Basin Environments & Google Earth Tour

Basin Environments Image

Basin Environments

 The Basin Environment of the Mekong is made up of multiple levels of physical, chemical, and biological processes, operating on different time and space scales. To understand how the basin operates at a day-to-day level, and how that functioning might change under future conditions, it is useful to seperate out the key component of the basin.

We can look at the basin through a Google Earth tour of the Virtual Mekong Basin.

And then we can look in some detail at the individual components of the basin, and how they combine to form the overall "template" that the life of the basin plays out over.

As rerpesented schemtically in this figure, the basic structure of the basin is determined by geological processes happeing over very long time scales. Soils are formed on somewhat shorter time scales, but still o ntime scales of decades to centries. Landcover and landuse reflect how vegetation changes on seasonal to annual and decadel time scales. Of course, human infrastructure is superimposed on the "natural" world. Climae is much more instantaneous, oftern on a nhourly to even minutes time scales. As river basins don't necessarily reflect naitonal boundaries, political divisions must be superimposed.

In the following sections, more general descrptions of thse basic landscape elements are reviewed. Elsewhere, this information is used as the basis to develop comuter models of how the basin functions.


The Basin Structure - What are the Building Blocks of the Mekong?

   The Mekong River basin is the world’s eighth largest in discharge (ca. 475 km3 year1), 12th largest in length (ca. 4800 km), and 21st largest in drainage area (ca. 795 000 km2). The headwaters of the Mekong River (Figure 1) are in the Tibetan Highlands, at nearly 5000 m elevation. Fed by melting snow, the Mekong runs down the steep Tibetan slopes through a narrow gorge in the Yunnan province of China. The Mekong drainage area is partitioned as follows: 21% in China, 3% in Burma, 25% in Laos, 23% in Thailand, 20% in Cambodia, and 8% in Vietnam. The portion of the basin lying within China, Burma, and the northern part of Laos, consists of mountainous terrain between 400 and 5000 m elevation and is referred to as the Upper Mekong Basin (189 000 km2), while the remaining 606 000 km2 of its drainage area form the Lower Mekong Basin.

     The ‘Northern Highlands’ include the region from southern Yunnan province in China through Burma, Laos, and Northern Thailand, and eastward into the northern end of the Annamite Range in Vietnam. The ‘Eastern Highlands’ mostly consist of the western slope of the Annamite Range shared by Laos and Vietnam. Next to western Cambodia, which receives, on annual average, over 3000 mm of rainfall (according to our estimates later in this paper), the Northern and Eastern Highlands, with elevations of up to about 2800 m, are the wettest regions in the basin. The high mountains have deep-cut valleys and topsoil consisting of a thin deposit of sandstone and igneous rocks. In contrast, the Korat Plateau, comprising the north-eastern region of Thailand, is a dry region. Despite annual rainfall totals between 1000 and 1600 mm, the rainless season lasts from October to April and evapotranspiration is high. The Plateau is mainly drained by the Chi and Mun rivers and includes areas of sandy and some saline soils.

     The basic physical structure, or template, of the Mekong provides the building blocks upon which all contemporary dynamics occur, such as cliamte, water movement, fisheires production, and so forth. This structure is represented by the geoloogy and geomorphology of the basin. Here is a general overview of the long-term physical processes that formed the Mekong. Click on:




What are the Soils of the Mekong Basin?

      FOLLOWING TO BE RE DONE The soil parameters required by the VIC model for each soil layer are the saturated hydraulic conductivity (Ks), porosity (s), field capacity (c), wilting point (w), and parameter n in the Brooks–Corey equation for unsaturated conductivity (Brooks and Corey, 1966). The soil parameters were estimated based on US Department of Agriculture (USDA) soil texture classes, using the conversion values by Schaake (2000). The Brooks–Corey n was obtained from the b parameter in Schaake’s table, which represents the slope of the moisture retention curve, in log space, using the relation n D 3 C 2b (see, e.g. Rawls et al., 1993, table V.1Ð1). Soil texture classes were obtained from soil type as described later.

     In this study, the top layer depth was set to 10 cm, and the depths of the second and third layers were established by calibration to observed hydrographs. soil texture data (see later) are partitioned into above and below 10 cm depths, making this a logical choice for the upper layer. Model sensitivity to the upper layer depth was found to be much less than to the depth of the second and third layers. For soil type, we used primarily a map by the Mekong River Commission (MRC, 2003), covering the Lower Mekong Basin, which utilizes the Food and Agriculture Organization (FAO) 1988 classification. This map is in polygon format and was converted to gridded format at 5 arc-minutes resolution in order to match the model grid used. Where re-classification of the MRC data into the FAO-1979 classes (required to use the soilprogram, later) was not feasible, the FAO/UNESCO digital soil map of the world (FAO, 1995), was used. Soil texture (%clay and %sand) and bulk density (b)were derived from the soil type map using the World Inventory of Soil Emission (WISE) potentials pedon data base (Batjes, 1995) with the aid of the soilprogram (Carter and Scholes, 1999). The methods used to derive soil parameters from this data set are described in (Nijssen et al., 1997). From %clay and %sand, each of our 5 arc-minutes pixels was assigned one of the 12 USDA soil texture classes for the top soil layer (<10 cm) and bottom soil layers (>10 cm) (Figure 3).


