Effects of landuse change on the hydrologic regime of

the Mae Chaem river basin, NW Thailand.

P. Rattanaviwatponga, J. Richeyb,* , D. Thomasc, S. Roddab, B. Campbellb , M. Logsdonb

a Chemical Engineering, University of Washington, Seattle, Washington 98195, USA

b School of Oceanography, University of Washington, Seattle, Washington 98195, USA

c World Agroforestry Center ( ICRAF), Southeast Asia Regional Programme,

Chiang Mai, Thailand

Received Accepted

Abstract

Upland shifting cultivation, upland commercial crops, and lowland irrigated agriculture are

believed to be causing water resource tension in the Mae Chaem watershed in Chiang Mai, Thailand.

In this paper, we assess hydrologic regimes and buffering characteristics of the Mae Chaem River

with land-use change. Three plausible future forest-to-crop expansion scenarios and a scenario of

crop-to-forest reversal were developed based on the landcover transition from 1989 to 2000, with

emphasis on influences of elevation bands, irrigation diversion and soil compaction. Basin hydrologic

responses and river buffering indicators were simulated using the Distributed Hydrology Soil

Vegetation Model (DHSVM). Meteorological data from 6 weather stations inside and adjacent to the

Mae Chaem catchment during the period 1993-2000 were the climate inputs. Computed stream flow

was compared to observed discharge at Kaeng Ob Luang, located downstream from the district town

in Mae Chaem, and to hydrologic simulations by the GenRiver model. We observe altitudinal increase

in evapotranspiration and soil moisture. With current assumptions, expansion of highland crop fields

led to increasing unregulated annual and wet-season water yields and lowered buffering indicators,

compared to similar expansion in the lowland-midland zone. Actual downstream water availability

and stream peak flow were sensitive to irrigation diversion. We did not detect significant difference

after accounting for soil compaction. This modeling approach can be a useful tool for water allocation

and flood forecasting for small catchments undergoing rapid commercialization, because it alerts land

* Corresponding author. Tel.: +206-543-7339; fax: +206-685-3351 Email address: jrichey@u.washington.edu (J.E.Richey)

1

managers to the potential range of water supply in wet and dry seasons, to river buffering

characteristics, and to spatial distribution of basin moisture storage behaviors.

Keywords: Crops; Deforestation; Watershed hydrology; Landuse change; Northern Thailand; Stream

flow; Resource management

1. Introduction

Landscape and water resource management are major challenges for socio-economic

development of upland watersheds in Southeast Asia due to their association with downstream

environmental impacts. During recent decades, concerns about the impact on water resources of

changing patterns of land-use associated with deforestation and agricultural transformation have

created social and political tensions from local to national levels. In response, public policy decision-

making processes are now seeking both economic and conservation goals. Major concerns focus on

consequences of land-use change for water supply and demand, for local and downstream

hydrological hazards, and for biodiversity conservation. Walker (2002-a) cites numerous sources of

views that logging, shifting cultivation by mountain ethnic minorities, and commercial agriculture in

highland watersheds are believed to cause severe dry-season water supply shortages in northern

Thailand. Ziegler et al. (2004) refers to studies by Sharma in 1992 and by Tuan in 1993, which

conclude that similar types of land-use activities in highlands of Vietnam result in watershed

degradation such as soil and nutrient loss. Controversy over the eight-dam hydropower cascade

system on the Lancang river of the upper Mekong basin in China is an example at a wider scale where

potential effects on downstream river flows and sediment transport are an international issue for the

five countries sharing the lower Mekong river.

Water demand is the other side of the equation, as it also places constraints on water

availability. Dynamics of water use also relate to land-use change, especially through expansion of

lowland cultivation, irrigated upland fields, urban areas, and industrialization. Walker (2002-b) points

out that public debate is mostly centered on consequences of highland activities on water supply, but

there is little focus on increasing levels of stream water diversion by lowland dry-season irrigated

2

agriculture.

In this paper, we seek to evaluate seasonal patterns and the composition of hydrologic

components of the Mae Chaem River in northern Thailand. The second objective is to assess effects

of land-use conversion between forest and croplands on the composition of basin hydrology, on water

availability in terms of annual and seasonal water yields, and on the buffering capacity of river

discharge. Specifically, we assess the influence of elevation bands of agricultural fields (highlands

versus lowlands), irrigation diversion, and soil compaction. These goals are achieved through use of

the Distributed Hydrology Soil-Vegetation Model (DHSVM) to simulate water components in the

basin, based on past, current, and hypothetical future land-use scenarios. Scenario analysis eliminates

interpretation problems associated with direct comparison of stream flow in paired watershed analyses

where basins have different underlying geological settings (Brunijnzeel, 2003). Some simulation

results are also compared to those from the Generic River Flow (GenRiver) model.

2. Mae Chaem basin: the study area

The Mae Chaem (Chaem River) watershed is located in Chiang Mai province of northern

Thailand (Fig. 1). It is a major upper tributary sub-basin of the Ping River, which in turn, is the largest

tributary of central Thailand’s Chao Phraya River. The Mae Chaem sub-basin is bounded by

coordinates 18o 06’ - 19o 10’ N and 98o 04’ - 98o 34’ E, and includes a total area of 3,853 km2 above

the Royal Irrigation Department (RID) river gauge station P.14. Sharp relief and forest vegetation

(and relatively sparse data) characterize the Mae Chaem. The basin has a wide range of elevation,

from 282 m.a.s.l. at its lowest point to 2,535 m.a.s.l. at the summit of its highest peak, Doi Inthanon

(Mount Inthanon) (Kuraji et al., 2001). Altitude variation induces different climatic zones with

distinctive types of natural land cover. Dominant vegetation includes dry dipterocarp and mixed

deciduous forests below 1,000 m.a.s.l., tropical mixed pine forest from 900 – 1,500 m.a.s.l. alternating

with hill evergreen forest that extend up to 2,000 m.a.s.l., and tropical montane cloud forest above

2,000 m.a.s.l. (Dairaku et al., 2000; Kuraji et al., 2001). Steep hillsides with slopes exceeding 25% are

a common landscape element, resulting in rates of natural soil erosion that prevent advanced soil

development. Thus, soils are relatively shallow and have limited water-holding capacity (Hansen,

3

2001). Dominant soil textures are sandy clay loam and clay loam.

The population of Mae Chaem is ethnically diverse and distributed among numerous small

villages. The majority Karen and the Lua ethnic groups live primarily in mid-elevation zones

between 600 to 1,000 m.a.s.l., with some communities extending into higher elevations. Ethnic

northern Thai (khon muang) villages are mostly clustered in lowland areas below 600 m.a.s.l.,

whereas Hmong and Lisu ethnic groups live mostly in highland villages located above 1,000 m.a.s.l.

