Simulating the hydrologic effects of land cover in the ...



Simulating the hydrologic effects of land cover and land use change in the Mekong river basin with the VIC macro-scale hydrologic model

Mariza Costa-Cabral1, Jeffrey E. Richey2, Gopalakrishna Goteti1,3,

Dennis P. Lettenmaier1

1. Department of Civil and Environmental Engineering, University of Washington, Seattle, USA.

2. Department of Chemical Oceanography, University of Washington, Seattle, USA.

3. Now at University of California, Irvine, USA.

Abstract

Concerns with the high rates of deforestation in the tropics include the possible impacts on species diversity, hydrologic response, topsoil erosion, water quality, stream sediment load, biogeochemical cycles, atmospheric chemistry, and land surface-atmosphere interactions affecting climate. Our work addresses changes in hydrologic response due to widespread replacement of forest cover by agricultural land under the monsoonal climate of Southeast Asia. Field observations have documented examples of local hydrologic effects; but how widespread are these effects and where do they occur? Do localized changes in land cover affect the stream flow at a distance downstream? How does land cover affect the flow response to rainstorms in larger basins? And how does it affect soil moisture storage and dry season flows? We address these questions using the macro-scale Variable Infiltration Capacity (VIC) model to the largest basin in Southeast Asia, the Mekong (795,000 km2), at a spatial resolution of 5 arc minutes (roughly, 9 km) and a daily time step. In a separate publication, we have successfully applied VIC to the Mekong basin for the period 1979-2000 using a detailed land cover data set representing conditions in 1997 and satellite observations of seasonal leaf area index. Here, we apply the calibrated model to hypothetical land cover scenarios, and examine the simulated changes in hydrologic response. Scenarios of decreased as well as increased levels of forest cover (termed “deforestation” and “afforestation” scenarios) are studied. In the former, we consider forest replacement by either permanent agriculture or rotational swidden agriculture, in which a fallow period is allowed after one year cultivation of any land parcel. Our model simulations show increased flow yields annually and seasonally for a greater degree of forest replacement by either permanent or swidden agriculture. Differences in yield result in altered inflows to reservoirs and possible water shortages for afforestation scenarios in currently unforested areas. An increase is seen also in flood severity, while a higher soil moisture storage results in higher baseflow and heightened stream flows in the driest periods.

1. Introduction

Current deforestation rates in the humid tropics are the highest globally, especially in Southeast Asia and in the Amazon basin. In contrast with the Amazon, Southeast Asia is in the notable position of no longer retaining but a small fraction of its original rain forest, and it maintains a high deforestation rate today (FAO, 2001).

In the largest Southeast Asia basin, the Mekong (795,000 km2), rapid population growth and socio-economic development over the last several decades has been accompanied by extensive deforestation and agricultural expansion, as well as intensified irrigation, aquaculture, and streamflow regulation. Several currently projected new major reservoirs are in response to the region’s increasing needs for hydropower. Deforestation has been mostly due to the expansion of agriculture, but logging has also lead to considerable loss of forest; and, in the case of the Yunnan province of China, so has the mining activity of the 1950s and 1960s (Hori, 2000).

In this paper, we explore the following questions:

• How does land cover change in the Mekong affect the basin’s hydrologic regime, and the risk of flood and drought occurrence? In particular, what are the effects of replacing forest cover with permanent agriculture? What are the effects of the shortened fallow periods used in swidden agriculture? And how might hydrologic regimes in the past, when much of the basin was forested, have differed from the present time?

• How great an effect does the intense irrigation taking place in the Mun and Chi sub-basins of Northeastern Thailand have in the annual and in the dry season streamflow yields? Are those effects confined to the Mun and Chi rivers, or are there significant effects on the Mekong mainstream flows?

• How do the reservoirs currently operating in the Mekong basin influence seasonal flow regimes?

Resolution of these questions calls for the application of a distributed, macro-scale hydrologic model. The VIC model was implemented and calibrated for the Mekong basin, coupled to simple models of reservoir operation and irrigation consumption, and shown to adequately reproduce the streamflow record for the 22-year simulation period of 1979-2000 (Costa-Cabral et al., in review). In this paper, we use the calibrated VIC and the same reservoir and irrigation models to simulate the basin’s hydrologic behavior for hypothetical scenarios of land cover and scenarios where irrigation consumption is eliminated and reservoirs are removed.

While macro-scale hydrologic models have recently been incorporated into Global Circulation Models (GCM), use of such models to study the impacts on the hydrologic response of river basins (Mattheuson et al., 2000) has been limited, particularly in the humid tropics. Our interest in using a macro-scale model is two-fold. First, the calibrated model can be used to investigate hydrological effects of hypothetical land cover scenarios for any sub-basin of the Mekong, creating surrogate “paired basins” for comparison. Second, it is hoped that the model application to these land cover scenarios can shed light on how hydrologic impacts are felt at different basin scales: While streamflow measurements at point, plot, or experimental catchment scale have provided evidence of hydrological effects associated with a particular change in land cover, impacts have been increasingly difficult to detect for larger basin sizes – as the effects from contrasting land covers are aggregated together, and as the streamflow signals associated with land cover change over a limited area represent a progressively smaller fraction of total flow downstream (e.g., Schulze et al., 1998; Robinson et al., 2000). Additionally, the spatial and temporal variability of rainfall over a large tropical catchment can mask the signal of land cover change.

Our scenarios consider both permanent and swidden cultivation, as well as a scenarios where all irrigation water consumption is removed, and a scenario where the basin’s reservoirs are removed. Permanent agriculture has been the focus of most modeling studies to date, representing vegetation scenarios that are the most strongly contrasting with forest cover with respect to atmosphere interactions and hydrologic response. Much tropical deforestation is, however, not given such permanent active use, but instead is cyclically abandoned after one or two growth seasons of swidden agriculture (or, in the case of the Amazon basin, after a period of pasture use), allowing secondary regrowth to develop. This regrowth, if allowed a sufficiently long fallow period, will revert to forest. Far less well studied, the possible hydrologic effects of forest clearing followed by secondary regrowth are discussed.

2. The Mekong basin

In this section, we summarize the biophysical characteristics of this large and markedly heterogeneous basin, the major land cover and land use changes occurring in the past several decades, and the major water works in the basin.

2.1 Biogeophysical characteristics of the Mekong basin

The headwaters of the Mekong River (Figure 1) are in the Tibetan Highlands, at nearly 5,000 m altitude – an area covered by a deep snow mantle except for a brief period in August and September. Fed by the melting snow, the Mekong flow rushes down the steep Tibetan slopes and through the Yunnan province of China confined to a narrow gorge. At latitude 23˚30’N, still in Chinese territory, the river enters the torrid climatic zone.

After passing the joint borders of China, Burma, and Laos (latitude 22˚15’N), the basin gradually expands in width. Its flow velocity decreases progressively, until it reaches a very slow pace at its estuary. The journey from Tibet to the South China Sea takes about three weeks in the wet season, but more than three months to complete in the dry season. The basin drains a portion of China (containing about 21% of the basin area), Burma (3%), Laos (25%), Thailand (23%), Cambodia (20%), and Vietnam (8%).

