Academia.eduAcademia.edu

Figure 1 – uploaded by Anthony Kiem

See full PDFdownloadDownload figure
Figure 1 Schematic diagram of the S—W model. From right to left, rg and r§ bulk resistances of canopy stomatal and boundary layer respectively, ri and r? aerodynamic resistances from soil to canopy and from canopy to reference height respectively, r: soil surface resistance, Zo, ground roughness length, Zo + do effective height of canopy source, za reference height (=h, +2 m), h, vegetation height, R, and R* net radiations above canopy and to soil surface respectively, G soil heat flux, ET, transpi- ration from canopy, ET, evaporation from soil, ET total evapotranspiration, H sensible heat, Twin and Tmax daily min and max air temperatures respectively, e, actual vapour pressure, C cloud cover, uy wind speed at z, height, hy, observation height for other meteorological variables (usually, w=10m and h,=2m), Zow ground roughness length at weather station.

Figure 1 Schematic diagram of the S—W model. From right to left, rg and r§ bulk resistances of canopy stomatal and boundary layer respectively, ri and r? aerodynamic resistances from soil to canopy and from canopy to reference height respectively, r: soil surface resistance, Zo, ground roughness length, Zo + do effective height of canopy source, za reference height (=h, +2 m), h, vegetation height, R, and R* net radiations above canopy and to soil surface respectively, G soil heat flux, ET, transpi- ration from canopy, ET, evaporation from soil, ET total evapotranspiration, H sensible heat, Twin and Tmax daily min and max air temperatures respectively, e, actual vapour pressure, C cloud cover, uy wind speed at z, height, hy, observation height for other meteorological variables (usually, w=10m and h,=2m), Zow ground roughness length at weather station.

Related Figures (27)

where h, is the vegetation height (m), n is the eddy diffusiv- ity decay constant of the vegetation, K;, is the eddy diffusion coefficient at the top of canopy (m*s~'), Zog is the rough- ness length of ground (m), varying with the vegetation type, Zo is the ‘'preferred’’ roughness length (=0.13h-.) (m), d, is the ‘‘preferred’’ zero plane displacement (=0.63h,) (m), « is von Karman’s constant (x = 0.41), u. is the friction veloc- ity (ms~'), z, is the reference height (m), 2m above vege- tation, do is the zero plane displacement of canopy (m). All items are parameterized as follows (Monteith, 1973; Cho- udhury and Monteith, 1988; Shuttleworth and Gurney, 1990; Federer et al., 1996): where ug is the wind speed at the reference height (ms~'), Zp is the roughness length of canopy while Zo, is that for a ‘*closed’’ canopy (m), cg is the mean drag coefficient for individual leaves. Eqs. (22) and (23) combine two cases: closed canopy (LAI > 4) and sparse growing vegetation, using the ‘‘preferred’’ values or related to LAI in the sec- ond-order closure theory. Eq. (24) uses the ‘‘preferred’’ values but differentiates the canopy from short and tall veg- etation types and linear interpolated between. Eq. (25) uses the value of open water surface for bare surface (Brutsaert, 1982, pp. 118). Wilson and Shaw (1977) assumed cg as 0.2 for maize leaves, Jones (1992) gave a typical value in the range between 0.03 and 0.6, whereas Shuttleworth and Gur- ney (1990) set it to 0.07. Eq. (25) gives 0.05 around for short vegetation (h.<1m), but 0.412 for tall vegetation (he > 10m). Eq. (26) uses the typical values to represent for short vegetation and tall vegetation, and linear interpo- lation between: 2.5 of agricultural crop (Monteith, 1973; Uchijima, 1976) and 4.25 of pine forest (Brutsaert, 1982, pp. 106). el ee ee EB cece Be ae ee ee a a
