the river system, lake, reservoir or groundwater to meet the crop requirement.

69

How much water is actually diverted from the river system depending on the water availability. How much water is either returned to the river system as irrigation return flow or farm losses. The approach differs from the MRC DSF approach in a number of ways: Simulation of the crops, irrigation demands and hydrology is done within one model. In the DSF reference evapotranspiration is calculated with SWAT, then provided for the IQQM for crop demand calculation which is in turn provided back to the SWAT. Crop modelling is fully coupled with hydrological simulation. In the IQQM hydrology is not simulated. Full coupling enables crop and irrigation feedback on the hydrology, for instance irrigation return flow can affect available irrigation water or water availability can impact crop growth and crop status in turn water demand. In SWAT a watershed is divided into conceptual HRUs (Hydrological Response Units) that describe characteristic areas with similar hydrology. In the IWRM a watershed is divided into grid cells that use elevation, land use and soil type specified for each location. In practice the description of the watersheds and river systems is much more accurate in the IWRM than in the SWAT. For instance in the basinwide application SWAT modelling area has been divided into about 800 sub-basins but in the IWRM with 5 km resolution in about 33‟000 grid cells and in the 2 km resolution into more than 200‟000 cells.

9.2

CROP MODELLING Following the FAO56 the crop water demand modelling is based on calculation of the reference evapotranspiration. The Penman-Montheit method for the evapotranspiration has been identified as the most universal and applicable one: “The analysis of the performance of the various calculation methods reveals the need for formulating a standard method for the computation of ET o. The FAO Penman-Monteith method is recommended as the sole standard method. It is a method with strong likelihood of correctly predicting ETo in a wide range of locations and climates and has provision for application in data-short situations. The use of older FAO or other reference ET methods is no longer encouraged.” (Allen et. al. 2000) The actual crop water demand is obtained with the help of a specified crop coefficient: ETc

Kc ETo

where ETc is crop evapotranspiration [mm/d], Kc crop coefficient [dimensionless], and ETo reference crop evapotranspiration [mm/d]. The irrigation demand (Dreg) needs to take into account farm loss (lost water between diversion point and irrigation area) and excess water use that is returned to the drainage system. The total demand is then

Dreq

Ireq * (1 FIR) * (1 FOF )

where Ireg is irrigation requirement, FIR is the factor for return water and FOF the factor for farm loss. Example of the return flow and farm loss impact on downstream river discharge is shown in Figure 56. The black line shows baseline 70

without irrigation and the red one with irrigation included for dry and wet season rice. Irrigation requirement is the difference between the naturally available, precipitation and natural flow provided water, and the actual crop requirement. The crop requirement includes crop evapotranspiration and ponding water in case of rice. In the model ponding water is transferred to available surface water and evapotranspiration is not double-counted in the crop water demand.

Figure 56. Irrigation water demand for dry and wet season rice as well as downstream flow in case of no irrigation (black line) and irrigation included (red line). Use can specify crop types and corresponding crop return and farm loss monthly factors (Figure 57 and Figure 59). Model system contains default DSF crops and their factor values that can be added by “Add set types”-button (Figure 57).

Figure 57. IWRM model crop type definitions.

71

Figure 58. IWRM model crop monthly factors. Similar to the crop types user can define ponding depth types and daily ponding water depths in cm (Figure 59 and Figure 60).

Figure 59. IWRM model rice paddy water ponding types.

72

Figure 60. IWRM model rice paddy water daily ponding depths [cm]. Each irrigation area can contain many different crop areas. The pump capacities and crop areas (“crop mixes”) can be defined in the model in a crop mix dialog (Figure 61). The pump capacities and crop areas can be specified either as constant or yearly time series.

Figure 61. Definition of the IWRM model pump capacities and crop mixes.

73

10 GROUNDWATER

Previously IWRM model (VMOD) has had different groundwater formulations, but there exist public domain groundwater models that could be also used in combination with the model. The two main issues with groundwater modelling are (i) applicability of any specific groundwater model to a selected modelling scale (basin-wide to sub-basin scale) and (ii) existence of necessary soil data. In general soil characteristics such as soil type and soil thickness are poorly known in the Mekong region, and groundwater model has been designed taking this into account. Also there is indication that aquifers play minor role in comparison to saturated soil water. Model results in a groundwater irrigation case corresponding to the Figure 54 are shown in the following figures. The time series in Figure 62s hows crop demand in case of using groundwater for irrigation. Observe that the demand is seriously limited by the groundwater availability. Figure 63 shows corresponding change in downstream irrigation area river discharge. Figure 64 shows the groundwater depth in the water diversion (groundwater pumping) point. The groundwater amounts are not able to provide for the large irrigation demand in the case study and ground water is depleted fast when irrigation is applied.

Figure 62. Irrigation demand in case of groundwater pumping.

74

Figure 63. Impact of groundwater irrigation on downstream irrigation area discharge. Black line without irrigation, red line with irrigation.