Land Cover and Land Use

     While land use changes in the Mekong River basin clearly occurred during our simulation period 1979–2000, their quantification is difficult. No reliable estimates of deforestation rates are available for the basin. Different data sources, including satellite sources, are generally not mutually compatible, and sometimes disagree due to seemingly subtle differences in processing methods, and to sensor characteristics. The exception is the data set for 1993–1997 by the Forest Cover Monitoring Project (Stibig, 1999) which yielded a rate of net loss of forest of 0Ð53% per year over the basin for those 4 years. Relative to the entire 1979–2000 simulation period, this rate is probably high, but if we were to apply it, it would imply a net change in forest cover of about 11% from the beginning to the end of our reference period, and of about 5% on average between our calibration and evaluation periods.

     For estimation of land cover during our study period, we used the data set prepared by the MRC (2001) for the Lower Mekong Basin (for which 1997 is the base year) combined with GLC2000 products (European Commission, 2003) for the upper basin (for which 2000 is the base year). Figure 4 shows the land cover map used in our simulations. This map was obtained as a Landcovr Classescomposite of the four maps (a)–(d), described later; hence data resolution differs for the Upper Mekong Basin (30 arc-seconds, or roughly 1 km) and the Lower Mekong Basin (0Ð0012°, or about 250 m). In all cases, the spatial resolution of the land cover is much higher than the spatial resolution at which the VIC model was applied (ca. 10 km), hence the implications of the differences in resolution are modest, given that the VIC model requires only information about the fraction of each of its grid cells covered by a given vegetation type.

(a) For the Lower Mekong Basin, the 1997 land covermap, with a resolution of 0Ð0012°, produced bythe MRC (2001) in cooperation with the German Gesellschaft f¨ur Technische Zusammenarbeit (GTZ). The methods used to produce this map were described in Stibig (1999). This map is henceforth designated the ‘MRC/GTZ map’ and represents the year 1997. The methods used to produce this map combined field observations, aerial photographs, and multi-seasonal satellite images at a 1 : 250 000 scale (Stibig, 1999). In the legend of Figure 4, classes numbered 1–26, and 43, are taken from the legend of the MRC/GTZ map.

(b) For the irrigated areas in the Lower Mekong Basin, the ‘irrigated lands 2001’ data of the Atlas by the MRC and WWF (2003). Class 42 in the legend of Figure 4 represents this map.

(c) For the Chinese portion of the Upper Mekong Basin, the 20 arc-second resolution land cover map of the Global Land Cover 2000 (GLC2000) map for China (European Commission, 2003), which was produced from satellite data collected from January to December of 2000, by SPOT Vegetation S10. The GLC2000 maps use the United Nations Food and Agriculture Organization Land Cover Classification System (FAO-LCCS) ( 2000/legend.htm). Classes numbered 27–41 in Figure 4 are taken from this map.

(d) For the Burmese portion, the 30 arc-second resolution land cover map of the GLC2000 map for southeast Asia (European Commission, 2003), which was produced from the satellite SPOT Vegetation S10 data using dry-season images (January to March) for 1999/2000. Similarly to (c), it uses the FAO-LCCS. This map was chosen over the MODIS 2000 satellite derived map because, judged visually, it is in best agreement with the MRC/GTZ map over the Lower Mekong Basin. Only two land cover classes are represented in the Burmese portion of the Mekong: the class designated ‘tree cover, broadleaved, evergreen, closed and closed to open’ was assigned to MRC/GTZclass 1, ‘Evergreen forest, high cover density’; and the class designated ‘cultivated and managed, nonirrigated (mixed)’ was assigned to MRC/GTZ class 23, ‘agricultural land’.