where h, is the vegetation height (m), n is the eddy diffusiv- ity decay constant of the vegetation, K;, is the eddy diffusion coefficient at the top of canopy (m*s~'), Zog is the rough- ness length of ground (m), varying with the vegetation type, Zo is the ‘'preferred’’ roughness length (=0.13h-.) (m), d, is the ‘‘preferred’’ zero plane displacement (=0.63h,) (m), « is von Karman’s constant (x = 0.41), u. is the friction veloc- ity (ms~'), z, is the reference height (m), 2m above vege- tation, do is the zero plane displacement of canopy (m). All items are parameterized as follows (Monteith, 1973; Cho- udhury and Monteith, 1988; Shuttleworth and Gurney, 1990; Federer et al., 1996): where ug is the wind speed at the reference height (ms~'), Zp is the roughness length of canopy while Zo, is that for a ‘*closed’’ canopy (m), cg is the mean drag coefficient for individual leaves. Eqs. (22) and (23) combine two cases: closed canopy (LAI > 4) and sparse growing vegetation, using the ‘‘preferred’’ values or related to LAI in the sec- ond-order closure theory. Eq. (24) uses the ‘‘preferred’’ values but differentiates the canopy from short and tall veg- etation types and linear interpolated between. Eq. (25) uses the value of open water surface for bare surface (Brutsaert, 1982, pp. 118). Wilson and Shaw (1977) assumed cg as 0.2 for maize leaves, Jones (1992) gave a typical value in the range between 0.03 and 0.6, whereas Shuttleworth and Gur- ney (1990) set it to 0.07. Eq. (25) gives 0.05 around for short vegetation (h.<1m), but 0.412 for tall vegetation (he > 10m). Eq. (26) uses the typical values to represent for short vegetation and tall vegetation, and linear interpo- lation between: 2.5 of agricultural crop (Monteith, 1973; Uchijima, 1976) and 4.25 of pine forest (Brutsaert, 1982, pp. 106). el ee ee EB cece Be ae ee ee a a
For the environmental stress functions, CO2 concentra- tion is ignored and others are outlined as follows: As shown in Eq. (30), the hyperbolic solar radiation func- tion is used (Jarvis, 1976; Stewart, 1988; Stewart and Gay, 1989), where S is the incoming photo-synthetically active radiation (PAR) (in the range of 0.4—0.72 um) flux (W m2), d=1+c/1000, c=100 for forests and 400 for crops. For simplicity, the net radiation rather than the solar irradiation is used (Monteith, 1995) and averaged for the day time in daily simulation. The coefficient of vapour pressure deficit function, Eq. (31), is represented for short vegeta- tion obtained from the Konza Prairie in Kansas (FIFE data) Stewart and Gay, 1989), and for tall vegetation by the coniferous forest from the Hydrologic Atmospheric Pilot Experiment/Modelisation du Bilan Hydrique (HAPEX-MOBIL- HY) data (Noilhan and Planton, 1989; also see Lhomme et al., 1998), where D = e, — e,. When the air temperature, T, is higher than 25 °C, the stomatal openness is not limited; ess than 0 °C, stomatal closes completely; and varies line- arly between, as shown in Eq. (32) where T in K. The stress function of soil moisture content is shown in Eq. (33), where 0 is the soil moisture content in root zone, 0; is the field capacity below which the transpiration is stressed, 6, is the residual soil moisture content. For PET, the soil mois- ture is assumed at field capacity.
For the environmental stress functions, CO2 concentra- tion is ignored and others are outlined as follows: As shown in Eq. (30), the hyperbolic solar radiation func- tion is used (Jarvis, 1976; Stewart, 1988; Stewart and Gay, 1989), where S is the incoming photo-synthetically active radiation (PAR) (in the range of 0.4—0.72 um) flux (W m2), d=1+c/1000, c=100 for forests and 400 for crops. For simplicity, the net radiation rather than the solar irradiation is used (Monteith, 1995) and averaged for the day time in daily simulation. The coefficient of vapour pressure deficit function, Eq. (31), is represented for short vegeta- tion obtained from the Konza Prairie in Kansas (FIFE data) Stewart and Gay, 1989), and for tall vegetation by the coniferous forest from the Hydrologic Atmospheric Pilot Experiment/Modelisation du Bilan Hydrique (HAPEX-MOBIL- HY) data (Noilhan and Planton, 1989; also see Lhomme et al., 1998), where D = e, — e,. When the air temperature, T, is higher than 25 °C, the stomatal openness is not limited; ess than 0 °C, stomatal closes completely; and varies line- arly between, as shown in Eq. (32) where T in K. The stress function of soil moisture content is shown in Eq. (33), where 0 is the soil moisture content in root zone, 0; is the field capacity below which the transpiration is stressed, 6, is the residual soil moisture content. For PET, the soil mois- ture is assumed at field capacity.