The portion of the basin lying within China, Burma, and the northern part of Laos, consisting of mountainous terrain between about 400 and 5,000 m elevation, makes up 189,000 km2 and is know as the “Upper Mekong Basin”. The remainder 606,000 km2 territory make up the “Lower Mekong Basin”.

The “Northern Highlands” include the region from southern Yunnan through Burma, Laos, and Northern Thailand, eastward into the northern end of the Annamite Cordillera in Vietnam. The “Eastern Highlands” form a southern extension of this mountainous landscape, about 700 km from Laos through Vietnam. Next to western Cambodia, which often receives over 3,000 mm of rainfall annually (according to our estimates obtained by interpolation of raingauge data), the Northern and Eastern Highlands, with elevations of up to about 2,800 meters, are the wettest regions in the Mekong basin, with annual precipitation between about 2,500 and 3,000 mm. The high mountains have deep-cut valleys and topsoil consisting of a thin deposit of sandstone and igneous rocks. Their slopes are covered by dense evergreen tropical rainforests. These mountainous regions are sparsely populated, and cultivation is limited to narrow strips of land on the banks of mountain rivers and streams. The topography precludes irrigation crops. Swidden agriculture on hillslopes uses upland rice. The Nam Mae Kok tributary of the Mekong originates in the mostly hilly and forested Burma territory. Several tributaries, including the Nam Ngum, Nam Ou, Nam Theun, Se Bang and Se Kong, originate in Laos, and contribute about a third of the total flow of the river at its mouth.

The northeastern region of Thailand, the Khorat Plateau – the poorest region in that country – belongs to the basin as does the northern tip of Thailand. The Khorat Plateau is a dry region, with annual rainfall between about 1,000 and 1,600 mm (our estimated average is 1,300 mm yr-1), and with intense evapo-transpiration. The Plateau is mainly drained by the Chi and Mun rivers. Sandy and saline soils occur and make extensive portions of the plateau unsuitable for water intensive crops like wet rice. Despite poor soil fertility, agriculture is intensive in this region, with glutinous rice, maize and cassava are the main crops. According to Pednekar (1997), forest cover dropped from 42% in 1961 to 13% in 1993 in this Thai region.

The Mekong and the Tonle Sap Lake (the largest fresehwater body of Southeast Asia) are the main features of the Cambodian plains, whose average elevation is less than 10 m. The Tone Sap river connects the lake to the mainstem Mekong river at Phnom Penh. Flow direction in this river reverses with season: roughly one-fourth of the wet season flow (July to September) from the Mekong is stored in the lake – an important mechanism of regulating floods; and in the dry season (October to April) flow is released from the lake into the mainstem – which guarantees maintenance of downstream navigability, irrigation, and prevention of saline intrusion. The size of the lake increases from roughly 2,500 km2 in the dry season to as much as 13,000 km2 in the rainy season MRC, 2003). The waters recede at the start of each year to reveal a rich bed of newly fertilized soil.

The lake is surrounded by a forested, low-lying valley with an area of about 6,500 km2. During the rainy season, the lake’s waters reach tree branch height, and fish populations use the forested area as their breeding grounds. These flooded forests are in turn surrounded by paddy fields which extend into the shallow portions of the lake. Beyond the rice fields is a sandy plain with sparse groves of trees that continue to the base of the surrounding mountains.

Rice is the predominant crop in Cambodia, where the agricultural sector accounts for half of the gross domestic product and employs 80-85% of the country’s labor force (Gartrell, 1997). Rice production in the flood plains surronding the Tonle Sap, Mekong, and Bassac rivers is the main component of agriculture. More than half of the Cambodian territory is covered with evergreen, mixed or deciduous forests. Due to commercial logging, shifting cultivation, expansion of agricultural land, and various other reasons, the forest cover in Cambodia has decreased greatly. According to Ikunaga (1999), forest cover dropped from 73% in 1973 to 63% in 1993.

The Mekong’s water resources currently sustain a relatively dense population of over 75 million people who obtain most of their protein from freshwater fish. Fish production, which is mainly from capture fisheries and partly from aquaculture, is a significant industry in the countries of the Lower Mekong basin – the main fisheries systems being the mainstem Mekong and its major tributaries, the Tonle Sap Lake, the floodplains in southern Cambodia, the reservoirs of Thailand and Laos, and the brackish water zone of the Mekong Delta. The estimated 1,200 species of fish found in the Lower Mekong basin compare to the 3,000 species of the Amazon basin, whose drainage area is about ten times larger. The combined effect of logging, agricultural expansion, stream regulation, use of agricultural chemicals, and urban sewage discharges is a concern for the future health of the fisheries resources in the basin.

Many people also rely heavily on the annual flooding cycle for crop irrigation. The population is increasing at an average rate of about 2% per year. At this rate, it will increase by about 65% to approximately 120 million people by the year 2025. Cambodia and Lao PDR have the highest growth rates – over 2.6% annually – while Thailand, Burma and the Yunnan province of China have rates below 1.6% annually. The anthropogenic changes taking place in the Mekong basin have significant implications for riverine resources. Population growth has resulted in widespread conversion of forests into agricultural uses to meet increased demand for food. Along with urbanization and industrialization, the increase in agricultural areas has led to an overall increase in demand for water.

Climate, runoff.....

|[pic] |

|Figure 1: The Mekong river basin: topography, country borders, stream gauges, and major dams in operation within 1979-2000. |

2.2 Major land cover and land use changes in the Mekong basin

Land cover in the Mekong basin has undergone dramatic changes in the last century, with the replacement of forest and other original vegetation cover with agricultural land. Logging has also lead to considerable loss of forest; and, in the case of the Yunnan province of China, so has the mining activity of the 1950s and 1960s (e.g., Hori, 2000).

Until the mid-20th century, the most widespread agricultural system in Southeast Asia was swidden agriculture – sometimes called shifting cultivation or slash-and-burn agriculture. It involved roughly one-third of the region’s territory (Spencer, 1966, cited by Rasul and Thapa, 2003), providing an environmentally sustainable way of meeting the subsistence requirements of the relatively sparse population (as was also the case in mountainous South Asia; Chazee, 1994). Forest regeneration occurs after a fallow period of a few decades; and the soil also recovers within 10-20 years, provided it hasn’t suffered excessive burn (Chazee, 1994).

More recently, either under pressure of population growth or due to government land policies, swidden agriculture was replaced by permanent sedentary agriculture in some of its traditional regions, such as northern Thailand (as had earlier happened in Indonesia and Malaysia). Today, permanent agriculture in the Mekong basin occupies the vast majority of northwestern Thailand (where rice, cassava, maize, sugar cane, and other crops are grown) and smaller areas in Cambodia (maize, rubber, soy beans, and other crops), Vietnam (rice, sugarcane, corn, cereals, and other crops), and Burma.

Swidden agriculture continues to be widely practiced in the mountains of Laos – where it remains the major source of livelihood today, with principal crops of rice, maize, cassava, soybeans, sesame, and vegetables (Rasul and Thapa, 2003) –, as well as in the upper Da watershed of northern Vietnam (Fox et al., 2000), and in the southern Yunnan province of China (Xu et al., 1999). (This is also still the case in the mountains of Bangladesh and northeast India.) In some areas of Laos, over 90% of households depend on swidden agriculture for their subsistence (Sandewall et al., 2001).