Vegetation parameters Leaf area index.
     The leaf area index (LAI) is a non-dimensional variable that represents the average (projected) leaf surface area covering each unit ground area. LAI is one of the vegetation parameters to which VIC is most sensitive. It controls not only precipitation (solid and liquid) interception, but also canopy resistance to transpiration, and the attenuation of solar radiation through the vegetation cover. The standard 8-day, 1 km MODIS satellite LAI product was used (, monthly averages having been taken over the period 1 March 2000 through 28 February 2001. The 12 monthly mean LAI values thus generated for each model grid cell are shown in Figure 5. The spatial patterns of LAI in Figure 5 show good correspondence with the land cover map in Figure 4, with forest having the highest LAI values, and agricultural land having the lowest. The regions with highest LAI are Laos and Vietnam (with the exception of the Mekong Delta), and some regions of Cambodia far from the Tonle Sap Lake. The Thai portion of the basin has a LAI mostly below 2Ð0 throughout the year, due to the strong dominance of agriculture there. The high-elevation areas of China’s Yunnan province also have a LAI mostly under 2Ð0 year-round, which is indicative of the sparse vegetation in the upper basin. It was determined that there was a major problem with the MODIS values during the May through October rainy season, probably due to difficulties in obtaining valid satellite readings in the cloudy conditions of this season. Therefore, we decided to use the MODIS LAI values for November 2001 to represent the months from June through October, and the MODIS LAI values for April 2000 to represent the month of May.


     Albedo has large impact on the VIC computed evaporation from the canopy, and plant transpiration. We based our assignment of seasonal albedo on previous relevant observations elsewhere, mainly from Giambelluca et al. (1999), and attempted to correctly represent the albedo values of the vegetation classes relative to each other. In tropical regions, albedo is generally observed to be highest during the dry season because of reduced LAI (Pinker et al., 1980; Barradas and Ad´1999),em, 1982; cited by Giambelluca et al., with seasonal variations due to changes in sun angle being relatively unimportant. Annual minimum and maximum albedo values were established for each of the land cover classes in Figure 4, corresponding to the dry season (November–May) and the wet season (June–October), respectively (see Table I). The values used ranged from a low of 0Ð085 for irrigated cropland (class 41) to 0Ð6–0Ð8 for old snow and 0Ð8–0Ð9 for new snow (e.g. Gray and Prowse, 1993).

Vegetation height, displacement height, roughness length, and architectural resistance.
     Vegetation height is used by VIC as the basis for determining roughness length and displacement height, both of which are important parameters in its evapotranspiration formulation.We were unable to identify studies reporting vegetation height for different land cover classes in Mekong regions, except for the approximate values mentioned in Stibig (1999) for classes 7, 9, 10, 16, 19 and 20. Values from Sellers et al. (1986) were therefore used for classes 1, 2, 27 and 28. For all other classes, values judged reasonable (from field observations by some of the authors) were utilized (Table I), but which have not been validated. Roughness length is defined as the height above the ground where wind speed is reduced to zero due to surface resistance. This parameter is used to determine the wind profile in VIC using a logarithmic approximation.

     Displacement height is defined as the height above the ground where wind speed is not significantly affected by surface roughness. Roughness length and displacement height were estimated by multiplying the vegetation height in Table I by the factors 0Ð123 and 0Ð67, respectively [following Brutsaert (1975) cited in Shuttleworth (1993, p. 4Ð12)]. The values used for architectural resistance (Rarc, Table I) are based on Ducoudre´et al. (1993).

Minimum stomatal resistance, RGL, and solar radiation attenuation.
     The minimum stomatal resistance (Rmin) is defined as the stomatal resistance that occurs with full sunlight and at saturation leaf water potential. VIC uses Rmin, together with LAI and current soil moisture, to calculate the canopy resistance to transpiration—based on the formulations of Blondin (1991) and Ducoudre´et al. (1993), as described in Liang et al. (1994). The values of Rmin used in our simulations are listed in Table I. RGL is the minimum incoming shortwave radiation at which there will be transpiration. For forest cover classes we used the value 30 W m2,and for agriculture we used 100 W m2. Intermediate values were used for other land cover classes, as specified in Table I. Solar radiation attenuation was approximated at 50%.

Wind and wind attenuation.
     The VIC parameter wind height is the height above ground at which wind observations were recorded. VIC expects this measurement to have been made at high enough elevation that the vegetation effects on wind speed are negligible. The model then estimates wind speed through and below the canopy using logarithmic wind profiles. As in Maurer et al. (2002) and Nijssen et al. (2001b), we used wind speed data from the NCEP-NCAR Reanalysis (Kalnay et al., 1996) and assumed the wind measurement height to be equal to 2 m above the vegetation height in the case of short vegetation classes (of grassland, shrubland, and agricultural land), and equal to 10 m above the vegetation height in the case of tall vegetation classes (forest classes). Wind speed attenuation through the overstory to a 2 m height was approximated at 50% when a tree canopy is present and at 10% when it is not.

Maximum rooting depth, and distribution of root mass with depth.
     Maximum rooting depth affects the ability of the vegetation to extract moisture from the three soil layers, and hence affects evapotranspiration. The values of maximum rooting depth used in our simulations are listed in Table I. The bulk of root mass was allocated to the top two soil layers.