where Reotar; Ro var and R,, are the solar, clear-sky solar and net solar radiations respectively, R,, is the net loss of energy in longwave radiation to the atmosphere, « is the albedo of land surface, o is the Stefan—Boltzmann constant (=4.903 x 10°? MJ K 4m? day~'), Tmax,k aNd Tmin,x are the max and min atmospheric temperatures in Kelvin, Tmax,k = Tmax + 273.16 and Tinin,k = Tmin + 273.16. All radia- tion terms are in MJ m-? day~'. Reotar depends on the Julian date, the latitude location and the cloud cover condition (see Appendix A). In Eq. (40), the first term is based on the Stefan—Boltzmann law, second term is the net emissiv- ity between the atmosphere and the land surface, and third term is the correctness factor for the cloud cover (Shuttle- worth, 1993; Allen et al., 1998). The land surface albedo is related to vegetation LAI (Uchijima, 1976) as:
where Reotar; Ro var and R,, are the solar, clear-sky solar and net solar radiations respectively, R,, is the net loss of energy in longwave radiation to the atmosphere, « is the albedo of land surface, o is the Stefan—Boltzmann constant (=4.903 x 10°? MJ K 4m? day~'), Tmax,k aNd Tmin,x are the max and min atmospheric temperatures in Kelvin, Tmax,k = Tmax + 273.16 and Tinin,k = Tmin + 273.16. All radia- tion terms are in MJ m-? day~'. Reotar depends on the Julian date, the latitude location and the cloud cover condition (see Appendix A). In Eq. (40), the first term is based on the Stefan—Boltzmann law, second term is the net emissiv- ity between the atmosphere and the land surface, and third term is the correctness factor for the cloud cover (Shuttle- worth, 1993; Allen et al., 1998). The land surface albedo is related to vegetation LAI (Uchijima, 1976) as:
where SR is the simple ratio of hemispheric reflectance for the NIR (near-infrared) light to that for the visible light, FPAR is the fraction of photo-synthetically active radiation, Fy is the fraction of clumped vegetation, SRmin and SRmax are SR with 5% and 98% of NDVI population. The values of Fa, NDVI at 5% and 98% population are adopted from SiB2 for all vegetation types (NDVI at 5% setting to 0.039 glob- ally, F., and NDVI at 98%, see Table 1). FPARmin = 0.001 and FPARmax = 0.950 consider the satellite-sensed NDVI sat- uration (Sellers et al., 1996). LAl a, is the maximum LAI when the vegetation develops fully, prescribed for each vegetation class by referring to the literature. Leaf area index (LAI). The LAI is used not only intensively in S—W parameterization but also in interception estimation. The LAI for each vegetation class can be derived from NOAA-AVHRR NDVI through FPAR (Myneni and Williams, 1994; Sellers et al., 1994; Andersen et al., 2002). Here the SiB2 method is used (Sellers et al., 1996):
where SR is the simple ratio of hemispheric reflectance for the NIR (near-infrared) light to that for the visible light, FPAR is the fraction of photo-synthetically active radiation, Fy is the fraction of clumped vegetation, SRmin and SRmax are SR with 5% and 98% of NDVI population. The values of Fa, NDVI at 5% and 98% population are adopted from SiB2 for all vegetation types (NDVI at 5% setting to 0.039 glob- ally, F., and NDVI at 98%, see Table 1). FPARmin = 0.001 and FPARmax = 0.950 consider the satellite-sensed NDVI sat- uration (Sellers et al., 1996). LAl a, is the maximum LAI when the vegetation develops fully, prescribed for each vegetation class by referring to the literature. Leaf area index (LAI). The LAI is used not only intensively in S—W parameterization but also in interception estimation. The LAI for each vegetation class can be derived from NOAA-AVHRR NDVI through FPAR (Myneni and Williams, 1994; Sellers et al., 1994; Andersen et al., 2002). Here the SiB2 method is used (Sellers et al., 1996):
Figure 2 Soil water balance and runoff generation in BTOPMC, where P gross rainfall, ETo interception evaporation, Imax interception storage capacity, / interception state, P, net rainfall on land surface, ET actual evapotranspiration, S,max available water storage capacity of root zone, S,z soil moisture state in root zone, SD and S,, soil moisture storage deficit and state in unsaturated zone, qo overland runoff, q, groundwater recharge, and q, groundwater release.