Land use patterns in Mae Chaem have undergone substantial change during the past several

decades. As recently as the 1960s the agriculture mosaic was comprised of highland (above 1,000

m.a.s.l.) pioneer shifting cultivation that often included opium, mid-elevation (600-1,000 m.a.s.l.)

rotational forest fallow shifting cultivation with a decade long fallow period, and paddy and home

garden-centered cultivation in the lowlands (van Noordwijk et al., 2003; Thomas et al. 2002; Walker

2002-b). In the 1980s, development projects and programs in Mae Chaem began building

infrastructure and promoting commercial agriculture, under programs to reduce rural poverty and

promote alternatives to opium cultivation and shifting agriculture. Results have included significant

increases in production of highland cash crops such as cabbages and carrots, expansion of industrial

field crops such as soybeans and maize up watershed slopes above lowland paddies into mid-elevation

zones, expansion of irrigated paddy fields wherever terrain allows, and planting of fruit orchards in

some areas of all altitude zones (Praneetvatakul et al., 2001; Pinthong et al., 2000; Thomas et al.

2002; Walker, 2002-b).

3. Development of geospatial landscape and hydrology model

3.1 Development of the geospatial model of the Mae Chaem basin

3.1.1. Topography and flow network

Topography for the Mae Chaem basin was initially acquired as a 30-meter digital elevation

model (DEM) constructed by the World Agroforestry Centre (ICRAF), Chiang Mai. This 30-meter

DEM was then aggregated to 150-meter0 resolution (Fig. 2) using a mean calculation. Flow direction,

flow accumulation, and stream network were derived from the DEM at 150-meter resolution. Soil

depth was generated by DHSVM, based on the DEM, and was adjusted during model calibration.

4

3.1.2. Soil map and attributes

Soil data in Mae Chaem are very sparse and restricted to the lowlands. Therefore, a soil map

containing physical and chemical properties was constructed using SOILPROGRAM software, which

derives 5-minute resolution (about 10 km.) soil data from the WISE pedon-database developed by the

International Soil Reference and Information Centre (ISRIC) and the FAO-Unesco Digital Soil Map

of the World (FAO-1974 legend). The soil map was re-sampled to 150-meter resolution, with the

number of soil types equal to the number of unique values of physical and chemical soil properties.

Soil texture was assigned based on the percent sand and clay. Porosity and field capacity were

estimated from the soil texture triangle hydraulic properties calculator (Saxton et al., 1986).

Infiltration rates and soil depth were quantified using a local descriptive soil survey (Putivoranart,

1973), and was adjusted during calibration.

3.1.3. Vegetation and land use: 1989, 2000, future

Two landcover datasets (Fig. 3) form the basis for the landcover change scenarios in the

hydrology model. The original classification schemes of these data (Fig. A1 and A2) vary

significantly, so scheme modifications were made to achieve similarity between landcover data.

The first dataset in the land cover time series is a historical 1989 dataset, acquired from the

Land Development Division (LLD), Ministry of Agriculture and Cooperatives, Thailand. This data,

subsequently referred to as Veg 1989, originated as polygons, which were converted to a 150-meter

raster grid representation using a nearest-neighbor assignment algorithm. Data were then generalized

into 11 classes from its original 39. The second dataset represents landcover for the year 2000,

referred to as current landcover or Veg 2000. This dataset, also from the LDD, was prepared for the

model using the same procedure as that of the 1989 data. However, since the original 47 class names

in this dataset differed from those in the 1989 data, class names were reconciled by performing a

combinatorial analysis between the 1989 reclassified dataset and the 2000 original data. In this way

we were able to establish a correlation between the 11 classes in 1989 and the 47 original classes in

5

2000. This type of spatial overlay analysis returns not only the frequency of all unique combinations

of land cover types, but also a map product of the spatial commonalities. A plot was made to identify

the frequency of occurrence between a 2000 value (1 to 47) and a 1989 value (1 to 11). Based on this

plot, the 2000 vegetation values were re-assigned a value consistent with the frequency distribution of

shared space with the 1989 dataset.

Veg 2000 is employed as the reference land-use case, and four future scenarios (Fig. 4) were

created based on the transition from Veg 1989 to Veg 2000, with a focus on forest-to-crop conversion.

The first scenario represents reversal of all croplands back to evergreen needleleaf forests in zones

above 1,000 m.a.s.l., and to deciduous broadleaf forests below 1,000 m.a.s.l. Selected forest types

were generally in accord with actual dominant vegetation in the respective elevation zones. The

second scenario forecasts the doubling of cropland area in Veg 2000 by growing a buffer of new crop

cells around all existing crop patches. This ultimately increased the cropland share of total basin area

from 10.4% in 2000 to 19.9%. Finally, the third and fourth scenarios depict a doubling of cropland

that is limited to either highland zones of the basin (above 1,000 m.a.s.l.), or to lowland and midland

basin zones (below 1,000 m.a.s.l.). Growth of croplands limited to highland and to lowland-midland

basin zones increased cropland shares of total basin area to 18.0% and 19.1%, respectively. In both

cases, a buffer was grown around existing crop patches in the selected elevation range, while crop cell

area outside the selection remained the same as in 2000.

3.2. Climate Forcing and Hydrology

The climate of this mountainous basin is defined by large variations in seasonal and annual

rainfall that are influenced by tropical monsoon and maritime patterns (Walker, 2002-a), whereas the

orographic effect induces an altitudinal increase of spatial rainfall distribution (Dairaku et al., 2000;

Kuraji, et al., 2001). The average annual temperature ranges from 20 to 34 oC and the rainy season is

from May to October.

To best represent climatic variation within the catchment, we used daily rainfall, and

maximum and minimum air temperature records for the period of 1993-2000, which were obtained

from five meteorological stations and one agro-meteorological station (Fig. 5). Doi Inthanon (DO)

6

and Wat Chan (WA) stations are operated by the Royal Project Foundation, and their recorded values

were obtained from both ICRAF and the Royal Project Foundation. The Research Station (RE)

belongs to the Asian Monsoon Experiment-Tropics (GAME-T), led by the University of Tokyo in

Japan under the Global Energy and Water Cycle Experiment (GEWEX). Mae Jo Agromet

(TMD327301), Mae Hong Son (TMD300201), and Mae Sariang (TMD300202) stations are

managed by the Thai Meteorological Department (TMD). Data for these stations were acquired

directly from the respective agencies.