The curtailment of swidden agriculture in Thailand, where in the early 1970s it still flourished, owed to the notable scarcity of remaining forest in that country (after decades of intensive commercial logging, and large expansion of permanent agricultural fields), and to a variety of targeted policies by the Thai government. In 1972, national parks and wildlife reserves were established in Thailand covering 2 million ha of forest. Restrictions were imposed on cultivating steep slopes, and watersheds have been classified for erosion vulnerability. An extensive road network built in the 1970s and 1980s in northern Thailand, offered sedentary cultivators access to produce markets (Van Turkelboom et al., 1996, cited by Rasul and Thapa, 2003). Usufruct rights to permanent agricultural land were granted to former swidden cultivators to further incentivate their conversion to sedentary agriculture (Feder et al., 1988). The increasing scarcity of land for swidden agriculture in mountainous Thailand – further aggravated by government afforestation and soil conservation projects – at first forced a shortening of fallow periods. And as the benefits of converting to sedentary agriculture grew larger, swidden agriculture was finally abandoned by nearly all. The exceptions that remain as swidden cultivators in Thailand today are mostly those groups living in mountainous areas near the Burmese and Laotian borders who are classified as illegal aliens by the Thai government, and hence do not have access to usufruct rights to Thai agricultural land (Rasul and Thapa, 2003).

In predominantly mountainous Laos, where population density remains low and forested land remains relatively plentiful, the area subject to swidden agriculture was estimated at 2,189 km2 in the mid-1990s, employing some 1.24 million people, or about one-third of the country’s population (Pravongviengham, 1998, cited by Rasul and Thapa, 2003). (Boupha, 1995, cited by Hori, 2000, indicates a similar value of 1,900-2,000 km2 for 1995.) In Laos, this type of agriculture depends on the cyclic clearing of secondary vegetation and sometimes of primary forest (Rasul and Thapa, 2003), and involves very limited application of fertilizers or pesticides (Pravongviengham, 1998, cited by Rasul and Thapa, 2003). Invoking economic and environmental problems, the Laotian government, under much foreign pressure, had announced a plan to eradicate swidden agriculture by the year 2000. But the effects of its legal measures were minor due to weak enforcement and popular opposition (see Thapa, 1998). Forest area available to swidden cultivators in Laos has, however, become considerably limited, hence the rotational cycle in that country has been greatly shortened in the last few decades, from a traditional value of 15-25 years of fallow, to only 3-5 years today (Hansen, 1998). It is not known whether such a short fallow period will allow retaining of soil fertility and continuation for much longer of swidden cultivation at the current intensity.

A notable characteristic of swidden agriculture is that it does not involve large continuous land areas but is fragmented, creating a patchwork of secondary vegetation and primary forest; and the secondary vegetation itself is frequently a mosaic of crops, grassland, and forest regrowth at various stages of development. Fragmentation is often also seen in areas with permanent agriculture. In fact, Laurance and Bierregaard (1997), editors of work surveying fragmented forests worldwide, conclude that “fragmented landscape is becoming one of the most ubiquitous features of the tropical world – and indeed, of the entire planet.” Other authors (Brown and Lugo, 1990; Whitmore, 1997; and Giambelluca, 2002) have found that most tropical countries now have larger areas of secondary vegetation than primary forest. There has been considerable interest on the ecological effects of tropical land-cover fragmentation (Laurance et al., 1988) as well as possible effects on land-atmosphere interactions (Avissar and Pielke, 1989). Hydrological effects and their modeling have, however, not received much attention to date (section 2.2).

2.3 Major water works in the Mekong Basin

The major water infrastructures in the basin are the dams, which were constructed over the last four decades in China, Laos, Vietnam and Thailand (Figure 1). Massive hydroelectric dams are planned or have already been constructed along the Mekong and major tributaries (Figure 2). Environmentalists have been debating negative environmental impacts.

| |

| |

| |

|Figure 2: Existing and proposed dams in the Mekong basin. The dams in operation for part of our simulation period (1979-2000) are |

|Manwan, Nam Ngum (1), Ubol Ratana, and Pak Mun (also shown in Figure 1). |

The countries of this region view export of electricity, especially to relatively rich Thailand, as an important means of acquiring foreign currency. Construction of dams and reservoirs in the basin could significantly alter the mainstem flow during the rainy and dry seasons. Wet season flow is critical to fish and rice production in the Mekong Delta and the Cambodian plains. Decrease in dry season flows could affect transport and commerce and increase seawater intrusions in the Delta.

Hydroelectric power dams have been planned at eight sites in China’s Yunnan province (Chapman and He, 2000). Of these eight sites, the Manwan dam was completed in 1993, near the city of Lincang in Yunnan. The damming of the upper portions of the Mekong has the potential to affect the downstream discharge patterns significantly, and concerns have been raised about possible environmental damage (Champan and He, 2000; Hinton, 1998).

Dams at various locations in Lao PDR have been proposed by the government, to produce power for export to neighboring Thailand (Dansie, 1994; IRN, 1999). The Nam Ngum dam, which was completed in 1985, and the Nam Theun Hinboun and Houay Ho trans-basin diversion schemes, which were completed in 1998, were designed to supply power to Thailand. Poor financial viability, problematic environmental impact assessments and inadequate resettlement practices are some of the causes for public protest against some of these projects (IRN, 1999).

In the Thailand part of the Mekong basin, which has relatively low hydropower potential, most of the hydropower has already been exploited. Building of new dams is unlikely in this region because of environmental movements against the construction of large dams (Rigg, 1995). The Pak Mun Dam on the Mun river, which was completed in 1994, is one such example. Fish populations, which are an important source of income for the local people, have decreased significantly following completion of the dam (reference?). Furthermore, the dam has not functioned at its designed capacity for various reasons (Amornsakchai et al., 2000). As a result of public protest, the gates of the Pak Mun Dam were opened temporarily in June 2001 to asses the impacts of the dam on fisheries of the Mun river.

Political and economic stability of Cambodia is a prerequisite for hydropower development in the Cambodian portion of the Mekong basin. Hydropower development projects have been proposed in several parts of Cambodia (FAO Cambodia, 1996). The existing Yali Falls dam on Se San River and the proposed Pleikrong dam in Vietnam are designed to provide electricity for central and southern parts of Vietnam (Miller et al., 1996). The irregular water releases from the reservoir of the Yali falls dam have been a serious concern for the downstream people of Cambodia (NTFP, 2000).

3. Description of the hydrologic and water management models

To simulate and quantify some of the land cover change effects on hydrologic fluxes described qualitatively in the previous section, we use the Variable Infiltration Capacity (VIC) hydrologic model (Liang et al., 1994 and Liang et al., 1996), which incorporates and is sensitive to vegetation parameters such as vegetation height, rooting depth, leaf area index, as well as albedo. VIC is suitable for application to large river basins, or entire continents. The VIC model has been applied to all river basins in the US (Abdulla et al., 1996; Nijssen et al., 1997; Cherkauer and Lettenmaier, 1999; Maurer et al., 2002); to the entire country of China (Su and Xie, 2003); to the pan-Arctic region (Su et al., 2004); and to the entire globe (at 2° resolution; Nijssen et al., 2001).