Figure 2 Soil water balance and runoff generation in BTOPMC, where P gross rainfall, ETo interception evaporation, Imax interception storage capacity, / interception state, P, net rainfall on land surface, ET actual evapotranspiration, S,max available water storage capacity of root zone, S,z soil moisture state in root zone, SD and S,, soil moisture storage deficit and state in unsaturated zone, qo overland runoff, q, groundwater recharge, and q, groundwater release.
where EX(t) is the storage excess in root zone (mm day‘) Saturation zone Soil moisture content in this zone does not change. How- ever, it receives the recharge from unsaturated zone at q(t) (mmday~') and releases a base flow at gq,(t) (mm day”):
where EX(t) is the storage excess in root zone (mm day‘) Saturation zone Soil moisture content in this zone does not change. How- ever, it receives the recharge from unsaturated zone at q(t) (mmday~') and releases a base flow at gq,(t) (mm day”):
The storage state of interception, I(t) (mm), is updated and the net rainfall on the land surface, P,(t) (mm day~'), is calculated as follows: where /(t — 1) is storage state of interception at previous time step (mm). Eq. (62) means when the gross rainfall is (i) larger than the interception deficit, a part of rainfall is captured by the canopy and saturated (reaching Imax); (ii) less than the deficit, all rainfall is intercepted. The initial state of interception storage can be set to /(0) = 0.
The storage state of interception, I(t) (mm), is updated and the net rainfall on the land surface, P,(t) (mm day~'), is calculated as follows: where /(t — 1) is storage state of interception at previous time step (mm). Eq. (62) means when the gross rainfall is (i) larger than the interception deficit, a part of rainfall is captured by the canopy and saturated (reaching Imax); (ii) less than the deficit, all rainfall is intercepted. The initial state of interception storage can be set to /(0) = 0.
Table 2 Parameters to be calibrated in the BTOPMC model where oer is an empirical constant and Vorosmarty et al. (1998) set to 5.0 so that the drying curve would resemble that of Pierce (1958) with the meadow crops when 8,2 = ET/PET is plotted as a function of S,,/S, max during peri- ods of no precipitation. Because the knowledge is very lim- ited on the drying function, here we obtain it by calibration and apply it uniformly over the whole basin.
Table 2 Parameters to be calibrated in the BTOPMC model where oer is an empirical constant and Vorosmarty et al. (1998) set to 5.0 so that the drying curve would resemble that of Pierce (1958) with the meadow crops when 8,2 = ET/PET is plotted as a function of S,,/S, max during peri- ods of no precipitation. Because the knowledge is very lim- ited on the drying function, here we obtain it by calibration and apply it uniformly over the whole basin.
Figure 3. Mekong River basin. The solid points indicated by the numbers are pan observation sites and stars hydrological stations. The pan observation sites are located in the Lower Mekong basin, most in Thailand.
Figure 3. Mekong River basin. The solid points indicated by the numbers are pan observation sites and stars hydrological stations. The pan observation sites are located in the Lower Mekong basin, most in Thailand.
Figure 5 Yearly change of (a) daily mean temperature and (b) diurnal temperature range over the Mekong River basin. Since the late 1970s, the temperature has an obvious increasing trend. The data of diurnal temperature range are lack before 1915 and three different periods can be identified: before 1930, between 1930 and the mid 1950s, and after the mid 1950s.
Figure 5 Yearly change of (a) daily mean temperature and (b) diurnal temperature range over the Mekong River basin. Since the late 1970s, the temperature has an obvious increasing trend. The data of diurnal temperature range are lack before 1915 and three different periods can be identified: before 1930, between 1930 and the mid 1950s, and after the mid 1950s.
Figure 6 Yearly change of (a) cloud cover and (b) actual vapour pressure over the Mekong River basin. As a secondary variable, the change trend of cloud cover follows the diurnal temperature range (Fig. 5(b)) inversely and of the actual vapour pressure follows the daily minimum temperature (not shown) to a high extend.
Figure 6 Yearly change of (a) cloud cover and (b) actual vapour pressure over the Mekong River basin. As a secondary variable, the change trend of cloud cover follows the diurnal temperature range (Fig. 5(b)) inversely and of the actual vapour pressure follows the daily minimum temperature (not shown) to a high extend.
rectly from station observations. The secondary variables include cloud cover and vapour pressure and were con- structed by merging the station observation where available with the synthetic data derived from the gridded primary variables. For the synthetic data, the cloud cover was re- lated to the diurnal temperature range using an empirical equation and the actual vapour pressure to the daily mini- mum temperature using a conceptual equation (New et al., 2000). The CRU data sets are extracted at the Mekong River basin and transferred into the Lambert Azimuthal Equal Area projection at 8 km resolution without interpola- tion involvement. The monthly climatic data are applied to both the 10-day and month intervals.
rectly from station observations. The secondary variables include cloud cover and vapour pressure and were con- structed by merging the station observation where available with the synthetic data derived from the gridded primary variables. For the synthetic data, the cloud cover was re- lated to the diurnal temperature range using an empirical equation and the actual vapour pressure to the daily mini- mum temperature using a conceptual equation (New et al., 2000). The CRU data sets are extracted at the Mekong River basin and transferred into the Lambert Azimuthal Equal Area projection at 8 km resolution without interpola- tion involvement. The monthly climatic data are applied to both the 10-day and month intervals.