Figure 64. Groundwater (soil saturated water) depth in case of no irrigation (black line) and with groundwater pumping (red line) in the groundwater diversion point.

75

REFERENCES

1

2

3

4 5 6

7 8

9

10

11 12

13 14 15

Adl, S.M., Simpson, A.G.B., Farmer, M.A., Andersen, R.A., Anderson, O.R., Barta, J.R., Bowser, S.S., Brugerolle, G. et al. 2005. The new higher level classification of Eukaryotes with emphasis on the taxonomy of protists. Journal of Eukaryotic Microbiology 52, 399-451. Allen, R., Pereira, L., Raes, D., Smith, M. 2000. FAO Irrigation and Drainage Paper No. 56. Crop Evapotranspiration (guidelines for computing crop water requirements). FAO Report, Rome. Baran, E., van Zalinge, N., Bun, N.P., Baird, I., Coates, D., 2001. Fish resources and hyrdobiological modelling approaches in the Mekong basin. Research report, Mekong River Commission Secretariat, Phnom Penh, Cambodia. Beecham, R., 2003. IQQM Irrigation Demand Model for Mekong Rvier System. Draft Technical Note for WUP-A Project. MRC Report, Phnom Penh. Driessen, P., Deckers, J., Spaargaren, O., Nachtergaele, F. (eds.) 2001. Lecture Notes on the Major Soils of the World. FAO. Flynn, D. H., McNamara, J. P., O'Kane, Pokrovskii, A. 2005. Application of the Preisach model to soil-moisture hysteresis. in "The Science of Hysteresis", Vol. 3, pp. 689–744, Elsevier Science. Hagemann, S. 2002. An improved land surface parameter dataset for global andregional climate models. Max-Planck-Institute for Meteorology, Report 336, Hamburg. Hewitt, K. 2009. Glaciers and climate change in the Karakoram Himalaya: developments affecting water resources and environmental hazards. http://wilsoncenter.org/events/docs/Hewitt_presentation.pdf Kidson, R.L., Richards, K.S., Carling, P.A., Hydraulic model calibration using a modern flood event: the Mae Chaem river, Thailand. PHEFRA (Palaeofloods, Historical Data and Climatic Variability: Applications in Flood Risk Assessment), Proceedings of the International Workshop Held at Barcelona, Spain, October 2002 (2003), pp. 171–176. Koponen, J., Sarkkula, J. 2009. MRC modelling requirements and tools analysis. Project report, Environmental and Socio-Economic modelling Tools for the Lower Mekong Basin Impact Assessment/ FINDS. Mekong River Commission and SYKE (Finnish Environment Institute) Consultancy Consortium, Vientiane, Lao PDR. 65 pp.Kuraji, K., Punyatrong, K., Suzuki, M., 2001. Altitudinal increase in rainfall in the Mae Chaem watershed, Thailand. Journal of Meteorological Society of Japan, Vol. 79, No. 1B, 2001, pp. 353-363. NIH – National Institute of Hydrology (India), 1998 – 1999. Determination of snow and ice melt factors in the Himalayan region through field investigations. 36pp. Sarkkula, J., Koponen, J., Keskinen, M., Kummu, M., Lauri, H., Rasphone, A., Souksakoun, K., Veijalainen, N. 2006. Nam Songkhram model applications. MRC/WUPFIN. Working Paper No.5. WUP-FIN Phase II – Hydrological, Environmental and SocioEconomic Modelling Tools for the Lower Mekong Basin Impact Assessment. Mekong River Commission and SYKE (Finnish Environment Institute) Consultancy Consortium, Vientiane, Lao PDR. Saxton, K. E., Rawls, W. J. 2006. Soil water characteristic estimates by texture and organic matter for hydrologic solutions. Soil Sci. Soc. Am. J 70:1569–1578. Tebakari, T., Yoshitani, J., Suvanpimol, C. 2005. Time and space analysis in pan evaporation over Kingdom of Thailand. J. Hydrologic Engrg, Vol. 10, Issue 3, 205-215. Thomas, D.E., Preechapanya, P., Saipothong , P., (eds.). 2004. Landscape agroforestry in northern Thailand: impacts of changing land use in an upper tributary watershed of montane mainland Southeast Asia. Studies based on the ASB-Thailand benchmark site 76

Mae Chaem district, Chiang Mai Province. Synthesis Report: 1996-2004. Chiang Mai, Thailand: ASB Thailand, 183pp. 16 Thanapakpawin P., Richey, J., Thomas, D., Rodda, S., Campbell, B., Logsdon b, M. 2006. Effects of landuse change on the hydrologic regime of the Mae Chaem river basin, NW Thailand. Journal of Hydrology 334, pp. 215– 230. 17 Vudhivanich, V, 1996. Monthly potential evapotranspiration of Thailand. Nat. Sci. J., 30: 392-399.

77