3.1 The VIC hydrologic model

VIC’s major distinguishing characteristics among macro-scale hydrologic models are that it represents soil moisture storage capacity as variable within each model grid cell, having a spatial probability distribution (following the Xinanjiang model of Zhao et al., 1980, described in Wood et al., 1992), and that it represents base flow as a nonlinear recession (Dümenil and Todini, 1992) like in the ARNO model (Todini, 1996). VIC describes a vegetation cover type through parameters including leaf area index, stomatal resistance, root mass distribution with soil depth, and others (section 3.2) and allows multiple vegetation cover types in each grid cell (the percentage of each type being specified by the user). Water balance and energy balance equations are solved simultaneously.

The subsurface is characterized in the vertical direction by a user-specified number of soil layers, usually two or three (in this study, three layers are used). The top soil layer contributes runoff via fast response mechanisms, whereas the deepest soil layer produces base flow. Drainage between the soil layers is modeled as gravity driven. Controls of vegetation on evapo-transpiration are represented explicitly using a Penman-Monteith formulation (Liang et al., 1994). Sub-grid variations in precipitation rate and in temperature, due to variations in elevation, are represented by sub-dividing each grid cell into a number of elevation bands. The effects of snow accumulation and melt are represented using an internal coupled snow model described in Storck and Lettenmaier (1999).

The VIC model can be operated in one of two modes: an energy balance mode and a water balance mode. In the energy balance mode, all the water and energy fluxes near the land surface are calculated, and the surface energy budget is closed by iterating over an effective surface temperature. In the water balance mode, the effective surface temperature is assumed to equal the air temperature and only the surface water balance fluxes are calculated. In this study, the VIC model is operated in the water balance mode, which is equivalent to the manner in which most operational hydrological models function. Precipitation, maximum and minimum temperature, and wind speed are the meteorological variables that drive the model in the water balance mode. Hourly temperatures are estimated by fitting a spline function to the time series of daily minimum and maximum temperatures. Daily precipitation inputs are distributed uniformly in time throughout the day.

The model used to simulate the routing of streamflow along the stream network is described in Lohmann et al. (1996; 1998). This model uses a triangular unit hydrograph and linearized St. Venant’s equations to route the streamflow from each individual grid cell separately to the basin outlet through the channel network. This model does not account for channel losses, extractions, diversions, or reservoir operations. The latter are represented in the water management model, described next.

3.2 The water management model

The Mekong Water Management Model (MWM) provides a simplified representation of the main dams and reservoirs existent in the Mekong basin during the simulation period of 1979-2000 and their operating practices. The MWM model operates on a monthly time scale and is driven by streamflow data simulated by the VIC model. The characteristics of the reservoirs included in the model implementation to the Mekong basin are summarized in Costa-Cabral et al. (in review).

Dams which are not run-of-the-river but are “storage reservoirs” are assumed to be operating in two major planning periods each year. In the wet season, when dams have more than sufficient flows to meet their particular demands, part of the inflow is used to fill the storage, the outflow thus being reduced compared to the inflow. In the dry season, when inflow is inadequate to meet the demand, water is released from storage to fulfill the demand.

The operation of storage reservoirs is prescribed by assumed rule curves, given that specific information was not available. The cross-section of storage reservoirs was assumed to be rectangular, and a minimum storage level was specified that corresponds to the minimum head required for operation of the dam. The power, P, generated by each dam is a function of the discharge that passes through its turbines, Q, and its associated hydrostatic pressure head, h. Noting that P=ρ Q g h η (where ρ is the density of water, g is the acceleration of gravity, and η is the efficiency of the power generating system), reservoir storage is estimated from h for the value of P produced by the reservoir. When the actual P is not known, the P for which the reservoir was designed is then used. Similarly, the water volume released for irrigation was the observed or, when unavailable, the designed release.

The only operational target to be met by the storage level of the hydropower dams is the observed, or designed, power output. The operational targets to be met by multipurpose (irrigation and hydropower) dams were the observed, or designed, irrigation release and the observed, or designed, power output.

3.3 Irrigation consumption model

Irrigation is essential to rice cultivation at several locations of the Mekong basin, even in the wet season when fluctuations of rainfall frequency occur, and especially in years of below-average rainfall totals. In the Khorat Plateau of Northeast Thailand and in the vicinity of the Tonle Sap Lake in Cambodia, direct rainfall is insufficient to sustain rice cultivation through much of the year. Irrigation is also conducted over much of the Vietnam Delta (see land cover class 41 in Figure 3).

Most irrigated areas in the Mekong basin correspond to rice paddies. Some shrimp farms may have been miss-classified as irrigated areas, based on the satellite images. Crop yield in irrigated areas is superior to that in rain-fed areas, where rainfall is often insufficient to guarantee good harvest of the entire planted area. Two varieties of rice are used: the wet season variety, and the more productive dry season variety. In most irrigated rice fields in Thailand, only the wet season variety is cultivated, with 2.33 million ha planted in 1995 and a yield of 3.38 ton ha-1; but in some irrigated areas the dry season variety is also cultivated, with 0.69 million ha in 1995 and the higher yield of 4.34 ton ha-1 (Charnvej, 1999). In these irrigation areas, two rice crops are hence produced annually. The fraction of the total irrigated area in which dry season rice is planted (which from the above 1995 figures is calculated at 23%) varies from year to year, a decision made by the cultivators based on the amount of water available in storage at the end of the wet season.

The water requirement to produce optimal crop yield must satisfy the evapo-transpiration of the crop and losses from the paddy field through percolation and seepage. The percolation losses for paddy field rice depend on the soil type, and are grossly estimated at 1mm day-1 for heavy soil (high clay content) and 2-3 mm day-1 for light soil (low clay content). We use the crop water requirements for rice estimated by Schreider et al. (2002) (Table 1). The estimate for dry season rice given by the Thai Royal Irrigation Department, for the country as a whole, is 6-7 mm day-1 (Charnvej, 1999), a lower value than the dry-season average in Table 1.

Table 1: Irrigation water demand (mm) of crops in northern Thailand (from Schreider et al., 2002).

| |

|Figure 3: Land cover map of the Mekong basin (see Costa-Cabral et al., in review). Classes 21, 22 and 23 (cropping mosaics and |

|agricultural land) are also present in the Chinese section. |

The methods used to produce the map portion for the Lower Mekong Basin were described by Stibig (1999), and are summarized next based on that publication. The classification of forest cover (classes 1-7 in Figure 3) was given whenever the fraction cover of tree crown was at least 20% and the tree height (estimated from canopy texture in satellite images) above 10 meters. The 20% crown cover threshold was chosen due to the open nature of natural dry Dipterocarp forests, while the 10% threshold used by FAO was judged too low for the tropical forests of this region.