Figure 7 Climate changes in year round over the Mekong River basin (a) precipitation (over upper basin of Pakse from 1972 to 2000), (b) cloud cover, (c) mean temperature, (d) actual vapour pressure, (e) diurnal temperature range and (f) wind speed (from 1961 to 1990). study, daily precipitation at 64 gauged stations throughout the basin was obtained from 1972 to 2000 from the Mekong River Commission, the China Meteorological Administration, and the Institute of Atmospheric Physics at the Chinese Academy of Sciences. The scatter of these rain gauges over the basin will be showed in ‘*Hydrological modeling with S— W estimated PET, and PET as input’’ section. The yearly and seasonly change of spatially averaged precipitation is first presented here. The CRU (Climate Research Unit, University of East Anglia in UK) TS 2.0 provides with the time series data of the monthly mean temperature, diurnal tempera- ture range, cloud cover and actual vapour pressure from 1901 to 2000 and mean monthly wind speed averaged in 1961—90 for the globe at 0.5x0.5° grids (New et al., 1999, 2000). The wind speed was measured at a majority of 10m height (New et al., 1999). In construction of CRU TS 2.0, climatic variables are separated into two categories: primary and secondary. Regarding to the data used in this study, the primary variables include mean temperature and diurnal temperature range and were constructed di- Precipitation. The distributed daily precipitation is ob- tained by applying the Thiessen polygon method to the 64 rain gauges. Because very few are located downstream of Pakse, the spatial average precipitation is made only over upper basin of Pakse, as shown in Fig. 4. It changes much from year to year, with the average of 1322 mm/year, close
Figure 7 Climate changes in year round over the Mekong River basin (a) precipitation (over upper basin of Pakse from 1972 to 2000), (b) cloud cover, (c) mean temperature, (d) actual vapour pressure, (e) diurnal temperature range and (f) wind speed (from 1961 to 1990). study, daily precipitation at 64 gauged stations throughout the basin was obtained from 1972 to 2000 from the Mekong River Commission, the China Meteorological Administration, and the Institute of Atmospheric Physics at the Chinese Academy of Sciences. The scatter of these rain gauges over the basin will be showed in ‘*Hydrological modeling with S— W estimated PET, and PET as input’’ section. The yearly and seasonly change of spatially averaged precipitation is first presented here. The CRU (Climate Research Unit, University of East Anglia in UK) TS 2.0 provides with the time series data of the monthly mean temperature, diurnal tempera- ture range, cloud cover and actual vapour pressure from 1901 to 2000 and mean monthly wind speed averaged in 1961—90 for the globe at 0.5x0.5° grids (New et al., 1999, 2000). The wind speed was measured at a majority of 10m height (New et al., 1999). In construction of CRU TS 2.0, climatic variables are separated into two categories: primary and secondary. Regarding to the data used in this study, the primary variables include mean temperature and diurnal temperature range and were constructed di- Precipitation. The distributed daily precipitation is ob- tained by applying the Thiessen polygon method to the 64 rain gauges. Because very few are located downstream of Pakse, the spatial average precipitation is made only over upper basin of Pakse, as shown in Fig. 4. It changes much from year to year, with the average of 1322 mm/year, close
Figure 8 Spatial distribution of the input data over the Mekong River basin (a) Hydro1K DEM, (b) IGBP land cover, (c) FAO soil texture (transferred into USDA classification), (d) NOAA-AVHRR NDVI (June, 2000), (e) gauge precipitation (June, 2000), (f) CRU mean air temperature (June, 2000), (g) CRU diurnal temperature range (June, 2000), (h) CRU cloud cover (June, 2000), (i) CRU actual vapour pressure (June, 2000), (j) CRU wind speed (June, 1961—90), (k) calculated extraterrestrial radiation (June 15), (l) calculated maximum daylight duration (June 15). Cloud cover and actual vapour pressure. As secondary vari- ables, the change trends of cloud cover and vapour pressure All climatic variables change significantly in the year, as shown in Fig. 7, averaged in 1901—2000, or in 1972—2000 (precipitation), or in 1961—90 (wind). Most precipitation oc- curs in May to October (Fig. 7 (a)). Accordingly the cloud cover and vapour pressure are high (Fig. 7 (b) and (d)), and the diurnal temperature range is low (Fig. 7 (e)) during this time. The wind is also weak in this time but strong dur- ing the dry northeast monsoon from November to April (Fig. 7 (f)). The mean temperature varies from 17.2 °C (Jan- uary) to 24.1 °C (June) with an average of 21.5 °C in the