The hydrology model runs on a sub-daily time step. Thus, disaggregated temperature,

radiation, and relative humidity are generated from daily records using a diurnal interpolation

scheme from the Variable Infiltration Capacity (VIC) model (Liang et al., 1994; Maurer et al., 2002).

Total daily rainfall was evenly distributed through sub-daily intervals. Calculated 3-hour

precipitation values in 1998 – 2000 were then replaced by observed records for all TMD stations.

Wind speed was assumed to be 2 m/s for all stations except RE and TMD327301, where actual daily

wind speed records are available.

The Mae Chaem hydrologic regime consists of high flow from May to October and base-flow

from November to April (Fig. 6), contributing 70% of the total flow and an average annual water

yield of 270 mm (we consider the water year to begin in November of the year previous to the year

cited). Due to the strong orographic effect on precipitation (Fig. 7), the surface runoff ratio could be

between 12 – 25%, depending on selection of reference rainfall stations and the interpolation scheme.

Van Noordwijk et al. (2003) and Walker (2002-a) provide thorough discussions on the long-term

rainfall-discharge relationship in Mae Chaem.

For model calibration and testing, 1989 – 1999 daily average stream flow measurements were

acquired from GAME-T for the basin outlet at Kaeng Ob Luang (gage P.14), operated by the RID.

The estimated daily discharge from the stage height observation in 2000 was obtained from the RID

Hydrology and Water Management Center for the Upper Northern Region.

3.3. DHSVM: the hydrology model

To examine problems at this spatial scale, we use the Distributed Hydrology Soil Vegetation

Model (DHSVM) (Wigmosta et al., 1994), a regional-scale hydrologic model that can recognize the

7

spatial heterogeneity of the watershed. The model can be seen as a multi-layered representation of

characteristics and processes in a drainage basin that allows us to examine intersections of the data

required for describing both the basin’s topographic features and its climatology patterns. DHSVM is

intended for small to moderate drainage areas (typically less than about 10,000 km2), over which

digital topographic data allows explicit representation of surface and subsurface flows. DHSVM is

utilized for stream flow forecasting and for addressing hydrologic effects of land management or of

climate change. It is tested on several basins in the USA (Bowling et al., 2000; Bowling and

Lettenmaier, 2001; Storck, 2000; VanShaar et al., 2002) and in British Columbia (Schnorbus and

Alila, 2004).

The model simulates soil moisture, snow cover, runoff, and evapo-transpiration on a sub-daily

time scale. It accounts for topographic and vegetation effects on a pixel-by-pixel basis, with a typical

resolution of 30 to 150 m. Snow accumulation and snow melt are calculated by a two-layer energy-

balance model. Evapo-transpiration follows the Penman-Monteith equation. The multi-layer soil

column in each pixel is a series of soil moisture reservoirs, and saturated sub-surface flow exists only

in the deepest soil layer. Runoff generation is represented by saturation excess and infiltration excess

mechanisms. Stream water balance was derived using Linear-reservoir routing.

The hydrology model sets requirements for the climate data model and the landscape model

based on basin characteristics.

3.3.1. Simulation conditions

The spatial domain was partitioned into 150-meter grid cells, and 3-hourly-timestep

simulations were run on the current landcover (Veg 2000), using climate data from June 1993 - May

1995 for calibration, and from June 1995 - October 2000 for model testing. Climate data across the

basin was computed from data of the 6 meteorological stations using a nearest-station interpolation.

The soil profile is divided into 3 root zones, 0-30 cm, 30-60 cm, and 60-100 cm. Lateral subsurface

flow was calculated using a topographic gradient. In the routing scheme, roads were not included, and

stream classification was based on Strahler stream order and segment slope, derived from the DEM.

Temperature and precipitation lapse rates were assumed constant for the entire catchment. After the

8

initial run, the approximate amount of irrigation diversion was subtracted from simulated stream flow

at the basin outlet, and then calibrated and tested against daily discharge observed at station P.14.

To observe seasonal and inter-annual variability of hydrologic responses, and to study effects

of land-use change, the same set of climate data and parameters were used for all vegetation

scenarios, both with and without irrigation. For future scenarios II, III, and IV, two sets of soil

conditions were applied. The first condition was consistent with soil properties of other scenarios and

the second reflected soil compaction by altering properties of the top-soil layer. In the latter, bulk

density was assumed to be 1800 kg/m3, the typical value that interferes with root growth for crops

grown in clay loam, and soil porosity was adjusted accordingly. The soil infiltration rate and vertical

hydraulic conductivity were reduced to 50% of calibrated values. In all cases, the soil matrix was

considered uniform and did not have a preferential flow path.

When irrigation was considered, daily irrigation consumption was calculated and subtracted

from computed daily discharge to account for water diversion to irrigated area. Crops were divided

into 3 categories based on their water demand: wet season rice, dry season rice, and cash crops (Table

1). Irrigated area was approximated from the number of pixels of each crop type in the original 1989

and 2000 landcover data sets (Fig. A1 and A2). Percentages of total irrigated areas in the basin in

1989 and 2000 were used to project a range of potential irrigated areas in the future scenarios. The

following assumptions were made in calculating irrigation diversion: First, only 1/8 of the area

designated as swidden cultivation in the original classification scheme was used for irrigated

cropping. Second, for general field-crop classes, half of the area was wet-season rice and the other

half was cash-crop; the composition of incremental cropland in future scenarios was divided in the

same manner. Diverted water was set equal to crop water demand, and this amount was then added to

simulated evapo-transpiration to maintain the water balance. Table 2 summarizes all simulation

conditions.

3.3.2. Calibration and validation

Calibration parameters were total soil depth, soil lateral and vertical hydraulic conductivities,

and vegetation parameters. Values of vegetation parameters were initially set to correspond with

9

Global Land Data Assimilation System (GLDAS) vegetation parameters mapped to the University of

Maryland (UMD) classification scheme, and were then tuned to northern Thailand based on forest

description by Gardner et al. (2000). Precipitation and temperature lapse rates were approximated

from linear regression of daily precipitation and temperature records and elevation gain. Final

calibration parameters are listed in Table 3.