For evergreen and for mixed forests, different classes of cover density were distinguished, based on the satellite images: high, low, and medium to low density. Small reductions of tree cover can in general, however, not be discerned in the satellite images, because openings are often filled by bamboo or other pre-existing or secondary vegetation. For deciduous forests, different density classes could not be distinguished.

• Classes 1, 2: Evergreen comprise the typical lowland tropical rain forests (Dipterocarpus spp., Shorea spp., Parashorea stellata, Anisoptera spp., Dyera spp.), the hill evergreen forests (Cinnamomum spp., Fraxinus spp., Podocarpus spp., Quercus spp.), and the dry evergreen (or ‘semi-evergreen’) forests (Dipterocarpus alatus, Hopea ferrea, Anisoptera costata, Afzelia spp.), and coniferous forests (Pinus keysia, P. merkusii). Coniferous forests (some of which are planted) are found in the Central Highlands of Vietnam and on the plateaus of Laos and Thailand above 800 meters elevation, and are often open forests, with some broadleaf species and allowing room for shrub and grass layers.

• Classes 4, 5: Mixed forests of evergreen and deciduous trees (Tectona grandis, Lagertroemia spp., Pterocarpus spp., Terminalia spp., Dalbergia spp.) may contain 30-70% of evergreen or deciduous trees. The range from the moist mixed deciduous forest (which, having a high fraction of evergreen trees and bamboo undergrowth, have an appearance close to that of an evergreen forest), to the mixed deciduous forest, and the more humid type of dry deciduous forest.

• Class 7: Deciduous forests lose most of their leaves during the dry season, and comprise the dry mixed deciduous forests and the dry Dipterocarp forests (Dipterocarpus obtusifolius, D. tuberculatus, D. intricatus, Shorea obtuse). Canopies are open, with 20-70% fraction of canopy coverage. Dry Dipterocarp forests occur in areas with annual precipitation under 1,200 mm, crown cover can be as low as 40% even in undisturbed forests, and tree height is often low (below 8 meters).

• Classes 9, 10: Forest regrowth consists of a continuous, usually dense layer of evergreen trees. This class is assigned whenever tree heights are between 5 and 10 meters. Class 10, forest regrowth, inundated, represents the inundated areas surrounding the Tonle Sap Lake.

• Class 11: Inundated forests are those inundated by the Tonle Sap Lake in the wet season. Most have low-height trees and are undisturbed, but some are degraded and only a mosaic remains of them (often caused by charcoal production). This class includes some swamp forest.

• Class 12: Mangrove forests appear in the Mekong Delta in Vietnam.

• Class 13: Forest plantations include rubber and eucalyptus and, when detectable, also teak and pine. Early growth stages were not included in this class.

• Classes 16, 19 and 20: Woodland and Shrubland have a mixture of shrubs, grass and trees, tree cover fraction below 20%, tree height under 5 meters. The evergreen type (Class 16) is found mainly on shallow soils, on mountain tops under climax conditions or as a result of non-sustainable land use, usually with a dense layer of shrubs and grass, small trees, and some bamboo. There is the possibility of it developing into forest again. The dry type (Class 19) is located in the dry plains or plateaus of the southern part of the basin, and on dry, sub-exposed slopes. The inundated type (Class 20) covers the degraded inundated areas surrounding the Tonle Sap Lake, often with a dense layer of small trees (too small to be classified as forest).

• Classes 21 and 22: Cropping mosaic represents shifting cultivation after land abandonment. It contains not only currently cultivated fields (30% for class 22), but also areas in various fallow stages, with shrubs and forest regrowth. These areas can once more become forests if they are not again cleared.

• Class 23: Agricultural land may contain various of crops, and includes permanent agricultural fields (mainly paddy fields), coffee and tea plantations, and mixed areas where agriculture is the dominant cover. Permanent agriculture on slopes, frequent in Vietnam’s Central Highlands, is difficult to distinguish from shifting cultivation.

• Class 30: Broadleaf deciduous forest has a crown cover varying from 60-70% and 10-20% depending on season. Class 31, Bush, consists of closed shrubland, and class 32, Sparse woods, consists of sparse tree cover.

• Classes 33-37, and 41 are all open grassland classes, growing on steep terrain (class 33, alpine and sub-alpine meadow), on less steep but sloping terrain (class 34, slope grassland), on level ground (class 35, plain grassland), on loose and shifting sands (class 36, desert grassland), meadows (class 37, meadow), or at high altitudes (class 41, alpine and sub-alpine grass).

Grouping these many land cover classes into broad categories, their distribution for the different countries is given in Table 2. The most forested country is Cambodia, followed by the Vietnam Highlands and Laos; while the most intensely cultivated country is Thailand, composed of nearly 80% of agricultural land, nearly as much as the Vietnam Delta region (84%).

Table 2: Percent distribution of land cover groups within the Mekong basin in 1997, within each country (calculated from the data in Figure 3).

|Land cover (%) |

|9A |

Table 4: Summary of the seven hypothetical land cover scenarios. Variables Ω and f are defined in the text. In this table, Ω0 denotes the value of Ω for the current scenario, equal to 26,249 km2.

|Scenario |

|Sc.1 |

|Sc.5 |

|A |

|[pic] |

|B |

|[pic] |

|C |

|[pic] |

|D |

|Figure 4: Average monthly flow rate in 1980-2000 at different gauge locations for the different simulation scenarios. (To eliminate|

|the effects of soil moisture initialization, the first year of simulation, 1979, was excluded from the computations of monthly |

|means.) |

Table 5: Simulated mean monthly flow rate (m3s-1) at each stream gauge for the current land cover scenarios; and percent change in monthly flow rate for each hypothetical land cover scenario relative to the current land cover scenario. See Table 4 for description of the scenarios. For gauge locations, see Figure 1.

|Scenarios |month |

| |

| |

| |

| |

| |

| |

| |

| |

| |

| |

| |

| |

| |

| |

|A |

|[pic] |

|B |

|Figure 5: Exceedance probability of the simulated daily average flow rate in 1979-2000, for different land cover scenarios. |

Drought periods

What are the implications of the lower flows seen in the afforestation scenario, for drought conditions at different time scales? Are the minimum daily annual flows also lower in this scenario (and higher in the deforestation scenario)? Is the driest week, the driest two-week period, and the driest month of the year considerably drier in the afforestation scenario? To answer these questions we look at the change in frequency of the minimum annual flows, shown in Figure 6 for selected stream gauges. Indeed, at a wide range of different aggregation periods, the driest periods are drier still for the afforestation scenario, but less dry for the deforestation scenario.

Stream flows in such dry periods are dominated by baseflow. Is the lowering of minimum flows in the afforestation scenario explained by conditions of greater soil moisture depletion? Figure 7 shows the frequency of annual minimum weekly mean soil moisture content over the sub-basin, for different scenarios and selected sub-basins. Indeed, in the driest week of the year, soil moisture is most depleted in the afforestation scenario, and least in the afforestation scenario, thus originating the differences in baseflow previously seen in Figure 6. The simulated mean monthly soil moisture for the current scenario is displayed in Figure 12 in Costa-Cabral et al. (in review).

|[pic] |

|A |

|[pic] |

|B |

|[pic] |

|C |

|Figure 6: Frequency of minimum annual flows at sample gauge locations, for different aggregation periods and different scenarios. |

|For each curve, the 22 annual values of minimum flows were ranked (mi being the rank of year i) and return period is given by |

|Tr=(22+1)/mi (see, e.g., Linsley et al., 1982,, page 375). |

|[pic] |

|A |

|[pic] |

|B |

|[pic] |

|C |

|Figure 7: Frequency of annual minimum weekly mean soil moisture content over the sub-basin, for different scenarios and example |

|sub-basins. Note that the y axis does not start at the origin. For each curve, the 22 annual values of minimum weekly soil moisture|

|content were ranked (mi being the rank of year i) and return period is given by Tr=(22+1)/mi (see, e.g., Linsley et al., 1982, page|

|375). |

Floods

How does the basin’s response to the most severe rainfall events change between land cover scenarios? How do flood magnitudes change for different return periods?

Here, we fit a Gumbel distribution to the series of 22 annual maximum flow values (for each scenario), and use the fitted distributions to extrapolate up to a return period of 100 years (Table 6 and Figure 8). The Gumbel distribution fit which uses only 22 annual values cannot be extrapolated for any larger return periods. Moreover, for floods of very large return period (>100 years), the effects of land cover may be reduced, as such floods are likely to be preceded by wet periods leading to soil saturation regardless of land cover, so that flood response possibly is dominated by the depth and intensity of the rainfall.

For a given return period, the flood magnitude computed is indeed higher for the deforestation scenarios, albeit by only a few percentage points (Table 6, and Figure 8a for Sc.2) and lower for the afforestation scenarios (Table 6 and Figure 8b for Sc.4). The increase in flood magnitude decays for larger return periods. The simulated impact of eliminating all current swidden agriculture (Sc.7) is very modest (Table 6).

The largest change is the reduction of about 40% in flood magnitude achieved by the afforestation scenario at Ubon – which drains the Mun-Chi basin, that currently is nearly totally dedicated to agriculture. Is this simulated drastic change realistic? How much less severe had flood events been in the Mun-Chi basin prior to its extensive deforestation and coversion to agriculture in the last few centuries (most marked over the 20th century)? We do not have the historical data to answer this question. One consideration is that the model calibration we performed (in Costa-Cabral et al., in review) led to soil depths of more than 3 meters over this basin, without which the low empirical runoff ratios, obtained by comparing rainfall and runoff totals, could not be reproduced, and agreement between observed and simulated hydrographs could not be achieved. The larger soil water content allowed by a deeper soil leads to higher evapo-transpiration rates. However, if there is a bias in the estimated rainfall rates, due to the sparse raingauge network available, then possibly these calibrated soil depths may be over-estimated. Shallower soil depths demonstrate less sensitivity of flood peaks to land cover, in trial simulations. Hence, over-estimated soil depths may possibly be leading to exaggerated differences between estimated flood magnitudes of the current and afforestation scenarios.

|[pic] |

|A |

|[pic] |

|B |

|Figure 8: Percent change in flood magnitude when current land cover is replaced by the deforestation scenario Sc.2 (panel A) and |

|the afforestation scenario Sc.4 (panel B). Results for return periods up to 100 years were obtained by fitting a Gumbel |

|distribution to each 22-year series of simulated maximum annual flows. |

Table 6: Computed maximum annual flood magnitude (m3s-1) for different return periods for selected stream gauge locations and scenarios, based on a fitted Gumbel distribution (see text), for different stream gauges and land cover scenarios.

|Scenarios |Return Period |

| |(yr) |

| |2 |10 |100 |

|Chiang Saen: |

| |Current |23,720 |41,689 |64,103 |

|Sc.2 |100% deforestation |24,694 |42,623 |64,986 |

|Sc.4 |100% afforestation |22,918 |40,824 |63,158 |

|Sc.7 |No swidden agric. |23,671 |41,567 |63,889 |

|Nakhon Phanom: |

| |Current |62,872 |83,814 |109,935 |

|Sc.2 |100% deforestation |64,879 |85,704 |111,680 |

|Sc.4 |100% afforestation |60,258 |80,788 |106,396 |

|Sc.7 |No swidden agric. |62,776 |83,505 |109,361 |

|Ubon: |

| |Current |6,888 |15,015 |25,152 |

|Sc.2 |100% deforestation |7,095 |15,416 |25,796 |

|Sc.4 |100% afforestation |4,293 |9,615 |16,254 |

|Sc.7 |No swidden agric. |6,886 |15,011 |25,146 |

|Stung Treng: |

| |Current |126,323 |180,138 |247,263 |

|Sc.2 |100% deforestation |129,990 |183,910 |251,167 |

|Sc.4 |100% afforestation |120,928 |174,680 |241,726 |

|Sc.7 |No swidden agric. |126,105 |179,966 |247,147 |

6. Impacts of reservoirs and irrigation on streamflow

6.1 Flow input to reservoirs

The four main Mekong reservoirs that were in operation for some portion of the 1979-2000 simulation period, and are represented in our water management model, were introduced earlier. These reservoirs and their operation-start dates are the Manwan dam (started 1995) on the main stem of the Mekong in China; the Nam Ngum dam (started 1985) in Laos; and the Ubol Ratana (started 1966) and the Pak Mun (started 1994) dams in Thailand. See Figure 1 or 2 for the location of these reservoirs.

The simulated flow input to each of these reservoirs is given in Table 7. Impact of the more extreme scenarios, Sc.2 (100% deforestation) and Sc.4 (100% afforestation), is considerable, especially of the latter in the two dams located in the Mun-Chi basin – Ubol Ratana and Pak Mun –, with severe drops in inflow rates throughout the year.

Table 7: Simulated mean monthly inflow rate to reservoirs (m3s-1).

|Scenarios |month |

| |

|Current |

|Current |

|Current |

|Current |70 |

| |

|With |

|Irrigation |

|With |

|irrigation |

With

irrigation |(m3s-1) |65 |63 |72 |116 |240 |457 |631 |1,414 |2,404 |1,719 |439 |124 | |Without irrigation |(m3s-1) |77 |73 |81 |116 |248 |466 |642 |1,419 |2,306 |1,721 |18 |18 | | |% change |+18.5% |+15.9% |+12.5% |0 |+3.3% |+2.0% |+1.7% |+0.4% |0 |+0.1% |+4.1% |+14.5% | |

7. Conclusions

In Costa-Cabral et al. (in review), the VIC model was calibrated for the Mekong basin, and was demonstrated to adequately reproduce the historical flow record, yielding our “historical simulation”. In this paper, the calibrated model was used to simulate hypothetical land cover scenario conditions, described in section 5.2. Use of macro-scale hydrologic models to study the impacts of land cover change on the hydrologic response of river basins has been limited, particularly in the humid tropics.

We used the calibrated VIC model to obtain estimated daily flows for the same period 1979-2000 for hypothetical land cover scenarios of various degree of forest replacement by permanent agriculture in the Mekong basin, ranging from an extreme 100% deforestation scenario – in which all forest cover is replaced with cropland – and an extreme 100% afforestation scenario – in which all current cropland, bushland and sparse woodland is replaced with low-density evergreen forest; and scenarios of rotative deforestation for swidden agriculture, followed by forest regrowth during a fallow period of specified length.

Simulation results indicate higher flow yields for the deforestation scenarios and lower for the afforestation scenarios, when compared to the current scenario (Table 4 and Figure 4). The differences in flow yields were seen when aggregating flows at any interval of time, from daily flows, weekly, monthly, to total annual yields.

Drought periods at all temporal aggregation scales (the driest day, driest week, driest month and driest season) were also rendered even drier by the afforestation scenarios, but less dry for the deforestation scenario. This results from corresponding differences in mean soil moisture content, whose simulated values were similarly affected at all aggregation periods.

Differences in yield were largest where currently deforested areas were completely afforested: the intensely agricultural Mun-Chi basin. Reduced yields following afforestation resulted in large drops in simulated inflows to reservoirs, throughout the year.

Eliminating all current swidden agriculture and allowing forest regeneration has a simulated impact of reducing flow at main stem gauge stations, by at most 2 percentage points (Table 5). The effect is however negligible at the Mun-Chi basin gauges (Ban Chot, Rasi Salai, Yasothon, and Ubon), given the current sparseness of total forest there. Swidden agriculture scenario Sc.5 – in which an area five times larger than that (estimated) for the current scenario would be available to swidden cultivators, and the current (estimated) fallow period of 5 years would be maintained –, has significant impact on flows at main stem gauges. The impact of scenario Sc.6, in which the fallow period is allowed to increase to 30 years, is much more modest.

Flood peaks were made more severe for the deforestation scenarios, by a few percentage points, but this difference became less pronounced for floods of higher return period (more severe floods) compared to smaller floods. Flood peaks became less pronounced for the afforestation scenarios, and especially so in the case of the Mun-Chi basin of Thailand.

For a given return period, the flood magnitude computed is indeed higher for the deforestation scenarios, albeit by only a few percentage points (Table 6, and Figure 8a for Sc.2) and lower for the afforestation scenarios (Table 6 and Figure 8b for Sc.4). The increase in flood magnitude decays for larger return periods. The simulated impact of eliminating all current swidden agriculture (Sc.7) is very modest (Table 6).

The largest change is the reduction of about 40% in flood magnitude achieved by the afforestation scenario at Ubon – which drains the Mun-Chi basin, that currently is nearly totally dedicated to agriculture. We do not know whether this dramatic simulated change is realistic, or whether it is amplified in our simulations by particular values of calibration values or vegetation parameters used, all of which are subject to considerable uncertainty.

Acknowldegements

Lauren McGeoch (U.W.) helped prepare Figure 1. Financial support for this work came in part from the Functional Value of Biodiversity Program, funded by the World Bank-Netherlands Partnership Program. All findings and interpretations belong to the authors alone and should not be attributed to the Bank, its Executive Board of Directors, or the countries they represent.

References

Amornsakchai, S., et al. (2000): Pak Mun Dam, Mekong River Basin, Thailand, A WCD Case Study prepared as an input to the World Commission on Dams, Cape Town. (Available online at )

Avissar, R., and R.A. Pielke (1989): A parameterization of heterogenous land-surface for atmospheric numerical models and its impact on regional meteorology. Monthly Weather Review 117, 2113-2136.

Becker, A., M. Bonnell, R.A. Feddes, V. Krysanova, J.J. McDonnell, R.E. Schulze, and C. Valentin (2004): Responses of hydrological processes to environmental change at small catchment scales. In Vegetation, Water, Humans, and the Climate, edited by Kabat et al., Springer, 301-338.

Boupha, P. (1995): Shifting cultivation extermination program in Lao PDR. Proceedings of Tokyo Symposium on Sustainable Agriculture and Rural Development, Tokyo, November.

Bowling, L.C. and D.P. Lettenmaier, 2001. The effect of forest roads and harvest on catchment hydrology in a mountainous maritime environment, in Land Use and Watersheds: Human Influence on Hydrology and Geomorphology in Urban and Forest Areas, Water Science and Application, The American Geophysical Union, 2, 145-164.

Brown, S., and A.E. Lugo (1990): Tropical secondary forests. J. Tropical Ecology 6, 1-32.

Bruijnzeel, L.A. (2004): Tropical forests and environmental services: not seeing the soil for the trees? Agriculture, Ecosystems & Environment (in press).

Bruijnzeel, L.A. (2001): Forest hydrology. In The Forests Handbook. Evans, J.C. (Ed.). Blackwell Scientific: Oxford, U.K.; Chapter 12.

Bruijnzeel, L.A. (1989): (De)forestation and dry season flow in the tropics: a closer look. J. Tropical Forest Science 1, 229-243.

Chazee, L. (1994): Shifting cultivation systems in Laos: present systems and their future. Shifting Cultivation Systems and Rural Development in the Lao PDR, report of the Nabong Technical Meeting, Vientiane, 66-97. (Available from UNDP Resident Representative, PO Box 345, Vientiane, LAO P.D.R.)

Costa-Cabral, M.C., J.E. Richey, G. Goteti, D.P. Lettenmaier, S.D. Rodda, and C. Feldkötter (in review): Landscape structure and use, climate, and water movement in the Mekong river basin.

Dansie, J. (1994): Laos hydro to power Thai economy. International Water Power and Dam Construction, pp 18-20, March 1994.

FAO (2001): Global Forest Resources Assessment 2000: main report. FAO Forestry Paper No. 140. Food and Agricultural Organization of the United Nations, Rome. Available at:

FAO Cambodia Online (2000): Environment: Concepts and issues. At

Feder, G., T. Onchan, and Y. Chalamwong (1988): Land policies and farm performance in Thailand’s forest reserve areas. Economic Development and Cultural Change 36, 483-501.

Fox, J., D.M. Truong, A.T. Rambo, N.P. Tuyen, L.T. Cuc, and S. Leisz (2000): Shifting cultivation: A new old paradigm for managing tropical forests. Bio Science 50: 521-528.

Gartrell, A. (1997): Resource management in the Cambodian Mekong basin, Working Paper Series, Asia Research Centre, Murdoch University. (Available online at )

Giambelluca, T.W. (2002): Hydrology of altered tropical forest. Hydrological Processes 16, 1665-1669. doi: 10.1002/hyp.5021. Published online in Wiley InterScience ( Interscience.)

Giambelluca, T.W., A.D. Ziegler, M.A. Nullet, T.M. Dao (2001): Transpiration in a small forest patch. Section 6, Draft final report of project DEB-9613613. Influence of forest fragmentation on watershed functions in northern Vietnam. (accessed November 2004).

Giambelluca, T.W., M.A. Nullet, A.D. Ziegler, and L. Tran (2000): Latent and sensible energy flux over deforested land surfaces in the eastern Amazon and northern Thankland. Singapore Journal of Tropical Geography 21, 107-130.

Giambelluca, T.W., J. Fox, S. Yarnasarn, P. Onibutr, and M.A. Nullet (1999): Dry-season radiation balance of land covers replacing forest in northern Thailand. Agricultural and Forest Meteorology 95, 53-65.

Giambelluca, T.W.,D. Hölscher, T.X. Bastos, R.R. Frazão, M.A. Nullet, and A.D. Ziegler (1997): Observations of albedo and radiation balance over post-forest land surfaces in eastern Amazon Basin. J. Climate 10, 919-928.

Hansen, P.K. (1998): Shifting cultivation development in northern Laos. In E.C. Chapman, B. Bouahom, and P.K. Hansen (Eds.) Upland Farming Systems in Lao P.D.R. – Problems and Opportunities for Livestock. Australian Centre for International Agricultural Research, Canberra. ACIAR Proceedings No. 87.

Helms, .A. (1988): The Dictionary of Forestry. Society of American Foresters, Bethesda, MD, U.S.A.

Hinton, P. (1998): Resource management in the Yunnan Mekong basin, Working Paper Series, Asia Research Centre, Murdoch University. (Available online at )

Hölscher, D., T.D.A. Sá, T.X. Bastos, M. Denich, and H. Fölster (1997): Evaporation from young secondary vegetation in eastern Amazonia. J. Hydrology 193, 293-305.

Hori, H. (2000): The Mekong: Environment and Development. United Nations University Press, 398 pp.

Ikunaga, M. (1999): The forest issue in Cambodia – current situation and problems: An analysis based on field research. IDRI Occasional Paper No. 12. FASID (Foundation for Advanced Studies on International Development), Tokyo, Japan.

International Rivers Network (1999): Power struggle: The impacts of hydro-development in Laos. Berkeley, California, 68 pp.

Klaassen, W. (1992): Average fluxes from heterogeneous vegetated regions. Boundary-Layer Meteorology 58, 329-354.

Lamarche J. and D.P. Lettenmaier (2001), "Effects of Forest Roads on Flood Flows in the Deschutes River Basin, Washington," Earth Surface Processes and Landforms, 26(2): 115-134.

Laurance, W.F., and R.O. Bierregaard, Jr. (Eds) (1997): Tropical Forest Remnants: Ecology, Management, and Conservation of Fragmented Communities. University of Chicago Press, 1997.

Liang, X., D.P. Lettenmaier, E.F. Wood, and S.J. Burges (1994): A simple hydrologically based model of land surface, water, and energy fluxes for general circulation models. J. Geophysical Res. 99 (D7), 14,415-14,428.

Liang, X., D.P. Lettenmaier, and E.F. Wood (1996): One-dimensional statistical dynamic representation of subgrid spatial variability of precipitation in the two-layer variable infiltration capacity model. J. Geophys. Res. 101 (D16), 21,403-21,422.

Linsley, R. K., Jr., M.A. Kohler, and J.L.H. Paulhus (1982): Hydrology for Engineers (Third Edition). McGraw-Hill, Inc., 508 pp.

Mattheussen, B. R.L. Kirschbaum, I.A. Goodman, G.M. O’Donnell, and D.P. Lettenmaier (2000): Effects of land cover change on streamflow in the interior Columbia River Basin (USA and Canada), Hydrological Processes 14, pp. 867-885.

Mekong River Commission (2003): State of the Basin Report ........XXXXXXX

Miller, F., N.V. Thinh, D.T.M. Duc (1999): Resource management in the Vietnamese Mekong basin, Working Paper Series, Asia Research Centre, Murdoch University. (Available online at )

Non-Timber Forest Products Project (NTDP, 2000): A Study of the downstream impacts of the Yali Falls Dam in the Se San river basin in Ratnakiri Province, Northeast Cambodia. The Fisheries Office. (Available online at )

Ohte, N. (2001): Flux observation in GAME-T. In 2001 Workshop on GAME-T, Phuket, Thailand, March 2001.

Pednekar, S. (1997): Resource management in the Thai Mekong Basin, Working Paper Series, Asia Research Centre, Murdoch University. (Available online at )

Piccolo, A., and A.Z. Teshale (1998): Soil processes and responses to climate changes. In: Climate change impact on agriculture and forestry. European Commission, EUR 18175, Luxemburg, pp. 79-92.

Pravongviengham, P.P. (1998): The role of animal husbandry and aquaculture in improvements of swidden-based livelihood system in the Lao PDR. PhD Thesis, Asian Institute of Technology, Bangkok.

Rasul, G., and G.B. Thapa (2003): Shifting cultivation in the mountains of south and southeast Asia: Regional patterns and factors influencing the change. Land Degrad. Develop. 14, 495-508.

Rigg, J. (1995): In the fields there is dust, Geography 80 (1), pp. 23-32.

Robinson, M., J. Boardman, R. Evans, K. Heppell, J.C. Packman, and G.J.L. Leeks (2000): Land-use change. In Acreman MC (Ed.) The hydrology of the U.K. Routledge, London, pp 30-45.

Sandewall, M., B. Ohlsson, and S. Sawathvong (2001): Assessment of historical land-use changes for purposes of strategic planning: a case study in Laos. Ambio 30, 55-61.

Schulze, R.E. (2004): Richer basin responses to global change and anthropogenic impacts, in Vegetation, Water, Humans, and the Climate, Kabat et al. (editors), IGBP series, Springer, 566 pp.

Schulze, R.E. (2001): Hydrological responses at river basin scale. School of Bioresources Engineering and Environmental Hydrology, University of Natal, Pietermaritzburg, South Africa, ACRUcons Report 38, 105 pp.

Schulze, R.E., M.J.C. Horan, S.N. Shange, R.M. Ndlela, and L.A. Perks (1998): Hydrological impacts of land use practices in the Pongola-Bivane catchment – Phase 3. University of Natal, Pietermaritzburg, South Africa. ACRUcons Report 26, 58 p.

Shuttleworth, W.J. (1988): Evaporation from Amazonian rainforests. Proceedings of the Royal Society, London Series B 233, 321-346.

Spencer, J.E. (1966): Shifting Cultivation in Southeastern Asia. University of California Press, Berkeley, California.

Thapa, G.B. (1998): Issues in the conservation and management of forests in Laos: the case of Sangthong District. Singapore Journal of Tropical Geography 19, 51-70.

Whitmore, T.c. (1997): Tropical forest disturbance, disappearance. In W.F. Laurance and R.O. Bierregaard Jr. (Eds.) Tropical Forest Remnants: Ecology, Management and Conservation of Fragmented Communities. University of Chicago Press, Chicago, U.S.A.

Xu, J., J. Fox, L. Xing, N. Podger, S. Leisz, and A. Xihui (1999): Effects of swidden cultivation, population growth, and state policies on land cover in Yunnan, China. Mountain Research and Development 19: 123-132.

Ziegler, A.D., T.W. Giambelluca, M.A. Nullet, D. Plondke, T.T. Vana, J. Fox, T.M. Dao, T.D. Vien, and S. Evett (2001): Near-surface hydrologic response in a fragmented landscape. Section 7, Draft final report of project DEB-9613613. Influence of forest fragmentation on watershed functions in northern Vietnam. (accessed November 2004).

Ziegler, A.D., and T.W. Giambelluca (1997): Importance of rural roads as source areas for runoff in mountainous areas of northern Thailand. J. Hydrology 196: 204-229.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download