Figure 8 Spatial distribution of the input data over the Mekong River basin (a) Hydro1K DEM, (b) IGBP land cover, (c) FAO soil texture (transferred into USDA classification), (d) NOAA-AVHRR NDVI (June, 2000), (e) gauge precipitation (June, 2000), (f) CRU mean air temperature (June, 2000), (g) CRU diurnal temperature range (June, 2000), (h) CRU cloud cover (June, 2000), (i) CRU actual vapour pressure (June, 2000), (j) CRU wind speed (June, 1961—90), (k) calculated extraterrestrial radiation (June 15), (l) calculated maximum daylight duration (June 15). Cloud cover and actual vapour pressure. As secondary vari- ables, the change trends of cloud cover and vapour pressure All climatic variables change significantly in the year, as shown in Fig. 7, averaged in 1901—2000, or in 1972—2000 (precipitation), or in 1961—90 (wind). Most precipitation oc- curs in May to October (Fig. 7 (a)). Accordingly the cloud cover and vapour pressure are high (Fig. 7 (b) and (d)), and the diurnal temperature range is low (Fig. 7 (e)) during this time. The wind is also weak in this time but strong dur- ing the dry northeast monsoon from November to April (Fig. 7 (f)). The mean temperature varies from 17.2 °C (Jan- uary) to 24.1 °C (June) with an average of 21.5 °C in the
whole year (Fig. 7(c)). Fig. 8 shows all the input data distri- bution over the Mekong River basin except Fig. 8(e) which shows only over upper basin of Pakse because of few gauges downstream. same level (Table 4) and the changes coincide (Fig. 9) over the whole basin and among these points, their magnitudes and variation amplitudes are quite different. The PET at the grassland point in Tibet is lowest and yearly changes gently due to the cold weather and short vegetation; at the forest point in middle stream, it is highest and yearly changes significantly though located in the north (lower so- lar radiation) but with taller vegetation than the cropland point in lower stream. Fig. 10 shows the basin-spatial distri- bution of annual PET averaged in 1901—2000. The least is lo- cated in the most north in Tibet with the open shrubland, only about 300 mm/year. The largest is in the most south of Laos with the evergreen broadleaf forest, about 2040 mm/year. At a nearby site in the O Thom | watershed in Kompong Thom province of Cambodia, the PET of ever- green broadleaf forest is estimated 4.3 mm/day in October, 7.3 mm/day in March and 5.3 mm/day in the year, averaged from 1901 to 2000, comparable to an independent field measurement made by Shimizu et al. (2004) at the same site that the maximum evapotranspiration levels were 5.5 mm/ day during 20—28 October, 2003 (late rainy season) and >6.0 mm/day during 1—9 March, 2004 (middle dry season).
whole year (Fig. 7(c)). Fig. 8 shows all the input data distri- bution over the Mekong River basin except Fig. 8(e) which shows only over upper basin of Pakse because of few gauges downstream. same level (Table 4) and the changes coincide (Fig. 9) over the whole basin and among these points, their magnitudes and variation amplitudes are quite different. The PET at the grassland point in Tibet is lowest and yearly changes gently due to the cold weather and short vegetation; at the forest point in middle stream, it is highest and yearly changes significantly though located in the north (lower so- lar radiation) but with taller vegetation than the cropland point in lower stream. Fig. 10 shows the basin-spatial distri- bution of annual PET averaged in 1901—2000. The least is lo- cated in the most north in Tibet with the open shrubland, only about 300 mm/year. The largest is in the most south of Laos with the evergreen broadleaf forest, about 2040 mm/year. At a nearby site in the O Thom | watershed in Kompong Thom province of Cambodia, the PET of ever- green broadleaf forest is estimated 4.3 mm/day in October, 7.3 mm/day in March and 5.3 mm/day in the year, averaged from 1901 to 2000, comparable to an independent field measurement made by Shimizu et al. (2004) at the same site that the maximum evapotranspiration levels were 5.5 mm/ day during 20—28 October, 2003 (late rainy season) and >6.0 mm/day during 1—9 March, 2004 (middle dry season).