Simulated hydrographs from May 1994 – October 2000, before subtracting irrigation

diversion, using Veg 2000 were compared to observed stream flows at gage P.14 (Fig. 8-a). Data from

May 1993 – April 1994 were excluded from the comparison because they were used in the start up

period. The model captures the on-set of the storm season, and estimated average wet-season flows

were relatively close to recorded values (Fig. 8-b). The average natural base flow was underestimated,

and the difference was higher for dry-seasons that follow a wetter storm season (Fig. 8-c). This is not

surprising since the model does not include deep groundwater components. In terms of percentage

deviation of prediction from observation within a season, the model is more consistent in estimating

daily dry-season flows, compared to wider-scattering in values of wet-season flows (Fig. 8-d). Thus,

we can infer that prediction accuracy is governed by climate inputs during the storm period, whereas

factors describing soil moisture storage and water transmission behavior dominate during the dry

season.

In other words, model performance depends on input data, operational parameters and

operational procedure. Rainfall measurement and appropriate basin-wide meteorological data

interpolation from weather station records are critical due to strong orographic effects. Estimation of

sub-daily climate input functions from daily data records is another source of uncertainty. In terms of

operational parameters, the simulations were sensitive to soil lateral hydraulic conductivity and soil

depth.

4. Hydrologic flow paths: current conditions and scenarios

Highland watersheds play an important role on controlling the streamflow regime (Tangtham,

1998) and its hydrologic behaviors, such as water yield and runoff generation, depend on the

10

vegetation cover, soil and geologic setting. We will evaluate the hydrologic response to the current

land cover and to the effect of forest-to-crop conversion in terms of the water yields and spatial

variation of soil moisture and evapotranspiration inside the basin. In addition, we analyze the

evenness of stream flow, represented by the buffering capacity of stream discharge relative to the

rainfall peaks, and the indicators of high peak discharge (van Noordwijk et al., 2003). These two

factors tell us how a river will respond to heavy storms regarding flooding potential.

4.1. Hydrologic dynamics under current conditions

For current conditions, the predicted values of basin hydrology were also compared to those

computed from the physically-based GenRiver model on Mixed-Landuse Scenario. The model

description, input data and landcover data set were described in the paper by van Noordwijk et al.

(2003).

4.1.1. Water yields and river buffering indicators

The runoff mechanism was mainly the sub-surface flow and no overland flow was produced.

High flow comprises 70% of total flow (Table 4) the surface runoff makes up about 20% of total

rainfall (Table 5). Predicted annual yields from DHSVM, accounting for irrigation, were 20% and

10% lower than observed and calculated values from GenRiver, respectively, due to lower estimate

for base flow. However, the magnitude of forecasted low flow was sensitive to the approximation of

irrigation consumption, and will be discussed in section 4.2.

In terms of river buffering capacity, the buffering of Mae Chaem river decreases as the basin

water yield increases (Fig. 9-a). However, we do not observe significant correlation of high peak

discharge to the surface runoff ratio over 6-year period (Fig. 9-b).

4.1.2. Evapotranspiration and soil moisture

Simulated average annual evapotranspiration was 760 mm, corresponding to 80% of basin-

wide estimated precipitation. Evapotranspiration was highest during June to September and reached

11

minimum values in January and February (Fig. 10-a). The seasonal trend is positively correlated with

rainfall seasonality.

Spatially, soil moisture was high near ridge areas and values were lower in lowlands (Fig.

11). Vertically, the deeper soil layer (60 - 100 cm) was moister and moisture storage was less

sensitive to precipitation level than in the upper layer (30 - 60 cm). The spatial pattern of

evapotranspiration also shows higher values on watershed boundaries and less along the main channel

path.

Direct observation indicates that soil moisture dynamics may follow spatial variation of

rainfall across the basin. To analyze if spatial relationships exist, basin elevation data was categorized

into 5 zones, and zonal means of soil moisture and evapotranspiration were computed. Results

showed an altitudinal increase in soil moisture (Fig. 12) and evapotranspiration (Fig. 13). Gaps in soil

moisture differences between lower and higher elevation zones were more significant during heavy

rain, but the reverse was true for the evapotranspiration. Seasonally, soil moisture was more static in

the lowland-midland zone, whereas in the highlands it followed the precipitation profile. In contrast,

evapotranspiration was dynamic in all elevation bands.

In summary, sub-surface flow was the main source of Mae Chaem river discharge, and

sensitivity to individual storms is due to short travel times caused by steep topography. The soil

column is relatively dryer in the lowlands, with less precipitation and less evapotranspiration. High

interannual variability during 1995-2000 resulted in simulated runoff ratios ranging from 0.15-0.25,

and buffering indicators decreased with increasing discharge fractions.

4.2. Effects of land-use change on hydrologic responses

The most important concerns regarding forest-to-crop land-use change relate to water

availability during the dry season and to potential downstream flooding hazard. Thus, we compared

buffering indicators and high peak flow responses among scenarios, as well as change in basin

moisture storage behavior. Scenario analysis based on the same climate forcing data limits the focus

of hydrologic change to only the influence from long term land-use change and excludes short-term

variation due to seasonal fluctuation.

12

4.2.1. Conversion from forests to croplands and influence of irrigation diversion.

Under the current set of model parameters, we demonstrated that increased croplands caused

greater natural (unregulated) seasonal and annual water yields, and lower evapotranspiration (Table 6)

compared to simulation using Veg 2000. The opposite trend was true when croplands were converted

to forests as in Scenario I. Highland crop expansion (Scenario III) yielded the highest annual and wet-

season water yields, while doubling of crop areas in lowland-midland zones (Scenario IV) resulted in

approximately the same water yields as under Veg 2000. If crops were converted back to forests

(Scenario I), the basin will provide less unregulated stream flow, and lose more water through

evapotranspiration. More forest increases river buffering capacity (Fig. 14-b) and yields a higher

range of discharge fractions. On average, expanded lowland-midland cultivation results in higher

buffering indicators and lower peak flows than the highland counterpart (Fig. 14-b, Fig. 14-d).

The magnitude of differences in stream flow behavior among scenarios depends on the

approximation of irrigation diversion. Low-season flow is a volatile component and available yield at

the basin outlet varies from 62% of regulated flow under Veg 2000, to 55% under Scenario III, and to

46% on Scenarios II and IV. Wet-season discharge is less sensitive and the flow remaining after

diversion is about 90% in each case. Evapotranspiration is 5% higher than the non-irrigated case for

Veg 2000, and about 8% higher for Scenarios II - IV. Crop-to-forest conversion clearly has the

highest buffering indicators, and lowest peak flow indicators. Overall, irrigation diversion decreases

by 20% the rate of reduction in natural flow-buffering indicators with incremental discharge fractions.

Among crop expansion scenarios (Fig. 14-a), Scenario IV has, on average, higher buffering indicators

on a lower range of discharge fractions, and has a 13% greater rate of decrease in buffering indicators

than does Scenario III. Scenario II provides results that fall in the middle of those from Scenario III

and IV. Trends among forest-to-crop scenarios are similar for relationships of highest flow fractions

and runoff ratios (Fig. 14-c).

We calculated zonal means of soil moisture in the upper soil column (30-60 cm depth) in each

elevation band in Scenario III and IV and compared the results to those of Veg 2000, on a 3-hour

period of dry (Table 7), and wet (Table 8) days, to analyze if the signal of vegetation composition

13

change stands out. No significant differences were observed between scenarios on both time periods.

The zonal means based on vegetation types, limited to each single zone of elevation band were also

compared. The results were indifferent and were not reported here.

4.2.2. Soil compaction

Forecasted natural annual and seasonal water yields increase by an average of 3% for all

scenarios (Table 9) when soil infiltration rate and top-soil hydraulic conductivity are reduced to half

of calibrated values. Estimated evapotranspiration, runoff ratio, buffering indicator, and highest

monthly flow remain about the same as for non-compacted cases. Again, Scenario III provides the

highest annual and seasonal yields.

5. Discussion

5.1. Effects of topography and vegetation types on hydrologic patterns

Mae Chaem’s simulated runoff mechanism is primarily sub-surface driven, so that the basin

acts like a pump, with about 75-80% of input rainfall returning to the atmosphere as

evapotranspiration. Ziegler et al. (2000) conducted experiments to generate runoff and sediment flow

on unpaved roads, foot paths and agricultural land surfaces in a nearby sub-basin, and found that

overland runoff was rarely generated from hoed fields and none from fallow fields. Runoff production

depends on infiltration characteristics (Ziegler et al., 2004), antecedent soil moisture, the conductivity

through the soil, and rainfall depth. According to a descriptive soil survey at a site in Mae Chaem

district (Putivoranart, 1973), the soil is relatively well-drained and has moderate water holding

capacity. Tangtham (1998) also found that tropical mountainous soils usually have infiltration rates

above 130 mm/hr and high permeability. This may explain why overland flow was not generated.

Estimated runoff ratios were consistent with the 15-25% runoff ratio published in Alford’s study of

annual runoff in mountainous regions of northern Thailand (1992).

Seasonal and annual patterns of evapotranspiration integrate effects of air temperature,

relative humidity, and vegetation. Each vegetation type differs in root depth, surface albedo, leaf area

index, stomatal and aerodynamic behavior (Giambelluca, 2002; Giambelluca et al., 2003), and these

14

factors govern potential evapotranspiration. Estimated monthly evapotranspiration was comparable to

literature values (Table 4) and the trend was similar to simulations by the GenRiver model (van

Noordwijk et al., 2003) (Fig. 10). The difference in magnitudes was partly due to the discrepancy in

estimated basin-wide rainfall.

Spatial variation of soil moisture and evapotranspiration follows altitudinal patterns and is

dynamic. During the wet season, topographic effects dominate and homogeneous profiles of both

hydrologic components were observed. On the other hand, the influence of local plants and soil

characteristics are clearer during the dry season.

5.2. Impacts of forest-to-crop conversion

Based on future scenario analysis, effects of land-use change on seasonal and annual water

yields are a net balance of change in basin moisture storage size, vegetation pumping effects, and flow

regulation. Tropical forests have higher evaporation from rainfall interception and transpiration than

other landcover types, especially in northern Thailand where forests are referred to as ‘pumps’ rather

than ‘sponges’ (Bruijnzeel, 2004). Consequently, forest-to-crop conversion reduces pumping effects

due to lower evapotranspiration rates, releasing more water out into streams. This assumes that

agricultural practices do not cause significant soil compaction, which lowers infiltration rate and

hydraulic conductivity. Soil compaction typically increases infiltration-excess overland flow, reduces

the size of basin moisture storage, and induces slower travel time of sub-surface water to a river

channel. Since overland runoff rarely occurs on simulations and observations, soil compaction

appears to be the least important factor in water yield dynamics. We find no evidence to support the

notion that selected soil compaction conditions alter hydrologic behavior at the whole-catchment

scale.

Irrigation diversion is the most direct influence on discharge magnitude, and it causes

significant difference on potential ranges of water yields among vegetation scenarios. Thus, discharge

magnitude is sensitive to assumptions on the percentage of area irrigated, crop types, and crop water

needs.

15

Crop field expansion in the highlands (> 1,000 m.a.s.l.) transmitted more natural sub-surface

lateral flow than lowland-midland expansion. Changes in spatial soil moisture distribution were not

detected, however, due to orographic effects that produce more excess soil moisture in the highlands,

which percolates through well-drained soils. Accumulation of water as it moves downslope toward

the river channel is observable, whereas soil moisture storage is a state of the basin and change is

more difficult to capture over snap shots in time. Furthermore, the fraction of landuse-change parcels

is so small compared to the total area in an elevation band that change in soil moisture did not affect

the zonal mean. The implications of increased peak flow are for potential flooding from the main

channel rather than from overland flow. Lowland-midland expansion provided more even stream

flow, both seasonally and annually, in this setting.

5.3. Model performance: comparison with GenRiver and IHACRES models

Previous hydrologic modeling and scenario analyses in the Mae Chaem basin includes

simulation on gauged and ungauged sub-basins using the IHACRES metric-conceptual rainfall-runoff

model by Schreider (2002) and Croke et al. (2004), and simulation of the whole basin by van

Noordwijk et al. (2003) using the physically-based GenRiver model. While the conceptual model uses

only a few input parameters, the parameters for each sub-basin need to be calibrated individually and

spatial variability of the specified land cover change was not testable. Flow prediction at the Mae

Chaem basin outlet was not executed due to sparse input data, and Croke pointed that this limitation

makes use of a physically-based model inapplicable. In contrast, we believe the physically-based

approach is appropriate for this scale, and the distributed feature allows us to adjust calculation based

on the availability of data and level of complexity.

Simulated total basin rainfall and results from DHSVM were compared to those from

GenRiver as another test of model reliability. The dynamics of rainfall and evapotranspiration (Fig.

10) were relatively well-correlated (Pearson’s r = 0.94 and 0.88 respectively). Water yields and river

buffering characteristics were in the same magnitude (Table 5 and Fig. 9), except for dry-season yield.

Estimation of basin-wide rainfall in each model appears to be an important factor in differences

between simulation results, as seen in the amount of monthly rain used in GenRiver versus in

16

DHSVM (Fig. 10-a). The SpatRain model which generates rainfall for the GenRiver model accounts

for local patchiness of rainfall events, spatial rainfall intensity based on the inverse distance from the

core of the storm, and elevation effects. Consequently, its accuracy relies on long-term statistics of

rainfall patterns from reference stations. On the other hand, the current setting of DHSVM depends on

orographic effects and interpolation from weather stations. Performance of both models should be

similar on seasonal and annual scales, due to the highly fluctuating nature of basin rainfall behavior. If

the purpose is to evaluate basin response to individual storm events, however, we would expect to see

more dramatic differences in prediction. Another factor contributing to differences in model

performance is the lack of slow subsurface flow in DHSVM. Spatially distributed soil moisture or

other soil hydraulic field measurements are necessary to check the accuracy of model predictions. On

the other hand, once the model is calibrated, the advantage of a physically-based approach is that the

final parameter set is transferable, and little adjustment is required to start a new application in similar

geological settings.

The strength of a distributed model is its ability to identify if spatial or spatially-explicit

relations exist among model input and output parameters. Examples of applications include

examination of effects of elevation bands in conjunction with connectivity among same-land-use

patches, proximity of agricultural fields to streams, or adjacency of certain landcover types. In this

paper, we identified altitudinal increases in soil moisture and evapotranspiration.

6. Conclusions

Land-use change in Mae Chaem has largely featured agricultural transformations in different

altitude zones. Highland pioneer shifting cultivation has been replaced by expanded permanent fields

producing commercial horticultural crops, often with seasonal sprinkler irrigation. While some

midland rotational forest fallow shifting cultivation systems remain, others have been replaced by

rainfed permanent plots producing subsistence and commercial field crops. Irrigated paddy has

expanded where terrain allows, and lowland agriculture has increased dry season water use for

irrigated rice, cash crops and fruit orchards. The DHSVM hydrology model was used as a tool for

analyzing impacts of forest-to-crop conversion, and vice versa, on basin hydrology and water

17

availability at the basin outlet, resulting in the following conclusions:

1. The utility of a spatially-explicit, process-based analytical modeling environment is demonstrated

by its ability to reproduce hydrographs across a range of conditions, in a basin where data are

relatively sparse. At a minimum, the efficacy of the model as an intelligent data-interpolation engine

is clear. That the model does as well as it does implies that the constituent dynamics are relatively

well-understood and some confidence can be placed in the quantitative implications of scenarios. This

modeling approach can be useful in assessing the influence of spatial configuration or fragmentation

of landcovers.

2. Topography is the primary factor controlling climatic, vegetation, and, consequently, spatial

variation of Mae Chaem’s hydrologic components. We found altitudinal increase in soil moisture and

evapotranspiration.

3. The basin hydrology is sensitive to changes in landcover attributes, with a general pattern of

increasing runoff with migration from trees to crops due to decreasing evapotranspiration. As there is

no evidence that soil compaction in agricultural fields has significantly altered overall infiltration

rates, rainfed upland agriculture, especially in the midland zone, does not appear to result in lower

water availabilities downstream.

4. Irrigation has a counter effect by withdrawing water, and its overall impact will relate to its

seasonality, and to water use efficiencies of the crops, irrigation systems, and management regimes

employed.

5. Highland crop expansion (>1,000 m.a.s.l.) may lead to lower buffering indicators, higher peak

flow, and higher seasonal and annual yields. This is due to increased water yields resulting from

reduced evapotranspiration. Since the increased water percolates through well-drained soils, primary

implications are for downstream main channel flooding rather than increased overland flow, and

extractions for irrigation may reduce downstream impacts.

Acknowledgements

This work was supported in part by the Bank Netherlands Partnership Program. Findings and

18

interpretations are those of the authors and do not necessarily reflect the views of the World Bank, its

Executive Board of Directors, or the countries they represent. The work was further supported by the

U.S. National Science Foundation and the SEA/BASINS Program of the University of Washington

and Chulalongkorn University, the Mekong River Commission, and the Puget Sound Regional

Synthesis Model. Meteorological and stream flow data were provided by the Thai Meteorological

Department, the Royal Irrigation Department, the Royal Project Foundation, and the GEWEX-Asian

Monsoon Experiments (GAME)-Tropics. The soil descriptive survey was available from the Land

Development Department of Thailand. The authors thank Meine van Noordwijk for sharing input data

and simulation results from GenRiver model and supporting documents of SpatRain and GenRiver

models.

19

References

Alford, D., 1992. Stream flow and sediment transport from mountain watersheds of the Chao Phraya

Basin, northern Thailand: a reconnaissance study. Mountain Research and Development 12 (3),

237-268.

Bowling, L.C., Storck, P., Lettenmaier, D.P., 2000. Hydrologic effects of logging in western

Washington, United States. Water Resources Research 36 (11), 3223-3240.

Bowling, L.C., Lettenmaier, D.P., 2001. The effects of forest roads and harvest on catchment

hydrology in a mountainous maritime environment. Land Use and Watersheds: Human Influence on

Hydrology and Geomorphology in Urban and Forest Areas: Vol. 2. Water Science and Application.

American Geophysical Union, Washington, D.C., pp. 145-164.

Bruijnzeel, L.A., 2004. Hydrological functions of tropical forests: not seeing the soil for the trees?

Agriculture Ecosystems and Environment 104 (1), 185-228.

Croke, B.F.W., Merritt, W.S., Jakeman, A.J., 2004. A dynamic model for predicting hydrologic

response to land cover changes in gauged and ungauged catchments. Journal of Hydrology 291 (1-2),

115-131.

Dairaku, K., Kuraji, K., Suzuki, M., Tangtham, N., Jirasuktaveekul, W., Punyatrong, K., 2000. The

effect of rainfall duration and intensity on orographic rainfall enhancement in a mountainous area: a

case study in the Mae Chaem watershed, Thailand. Journal of the Japan Society of Hydrology and

Water Resources 13 (1), 57-68.

Gardner, S., Sidisunthorn, P., Anusarnsunthorn, V., 2000. A field guide to forest trees of northern

Thailand. Kobfai Publication Project. Chulalongkorn Book Center, Bangkok, Thailand.

20

20

Giambelluca, T.W., Ziegler, A.D., Nullet, M.A., Truong, D.M., Tran, L.T., 2003. Transpiration in a

small tropical forest patch. Agricultural and Forest Meteorology 117 (1-2), 1-22.

Giambelluca, T.W., 2002. Hydrology of altered tropical forest. Hydrological processes 16 (8),

1665-1669.

Hansen, P.K., 2001. Environmental variability and agro-ecological stratification in a watershed in

northern Thailand. Capacity Building in Biodiversity Project. Technical paper no. 2.

Kuraji, K., Punyatrong, K., Suzuki, M., 2001. Altitudinal increase in rainfall in the Mae Chaem

watershed, Thailand. Journal of the Meteorological Society of Japan 79 (1B), 353-363.

Liang, X., Lettenmaier, D.P., Wood, E.F., Burges, S.J., 1994. A simple hydrologically based model of

land-surface water and energy fluxes for general-circulation models. Journal of Geophysical

Research-Atmospheres 99 (D7), 14415-14428.

Maurer, E.P., Wood, A.W., Adam, J.C., Lettenmaier, D.P., Nijssen, B., 2002. A long-term

hydrologically based dataset of land surface fluxes and states for the conterminous United States.

Journal of Climate 15 (22), 3237-3251.

Pinthong, J., Thomsen, A., Rasmussen, P., Iversen, B.V., 2000. Studies of soil and water dynamics.

Approach and methodology. Working paper no. 2. Research Center on Forest and People in Thailand,

Tjele, Denmark.

Praneetvatakul., S., Janekarnkij, P., Potchanasin, C., Prayoonwong, K., 2001. Assessing the

sustainability of agriculture a case of Mae Chaem catchment, northern Thailand. Environment

International 27 (2-3), 103-109.

21

21

Putivoranart, S., 1973. Soil survey report no. 167 for Wang-pa, Mae Chaem district, Chiang Mai

(translated from Thai title). The Land Development Department of Thailand, Thailand.

Saxton, K.E., Rawls, W.J., Romberger, J.S., Papendick, R.I., 1986. Estimating generalized soil-water

characteristics from texture. Soil Science Society of America Journal 50 (4), 1031-1036.

Schnorbus, M., Alila, Y., 2004. Forest harvesting impacts on the peak flow regime in the Columbia

mountains of southeastern British Columbia: an investigation using long-term numerical modeling.

Water Resources Research 40 (5), Art. no. W05205.

Schreider, S.Yu., Jakeman, A.J., Gallant, J., Merritt, W.S., 2002. Prediction of monthly discharge in

ungauged catchments under agricultural land use in the Upper Ping basin, northern Thailand.

Mathematics and Computers in Simulation 59 (1-3), 19-33.

Storck, P., 2000. Trees, snow, and flooding: an investigation of forest canopy effects on snow

accumulation and melt at the plot and watershed scales in the Pacific Northwest. Water Resources

Series Technical Report no. 161. University of Washington, Seattle, WA.

Tangtham, N., 1998. Hydrological roles of the highland watersheds in Thailand, in: Thaiutsa, B.,

Traynor, C., Thammincha, S. (Eds.), Highland ecosystem management: Proceedings of the

International Symposium on Highland Ecosystem Management, 26-31 May 1998. Royal Anghang

Agricultural Station, Chiang Mai, Thailand.

Thomas, D., Preechapanya, P., Saipothong, P., 2002. Landscape Agroforestry in Upper Tributary

Watersheds of Northern Thailand. Journal of Agriculture (Thailand) 18 (supplement 1), S255-S302.

Van Noordwijk, M., Richey, J., Thomas, D., 2003. Functional value of biodiversity – phase II.

Technical report for activity 2: landscape and (sub) catchment scale modeling of effects of forest

22

22

conversion on watershed functions and biodiversity in Southeast Asia. Alternatives to Slash-and-Burn

Programme, Nairobi, Kenya.

VanShaar, J.R., Haddeland, I., Lettenmaier, D.P., 2002. Effects of land-cover changes on the

hydrological response of interior Columbia River basin forested catchments. Hydrological Processes

16 (13), 2499-2520.

Walker, A., 2002-a. Forests and water in northern Thailand. Resource Management in Asia-Pacific

Working Paper no. 37. Resource Management in Asia-Pacific Program. The Australian National

University, Canberra, Australia.

Walker, A., 2002-b. Agricultural transformation and the politics of hydrology in northern Thailand: a

case study of water supply and demand. Resource Management in Asia-Pacific Working Paper no. 40.

Resource Management in Asia-Pacific Program. The Australian National University, Canberra,

Australia.

Wigmosta, M.S., Vail, L.W., Lettenmaier, D.P., 1994. A distributed hydrology-vegetation model for

complex terrain. Water Resources Research 30 (6), 1665-1679.

Ziegler, A.D., Sutherland, R.A., Giambelluca, T.W., 2000. Runoff generation and sediment

production on unpaved roads, footpaths and agricultural land surfaces in northern Thailand. Earth

Surface Processes and Landforms 25 (5), 519-534.

Ziegler, A.D., Giambelluca, T.W., Tran, L.T., Vana, T.T., Nullet, M.A., Fox, J., Vien, T.D., Pinthong,

J., Maxwell, J.F., Evett, S., 2004. Hydrological consequences of landscape fragmentation in

mountainous northern Vietnam: evidence of accelerated overland flow generation. Journal of

Hydrology 287 (1-4), 124-146.

23

23

Fig. 1. Location and river network alignment of Mae Chaem basin in northern Thailand

24

24

Fig. 2. DEM, soil depth, and stream network grids (left to right) represented by the 150-m resolution.

Fig. 3. Mae Chaem landcover (Veg 1989) re-processed 1989, (Veg 2000) re-processed 2000 data sets

25

25

Fig. 4. Landuse conversion (Scenario I) crops to forest, (Scenario II) Double crops, (Scenario III) More upland crops, (Scenario IV) More lowland crops

26

26

Fig. 5. Location of climate stations with elevation (m. a. s. l.)

27

27

Fig. 6. Historic records of stream flow at gage P.14 and precipitation from Research Station (RE)

28

28

Fig. 7. Orographic effects on average annual rainfall (1989-2000)

29

29

Fig. 8. Comparison of observed and predicted hydrographs on Veg 2000 as (a) daily discharge, (b) average wet-season flow, (c) average dry-season flow, and (d) % deviation of simulated from observed values. Error bars represent the standard deviation.

30

30

Fig. 9. Quantitative indicators of (a) river buffering capacity and (b) transmission of high peak discharge in relation to annual (Jan - Dec) basin water yield indicators

31

31

Fig.10. Seasonal patterns of monthly simulated basin-wide evapotranspiration (ET) and rainfall simulated by DHSVM on Veg 2000 with irrigation and compared to values from GenRiver model

32

32

Fig. 11. Illustration of the underlying dynamics changes in hydrographs, with soil moisture in the root zones at 30-60 cm, 60-100 cm, evapotranspiration, and precipitation (top to bottom). Values are at time = 0:00-3:00 and simulated on Veg 2000

33

33

Fig. 12. Soil moisture temporal dynamics in the 30-60 cm root depth and the correlation with elevation zone, simulated on Veg 2000

34

34

Fig. 13. Evapotranspiration temporal dynamics in correlation with elevation zone, simulated on Veg 2000

35

35

Fig. 14. Comparison of (a) - (b) river buffering indicators and (c) - (d) transmission of high peak discharge in relation to annual (Jan - Dec) basin water yield indicators among landcover scenarios

36

36

Fig. A1. Original landcover in 1989

37

37

Fig. A2. Original landcover in 2000

38

38

Table 1

Monthly irrigation water demand (in mm) of northern agricultural crops (Schreider et al.,

2002)

Crop typeMonth

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Wet season rice 0 0 0 0 250 300 350 150 50 50 0 0

Dry season rice 250 200 200 0 0 0 0 0 0 0 300 500

Cash crops 150 150 100 0 0 0 0 0 0 0 300 100

Table 2

Simulation scenarios to look at effects of land-use type, irrigation, and soil compaction

Secondary factorsPrimary factor : Land-use change

Veg 1989 Veg 2000 Scenario I Scenario II Scenario III Scenario IV

Irrigated areas

0% X X X X X X

20%a X X X

35%b X X X

Soil compaction, no irrigation X X X

a Approximate maximum percentage of croplands being irrigated based on Veg 1989

b Approximate maximum percentage of croplands being irrigated based on Veg 2000

39

39

Table 3

Final values for DHSVM model parameters

Parameter Values

Ground roughness, m 0.02

Reference height, m 40

Rain LAI multiplier 0.0005

Temperature lapse rate, oC/m -0.0015

Precipitation lapse rate, m/m 0.002

Overstory height, m 20-30

Understory height, m 0.2-5

Vegetation vapor pressure deficit threshold, kPa 4000

Maximum stomatal resistance for overstory, s/m 4000-5000

Minimum stomatal resistance for overstory, s/m 200-400

Maximum stomatal resistance for understory, s/m 600-3000

Minimum stomatal resistance for understory, s/m 120-175

Overstory LAI2.3-10.8 (broadleaf)

4.8-12.5 (needleleaf)

Understory LAI 0.8-10

Overstory albedo 0.2

Understory albedo 0.2

Lateral soil hydraulic conductivity, m/s 0.0005 - 0.005

Vertical hydraulic conductivity, m/s 10-6 – 10-3

40

40

Table 4

Comparison of hydrologic compositions between observation, simulation by DHSVM on Veg 2000 with

irrigation diversion, and simulation by GenRiver on Mixed-landuse, based on water year

Source of computed

hydrologic components

Average values of hydrologic components (1995 – 2000)

Annual yield

mm (m3/s)

High flow a

m3/s

Low flow

m3/s

Annual evapotranspiration

mm

Observed 281 (34.3) 50.7 18.5 750 b 1230 c

DHSVM 215 (26.2) 54.7 7.6 762

GenRiver 247 (30.3) 43.7 18.0 968

a Exception: high flow is an average from 1994-2000

b Hill evergreen forest in Chiang Mai (Tangtham,1998)

c Typical mountainous watershed, excluding cloud forests (Tangtham,1998)

Table 5

Comparison of average observed and simulated annual (January to December) runoff ratios, buffering

indicators, and highest-month flow fraction from DHSVM (Veg 2000 with irrigation) and GenRiver. For

observed and DHSVM values, basin-wide DHSVM-generated rainfall was used in the calculation. Rainfall data

in GenRiver was generated from SpatRain model (van Noordwijk et al., 2003)

Source of computed

hydrologic components

Average values of hydrologic components (1995 – 1999)

Runoff ratio Buffering indicatorHighest monthly discharge

relative to mean rainfall

Observed 0.26 0.832 3.13

DHSVM 0.21 0.815 4.22

GenRiver 0.18 0.893 3.91

41

41

Table 6

Potential rangesa of basin hydrology simulated on different landcover scenarios, with and

without irrigation based on water year (November – October)

Landcover scenarios

Average hydrologic components (1995 – 2000)

Annual yield,

mm (m3/s)

High flow,

m3/s

Low flow,

m3/s

Annual evapotranspiration,

mm

Veg 2000I 215 (26.2) 54.7 7.6 762

NI 249 (30.5) 58.6 12.0 727

Scenario I NI 223 (27.2) 53.3 11.1 752

Scenario III 202 (24.7) 53.6 5.8 781

NI 261 (31.8) 61.2 12.5 715

Scenario IIII 220 (25.6) 56.8 7.0 759

NI 269 (32.8) 63.1 12.7 707

Scenario IVI 193 (23.6) 51.6 5.6 786

NI 251 (30.7) 59.1 12.2 724

a Based on percentage of irrigated croplands in Table 2

I = With irrigation

NI = No irrigation

42

42

Table 7

Comparison of soil moisture in the 30-60 cm depth in all elevation bands among land-use scenarios for

simulations on March 9, 1999 time step 0:00 – 3:00

Elevation range, mLandcover scenario

Veg 2000 Scenario III Scenario IV

> 1600 0.227 0.229 0.227

1200 - 1600 0.180 0.186 0.180

800 - 1200 0.163 0.166 0.164

500 - 800 0.151 0.151 0.151

< 500 0.147 0.147 0.149

Table 8

Comparison of soil moisture in the 30-60 cm depth in all elevation bands among land-use scenarios for

simulations on October 10, 1999 time step 0:00 – 3:00

Elevation range, mLandcover scenario

Veg 2000 Scenario III Scenario IV

> 1600 0.366 0.366 0.366

1200 - 1600 0.331 0.331 0.331

800 - 1200 0.298 0.299 0.299

500 - 800 0.185 0.185 0.186

< 500 0.147 0.147 0.149

43

43

Table 9

Comparison of simulated hydrologic components among future forest-to-crop scenarios with soil compaction

(no irrigation)

Landcover

scenarios

Average hydrologic components (1995 – 2000)

Annual yield a,

mm (m3/s)

High flow a,

m3/sa

Low flow a,

m3/s

Annual

Evapotranspiration a,

mm

Runoff

ratio bBuffering indicator b

Scenario II 269 (32.8) 62.9 12.8 708 0.26 0.81

Scenario III 277 (33.8) 64.8 12.9 700 0.27 0.80

Scenario IV 259 (31.6) 60.9 12.6 717 0.25 0.81

a Calculation based on water year (November – October)

b Calculation based on Julian year (January – December)

44

44