* The extreme values of minimum and maximum occur in the year indicated in the parenthesis, and the percentage listed with the standard deviation (STDEV) is the ratio of its absolute value to the mean annual PET. able 4 Characteristic values of annual PET and LAI over the Mekong River basin and at some specific points
* The extreme values of minimum and maximum occur in the year indicated in the parenthesis, and the percentage listed with the standard deviation (STDEV) is the ratio of its absolute value to the mean annual PET. able 4 Characteristic values of annual PET and LAI over the Mekong River basin and at some specific points
Figure 10 Spatial distribution of annual potential evapo- transpiration averaged in 1901—2000, where monthly compos- ite NDVI is used.
Figure 10 Spatial distribution of annual potential evapo- transpiration averaged in 1901—2000, where monthly compos- ite NDVI is used.
Figure 9 Yearly change of annual potential evapotranspira- tion over the whole basin and at three points: 1 in Tibet, 2 in north Laos and 3 in east Thailand, where monthly composite NDVI is used.
Figure 9 Yearly change of annual potential evapotranspira- tion over the whole basin and at three points: 1 in Tibet, 2 in north Laos and 3 in east Thailand, where monthly composite NDVI is used.
Figure 12 Comparison of PET due to the albedo change of substrate soil surface at Points 1, 2 and 3. Figure 11 Seasonal change of (a) PET (monthly), (b) LAI (monthly), (c) PET (10-day) and (d) LAI (10-day) over the whole basin and at two specific points averaged in 1901—2000 for PET and in 1981—2001 for the LAI. Both Points 4 and 5 are located in south Laos.
Figure 12 Comparison of PET due to the albedo change of substrate soil surface at Points 1, 2 and 3. Figure 11 Seasonal change of (a) PET (monthly), (b) LAI (monthly), (c) PET (10-day) and (d) LAI (10-day) over the whole basin and at two specific points averaged in 1901—2000 for PET and in 1981—2001 for the LAI. Both Points 4 and 5 are located in south Laos.
Table 5 Parameter values calibrated using the flow from 1983 to 1987 at Vientiane, Mukdahan, Yasothon and Pakse
Table 5 Parameter values calibrated using the flow from 1983 to 1987 at Vientiane, Mukdahan, Yasothon and Pakse
Table 6 Calibration and validation of BTOPMC over Mekong River basin (%) ns is Nash—Sutcliffe efficiency, Vol is volume ratio of Q, to Qo, Q, and Qp are simulated and observed discharge respectively at the station, pis precipitation, qoft is total runoff, gq, is recharge to groundwater, qp is groundwater release, ET is actual evapotranspiration from root zone, ETg is evaporation from interception. Except Q, and Q,, all items are averaged over upper basin of the station.
Table 6 Calibration and validation of BTOPMC over Mekong River basin (%) ns is Nash—Sutcliffe efficiency, Vol is volume ratio of Q, to Qo, Q, and Qp are simulated and observed discharge respectively at the station, pis precipitation, qoft is total runoff, gq, is recharge to groundwater, qp is groundwater release, ET is actual evapotranspiration from root zone, ETg is evaporation from interception. Except Q, and Q,, all items are averaged over upper basin of the station.
Figure 14 Raingauge (solid points), Thiessen polygon (solid lines) and subdivision (shades).
Figure 14 Raingauge (solid points), Thiessen polygon (solid lines) and subdivision (shades).
Figure 15 Hydrological simulation at Pakse (a) model calibration from 1983 to 1987, (b) model validation from 1988 to 1992. tive discharge happens, inflow and channel-in-water are routed totally as the outflow. This is reasoned by that the concentration time in a small catchment is rarely beyond one day. In the long term simulation, the recharge to groundwater is almost released as base flow, as indicated by the values of q, and q,. The groundwater runoff contrib- utes total runoff about 40—60% except at Yasothon. The to- tal evapotranspiration from interception and soil moisture in root zone, ET + ETo, takes the precipitation about 63.6% over upper basin at Vientiane, 57% at Mukdahan and 54.5% at Pakse, decreasing from upper stream to downstream be- cause the basin becomes wetter (more rainfall) down- stream. What we leant from this application is that hydrological simulation accuracy in large basins much de- pends on the modeling of evapotranspiration processes, including the potential evapotranspiration from intercep- tion and from soil moisture in root zone and the drying func- tion. Only when the accuracy of evapotranspiration estimation is guaranteed, could we get the accurate sepa- rating between the surface runoff and groundwater runoff and the accurate routing. We have shown that the parame- ter values are not reasonable in Block 5, the upper basin at Yasothon, and the evaluation indices indicate the hydrolog- ical simulation is not successful at Yasothon. The main rea- son may lie in the different runoff generation mechanism or the significant water resources regulation over the Yasothon catchment. The observed flow only takes about 10% of the precipitation. It seems like a semi-arid basin. The precipita- tion and vegetation LAI are much lower in this catchment than its neighborhood. In dry area, the infiltration excess process is often observed and may be a principal runoff gen- eration. But in present BTOPMC, this process is not consid- ered. Moreover, the evapotranspiration process is also probably different from other Mekong River basin parts. It is questionable to use the same value of drying function parameter for this catchment, as indicated by the much higher simulated runoff than the observed flow in volume. This study developed the S—W model for estimating the long- term PET over large basins using the parameter values from the literature and publicly available database. Its applica- tion to the Mekong River basin shows that the PET is not only controlled by the climate patterns, but also changes with the vegetation types and their development. The CRU TS 2.0 meteorological database, IGBP land cover classification, and particularly the satellite NDVI which measures the dy- namic change of vegetation morphology with the environ- mental conditions (e.g. the prolonged water stress) and seasons, provide the S—W model with complete data sets over a large basin in a long term. Given these public data are reliable to the interesting regions, this development is applicable at the global scale, essentially to the data-poor
Figure 15 Hydrological simulation at Pakse (a) model calibration from 1983 to 1987, (b) model validation from 1988 to 1992. tive discharge happens, inflow and channel-in-water are routed totally as the outflow. This is reasoned by that the concentration time in a small catchment is rarely beyond one day. In the long term simulation, the recharge to groundwater is almost released as base flow, as indicated by the values of q, and q,. The groundwater runoff contrib- utes total runoff about 40—60% except at Yasothon. The to- tal evapotranspiration from interception and soil moisture in root zone, ET + ETo, takes the precipitation about 63.6% over upper basin at Vientiane, 57% at Mukdahan and 54.5% at Pakse, decreasing from upper stream to downstream be- cause the basin becomes wetter (more rainfall) down- stream. What we leant from this application is that hydrological simulation accuracy in large basins much de- pends on the modeling of evapotranspiration processes, including the potential evapotranspiration from intercep- tion and from soil moisture in root zone and the drying func- tion. Only when the accuracy of evapotranspiration estimation is guaranteed, could we get the accurate sepa- rating between the surface runoff and groundwater runoff and the accurate routing. We have shown that the parame- ter values are not reasonable in Block 5, the upper basin at Yasothon, and the evaluation indices indicate the hydrolog- ical simulation is not successful at Yasothon. The main rea- son may lie in the different runoff generation mechanism or the significant water resources regulation over the Yasothon catchment. The observed flow only takes about 10% of the precipitation. It seems like a semi-arid basin. The precipita- tion and vegetation LAI are much lower in this catchment than its neighborhood. In dry area, the infiltration excess process is often observed and may be a principal runoff gen- eration. But in present BTOPMC, this process is not consid- ered. Moreover, the evapotranspiration process is also probably different from other Mekong River basin parts. It is questionable to use the same value of drying function parameter for this catchment, as indicated by the much higher simulated runoff than the observed flow in volume. This study developed the S—W model for estimating the long- term PET over large basins using the parameter values from the literature and publicly available database. Its applica- tion to the Mekong River basin shows that the PET is not only controlled by the climate patterns, but also changes with the vegetation types and their development. The CRU TS 2.0 meteorological database, IGBP land cover classification, and particularly the satellite NDVI which measures the dy- namic change of vegetation morphology with the environ- mental conditions (e.g. the prolonged water stress) and seasons, provide the S—W model with complete data sets over a large basin in a long term. Given these public data are reliable to the interesting regions, this development is applicable at the global scale, essentially to the data-poor
Related topics:
HydrologyPotential EvapotranspirationLand cover classificationExperimental MeasurementDistributed Hydrological ModelSeasonal change
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved