Simulating effects of climate scenarios on hydrological processes in southern Brazil using a lysimeter


Correspondence to: A. Pinheiro, Fundação Universidade Regional de Blumenau, Rua São Paulo, 3250, 89030-350, Blumenau—SC, Brazil. E-mail:


The purpose of this article is to simulate how future climate scenarios will affect evapotranspiration, surface runoff and drainage in an agricultural area. The area's behaviour is represented by a lysimeter consisting of an undisturbed soil core with volume 1 m3, sited in the basin of the Concordia River near the town of Lontras in the southern Brazilian State of Santa Catarina. The simulation used the SWAP (Soil-Water-Atmosphere-Plant) model, calibrated and verified with data on rainfall, temperature, relative humidity, wind speed and direction, radiation and soil water content over a 34-month period. Hydrological processes characterizing the basin soils were determined from present climate as baseline, and from series generated for maximum and minimum scenarios of carbon emissions (scenarios A2 and B2, respectively). Simulations were made for conditions of bare soil and for maize cultivation. Results showed that for these two conditions A2 and B2 give annual mean rainfall depths, evapotranspiration and interception of the same order of magnitude as in the baseline period. Depths of surface runoff were greater than baseline for both scenarios and for both cultivation conditions. For bare soil, the increases were about 118 and 268% for B2 and A2, respectively. These differences were generated from frequency distributions of maximum daily rainfalls in scenarios A2 and B2 that were greater than those for baseline. However, drainage flow was lower for the two soil conditions, and for the two scenarios A2 and B2, relative to baseline conditions.

1. Introduction

Increased concentrations of atmospheric carbon dioxide, methane and nitrous oxide have been observed since the pre-industrial period, contributing significantly to global warming. The Intergovernmental Panel for Climate Change (IPCC) has found an increase of 0.74 °C in the planet's temperature over the last 150 years. On top of the trends observed in historic records, a further increase of 0.2 °C per decade in global temperature is also predicted (IPCC, 2007). Over southern Brazil, Campos (2010) shows that considerable increases will occur in mean air temperature and in temperature extremes, both maxima and minima. From analyses at annual, seasonal and monthly time-scales, it was shown that there could be a temperature increase of up to 5.8 °C in spring-time.

Regional models coupled with models of global circulation have been used to explore future climate, the scenarios that are most used being A2 and B2. The A2 scenario, with high emissions of greenhouse gases, is seen as pessimistic; B2, with low emission of greenhouse gases, as optimistic. Temperature anomalies greater than 3.5 °C were estimated for almost all the State of Bahia in the Brazilian north-east, except for the coastal area where anomalies were between 2.0 and 3.5 °C. A reduction in rainfall of 70% is estimated for the State's coastal strip, increasing to 80% in its northern part (Tanajura et al., 2010). In the metropolitan areas of São Paulo and Rio de Janeiro, the increase in mean annual temperature could reach to between 3 and 4 °C. In terms of rainfall, an increased number of consecutive dry days is expected, together with a reduction in the number of wet days in the year with more than 10 mm of rain, and an increase in the annual maximum rainfall accumulated over five consecutive days (Torres et al., 2009). For the township of Taubaté in Sao Paulo State, an increase in air temperature of between 0.5 and 2.7 °C is expected, with an increased rainfall of between 80 and 150 mm (Horikoshi and Fish, 2007). Future projections for the south of Brazil suggest significant positive increases in minimum temperature of between 0.5 and 0.8 °C per decade, and trends in maximum temperature of 0.4 °C per decade. At the seasonal scale, trends in maximum temperature could vary from 0.4 to 0.9 °C per decade in summer and from 0.6 to 0.8 °C per decade in winter. During summer, negative trends in diurnal temperature range of − 0.3 °C per decade are expected for the eastern part of the States of Rio Grande do Sul, Santa Catarina and Paraná (Marengo and Camargo, 2008). Campos (2010) found warming trends for the same region. In annual terms, increases under the A2 scenario should lie between 2.5 and 5 °C. In summer, maximum temperatures may increase by up to 5.2 °C under scenario A2, and of at most 4 °C under B2. In spring-time, increases should be still greater. Maximum values could reach 5.8 °C under A2 and 4.9 °C under B2.

It is expected that the increase in air temperature and changes in rainfall will be reflected in the runoff regime. Future projections show a trend towards a drier climate than at present in western Amazonia, with a reduction in atmospheric humidity and soil moisture, and a reduction in streamflow (Liberato and Brito, 2010). Salati (2010) showed a trend for flow to be reduced up to the year 2100 in most of 12 Brazilian drainage basins, but those draining to the South Atlantic, and the River Uruguay, are expected to show increased flows up to the end of this century. Horikoshi and Fish (2007) concluded that increased temperature together with increased rainfall would cause an increase in water deficit of about 50–80 mm, and a reduction in water excess, of about 200 mm in the Taubaté township of the State of São Paulo. Mello et al. (2008) found that the A2 scenario would give increased water availability in the Paracatu basin of Minas Gerais, with variations of up to 131% in seasonal runoff, whilst no significant changes were found under the B2 scenario.

Abbaspour et al. (2009) studied the effects of climate change on Iran's water resources for the periods 2010–2040 and 2070–2100, using output from the Canadian Global Coupled Model, for scenarios A1B, B1 and A2. Their results showed exceptionally high rainfall over most of the country during 2073–2099, under the A2 scenario. The calculated recharge to aquifers indicated a reduction of between 50 and 100% in the eastern half of the country, whilst the north-west it increased because of increased rainfall. Equally Li et al. (2009) compared the effects of A2 and B2 scenarios relative to the baseline period 1961–1990, in four basins of Taiwan, concluding that, in general, increases in surface runoff and evapotranspiration are to be expected under both future scenarios, together with reductions in groundwater flow.

There is now a growing concern about the availability of water resources in regions where economic growth is leading to high water demand, and in regions where water deficit has already brought social and economic problems, with the need better to understand how resources should be used in regions where trends and future projections show changes in rainfall. Even in regions where increased rainfall is expected, reduced flow may occur as a consequence of increased evaporation losses from higher temperature (Salati et al., 2010).

From this background, the aim of this article is to use numerical simulation to analyse the effects of future scenarios on surface runoff, evapotranspiration and drainage flow in a lysimeter containing 1 m3 of undisturbed soil, representative of an agricultural area. The numerical simulation uses the SWAP (Soil-Water-Atmosphere-Plant) model for one-dimensional flow through the soil profile. The lysimeter is sited in the basin of the Concordia river, near the township of Lontras in the State of Santa Catarina. Using climate series generated for conditions of maximum and minimum carbon missions, hydrological processes are generated that are characteristic of soils within the basin.

2. Materials and methods

The results reported in this article were obtained by calibrating the SWAP model, using series of hydrological data series collected from observation of the lysimeter mentioned above. Having verified the model's performance, it was applied to climate data from the period 1961–1990 as a baseline, representing present climate conditions. Simulations were then made for the future climate scenarios B2 and A2 developed by the Intergovernmental Panel on Climate Change (IPCC, 2007) for the period 2071–2100, based upon different projections in the future of greenhouse gas emissions. The A2 scenario, corresponding to high emissions, is considered pessimistic; the B2 scenario, with low emissions, is regarded as optimistic with respect to greenhouse gas emissions.

2.1. The SWAP model

The SWAP model is based on the Richards equation which is obtained by combining the Darcy equation with that for continuity (van Dam, 2000), ensuring mass balance and continuity in heterogeneous media. The SWAP model integrates flows of water and heat, solute transport, details of crop growth in different soil profiles, various levels of regional drainage and management of surface water. To simulate vertical movement of water, the model uses the Richards equation given by

equation image(1)

where h is soil water potential (cm); K(h) is hydraulic conductivity (cm d−1); θ is volumetric water content (m3 m−3); T is time (d); z is the vertical co-ordinate (cm) and S(h) is the rate of water extraction by roots (m3 m d−1).

The Richards equation requires knowledge of soil water hydrodynamics and of elements in the relation between matric potential and soil water content, through a curve of water retention and hydraulic conductivity. Water availability is determined by the soil's physical characteristics, which in turn determine the greatest quantity of water that the soil can store and supply to a crop. Water balance in the root zone is given by the difference between inputs (rainfall, interception and irrigation) and outputs (transpiration, evaporation, percolation and infiltration) of water to the soil (Barros, 2010).

SWAP simulates the water balance in a vertical column of cultivated soil storing a volume W of water at a given instant of time, using:

equation image(2)

where ΔW is the change in soil water storage (m), P is rainfall input (m), I is irrigation (m), R is surface runoff (m), Pi is water intercepted by vegetation (m), Ta is transpiration (m), Ea is evaporation of soil water (m) and Pe is drainage flow (m).

Potential evapotranspiration from bare soil and vegetation is estimated from the Penman–Monteith equation (Monteith, 1965). Its value is used to determine evaporation and actual transpiration. Actual evaporation depends on the soil's ability to transport water to the surface, and actual transpiration depends on conditions of soil moisture, salinity and root density (Kroes et al., 2008) which is different for different crops.

2.2. Experimental set-up

The area where the model was used lies in the Concordia river basin near the Lontras township of Santa Catarina, where the volumetric drainage lysimeter was installed, consisting of an undisturbed soil sample with volume 1 m3. The mean slope at its surface is of 4%. The soil is predominantly Cambisol. The lysimeter was constructed from acrylic sheets 8 mm thick, and was filled with undisturbed soil with cross-section '1. Introduction' m2 and depth 1 m. The lysimeter was constructed as two acrylic boxes, an empty one that was used when collecting the soil sample and the second to construct the lysimeter base, as described by Oliveira et al. (2010). Tubes were installed near the surface and at the soil base to collect surface runoff and drainage. Water draining from the lysimeter bottom represents aquifer recharge.

Surface runoff and lysimeter drainage were measured by commercially available tipping-bucket rain gauges, as in Braga et al. (2009). Volumes can be measured up to 50 ml per tip. A stable calibration curve was constructed for each rain gauge. Each tip is recorded on a data logger and transformed to a depth of flow using the calibration curve. Records are stored at each 15-min interval. In addition to the automated systems, 5 and 50 l containers were used to collect surface runoff and lysimeter drainage respectively and were connected to the lysimeter by Polyvinyl chloride (PVC) tubing. Volumes of water collected in the containers were removed after each rainfall event.

The lysimeter is equipped with pressure tensiometers to measure soil water tension. There are three UMS T4 pressure sensors at depths 10, 30 and 70 cm giving measurements that are recorded continuously by data logger. Soil water tension is recorded in hPa at 15-min intervals, so that soil moisture can be estimated from the retention curve. Water stored in the soil, and the way that soil water varies in time, is then determined by integrating over the soil profile.

Maize is sown in the lysimeter annually after removal of dead material and with soil cultivated to 30 cm depth. In 2008, 2009 and 2010 sowing was always in the first fortnight of November, with crop cycles of 159, 128 and 132 d, respectively. Yields were 1800, 7860 and 5700 kg ha−1.

2.3. Climate data

Records of temperature, relative humidity of air, wind speed and direction, atmospheric pressure and solar radiation used in the calibration and validation phases of the SWAP model were obtained from an automatic weather station installed in the Concordia basin. Rainfall is measured using a Davis® tipping-bucket gauge with Novus® data logger, recording at 5-min intervals. The calibration period was from 1 March 2008 to 31 December 2010, and validation from 1 January 2011 to 31 May 2011.

Daily data on radiation, maximum and minimum temperatures, relative humidity of air, wind speed and rainfall were used for the baseline period 1961–1990, and for the simulated period 2071–2100, under maximum (A2) and minimum (B2) carbon dioxide emissions obtained from the PRECIS System (Providing REgional Climates for Impact Studies), issued by the Brazilian space agency INPE (Instituto Nacional de Pesquisas Espaciais) over a grid 0.5 × 0.5°.

Two conditions of soil use were simulated using the lysimeter, the first corresponding to bare soil, and the second to annual maize cultivation repeated over the three years of observation. Under the first, bare-soil condition, there was no interception or transpiration by vegetation. All simulations used a daily time interval, and the phenological patterns of maize were assumed to be the same for all three climatic scenarios.

3. Results and discussion

3.1. SWAP model calibration

Table 1 shows observed and simulated components of the water balance over the calibration period. Simulated and observed values show no evidence of statistical differences, relative to the volume of drainage in the period analysed, indicating that the model adequately simulates flow. Measured evapotranspiration is given by water balance considerations, in which water intercepted by crop foliage over the lysimeter is included. When simulating, interception is calculated from the leaf area index, representing the depth of water intercepted by the maize crop.

Table 1. Water balance over the calibration and verification period (cm)
Water balance componentCalibrationVerification
Surface runoff1.642.690.790.02
Change in storage5.895.561.950.16

Figure 1 shows depths of drainage simulated by SWAP, together with drainage depths from the lysimeter. The coefficient of determination, calculated for daily time intervals, was 0.83. At the time of sowing, the soil was turned over to a depth of 30 cm, corresponding to conventional cultivation practice. During this period, the drainage depths simulated by SWAP were not adequately reproduced; this is because of changes in soil characteristics of density and porosity which altered the soil tensiometry and increasing drainage flow (Alves and Cabeda, 1999; Medeiros et al., 2009). These changes in soil characteristics could not be introduced in the data used as model input.

Figure 1.

Daily rainfall and drainage: measured and simulated by the SWAP model, (a) calibration and (b) verification period

Figure 2 shows the evolution over time in months of observed evapotranspiration, together with that simulated using SWAP. The monthly coefficient of efficiency is better than 0.85, showing that the model reproduces evapotranspiration adequately, even when the difference in interception is allowed for. Some fluctuation in the observe values is the result of flow times being out of phase during the last 2 d of each month, and to the periods during which maize was developing. It will be recalled that evapotranspiration is measured and simulated at a daily time-scale, but is integrated over months for ease of graphical exposition.

Figure 2.

Monthly rainfall and evapotranspiration: measured and simulated by the SWAP model

3.2. Simulation of climate scenarios

The scenarios B2 and A2 gave mean annual rainfalls of the same order of magnitude as in the baseline period. For the period 1961–1990, mean annual rainfall was 1577.9 mm; under scenario B2 it was 1469.7 mm and for scenario A2, 1507.4 mm. These mean annual values show a reduction, relative to baseline, of 6.5% for B2 and 4.2% for A2. These changes are not maintained over time, as shown in Figure 3. Regarding maximum and minimum annual totals of rainfall, the baseline period showed more extreme values, 2662.1 and 734.7 mm. In the A2 scenario, the maximum annual total was 2261.3 mm, greater than that for B2. However, the lowest annual rainfall total under the B2 scenario was less than the lowest under A2.

Figure 3.

Annual values for the baseline period (1961–1990) and for scenarios A2 and B2 (2071–2100): maize cultivation

Similarly, mean annual evapotranspiration in both A2 and B2 scenarios was close to that in the baseline period. Between 1961 and 1990, mean annual evapotranspiration was 523.2 mm; in the scenario B2, it was estimated as 509.3 mm and in A2, 506.1 mm. In percentage terms, these reductions relative to baseline were 2.6% for B2 and 3.2% for A2. Maximum and minimum annual evaporation totals in the baseline period were 616.6 and 440.4 mm, respectively. The expected largest and smallest annual evapotranspiration totals are both greater for scenario B2 than for scenario A2, the largest being 574.1 mm for B2 and 563.6 mm for A2. For the smallest annual totals, the lowest annual evapotranspiration under A2 was lower than that for B2, and was also lower (416 mm) that for the baseline period.

Similarly, scenarios B2 and A2 show mean annual evapotranspiration close to that of the baseline period, both for bare soil and for maize cultivation. For maize, mean annual evapotranspiration was 646.6 mm over the period 1961–1990, whilst under the B2 scenario it was estimated as 645.2 mm and, for the A2 scenario, 659.0 mm. The A2 scenario gave a wider range of values for annual evapotranspiration, ranging from 772.5 to 519.5 mm. Annual transpiration in the baseline period was 44% of mean annual evapotranspiration (Figure 4) but was greater under the B2 and A2 scenarios, reaching 45 and 46%, respectively. This suggests that future climate conditions may lead to greater losses of water by plant transpiration. However, the losses are not the consequence of greater vegetation growth since annual interception, determined as a function of plant leaf area, was 83.3 mm in scenario B2 and 77.0 mm in A2.

Figure 4.

Annual mean evapotranspiration for the baseline period (1961–1990) and for scenarios A2 and B2 (2071–2100)

Medeiros (2003) evaluated the sensitivity of surface runoff and soil moisture to climate change in the semi-arid Brazilian north-east, comparing the present scenario (with a reference quantity of CO2 emission) with future scenarios (double the reference quantity of CO2 emission) using outputs from two global circulation models. Estimates produced by the models showed increased potential evapotranspiration in both scenarios as a consequence of the probable temperature increase, but the direction of changes in actual evapotranspiration altered with changes in rainfall indices. On average, potential evapotranspiration showed a monthly increase of 10%, while rainfall diminished. In semi-arid regions where the evaporative demand is high, actual evapotranspiration becomes very sensitive to changes in rainfall patterns, becoming less as the soil dries out.

Reductions in rainfall depths relative to the present mean were also found by Salati et al. (2007) for the Amazonian region. Unlike what was found in the lysimeter simulation, typical of central Brazil, water balances given by the model Had CM3 for scenarios A2 and B2 estimated a reduction in excess water in the Amazon region of up to 73.4% in the period 2071–2100, when compared with water balance data for 1961–1990. Similar results were found by Liberato and Brito (2010) for western Amazônia.

Regarding surface runoff, both A2 and B2 scenarios generated depths of runoff greater than those in baseline. In general, the A2 scenario shows greater increases than B2 throughout the period simulated. In some years, however, surface runoff under B2 exceeded that of A2 (Figure 3). Under bare soil, mean annual surface runoff in the baseline period was 19.1 mm, increasing to 41.7 mm and 70.2 mm, respectively for scenarios B2 and A2. As percentages, these increases were 118.1% for scenario B2 and 268.1% for A2. In the case of maize cultivation, the increases in surface runoff were 103 and 152%, respectively. This shows that maize cultivation reduces conditions that give rise to surface runoff (Figure 5).

Figure 5.

Mean annual values for bare soil and maize cultivation

Surface runoff events are caused by extreme events in which daily rainfall is intense, and are the main components of flood hydrographs in drainage basins (Tucci, 2004). Under bare soil, the analysis of daily time series shows that annual maximum 1 d surface runoff in the A2 and B2 scenarios exceeds that of the baseline period which was 88.4 mm. In the B2 scenario, annual maximum 1 d surface runoff reached 182 mm, and in the A2 scenario, 351.7 mm.

The most destructive natural events in southern Brazil are related to intense rainfall events. In Brazilian metropolitan areas, such as Rio de Janeiro and São Paulo, where about 16% of the country's people are concentrated, the population is constantly subject to rainfall extremes leading to floods and land-slides, causing social, economic and environmental damage (Torres et al., 2009).

In 2008, drought in southern Brazil and northeastern Argentina afflicted soya and grain production, delaying the planting of wheat in some regions of Santa Catarina and Rio Grande do Sul and parts of Paraná. In the same year, an intense rainfall event struck the State of Santa Catarina between 22 and 24 November 2008, causing floods and land-slides. There were 135 fatalities, 32853 people were left homeless, 14 townships declared a state of public calamity with 63 emergency situations (CEDEC-SC, 2009). Maximum 1 d rainfall in this event was 250.9 mm, with 620.4 mm of rainfall accumulated over 6 d. The frequency of occurrence of 1- to 5 d rainfalls corresponded to a return period of more than 1000 years, denoting an event of exceptional severity at a daily time-scale (Pinheiro and Severo, 2010).

Figure 6 shows frequency distributions of the highest daily rainfalls. The highest daily rainfalls occurred under the A2 scenario, surpassing those of B2 which, in turn, exceeded those of the baseline period. Such rainfalls are responsible for generating surface runoff when their intensity exceeds soil infiltration capacity.

Figure 6.

Frequency of annual maximum daily rainfall for the baseline period, and for scenarios B2 and A2

These results are contrary to those obtained by Salati et al. (2010), who concluded that mean runoff in Brazilian rivers would diminish during the period 2011–2040. They found that mean flow in the Paraná River during this period would be reduced by about 20% relative to the period 1961–1990; in the São Francisco river basin, they estimated that the reduction would be 30%. Similarly, Medeiros (2003) found a substantial reduction in surface runoff in winter and spring months in the Paraguaçu River basin, under the scenario CCCII (of the Canadian Climate Center), with a 25% reduction in autumn and a reduction of up to 50% in summer, resulting in a reduction of about 40% in total annual runoff. Under the scenario UKHI (of the UK Meteorological Office), flow increased during autumn and was lower in the rest of the year. The reduction in surface runoff was the result of increased evapotranspiration caused by higher temperature. In this study, however, evapotranspiration does not change significantly between baseline period and the simulated scenarios.

This is in contrast with results given in the studies cited above, and can be explained by the different time-scales used in simulations. In earlier work, surface runoff was obtained from monthly and annual water balances as the difference between rainfall and evapotranspiration, with the latter estimated from meteorological variables, including monthly temperature. In this study, surface runoff was determined from the rate of infiltration into soil and rainfall intensity, so that evapotranspiration is controlled by meteorological variables, soil water content and the state of vegetal development at the soil surface.

Mean annual drainage from the soil profile under scenarios B2 and A2 is less than for the baseline period. Under bare soil conditions, the annual mean for 1961–1990 was 1033.9 mm; for the B2 scenario it was 923.5 mm and for A2, 932.8 mm. These values represent a reduction relative to baseline of about 10% when soil is bare; under maize cultivation, it was about 15%. For both scenarios, however, this reduction was not constant over the period simulated, being sometimes smaller and sometimes greater than in the baseline period. Under bare soil, the range of values was greater for the baseline period, with maximum and minimum values 1974.2 and 284.4 mm. Comparing the simulated periods for bare soil conditions, the B2 scenario showed values greater than those of A2, with maximum 1681.8 mm and minimum 470.3 mm. The annual maximum for the A2 scenario was 1508.6 mm. In a drainage basin, the simulated drainage would represent aquifer recharge which feeds to the river system during times of drought when water availability is commonly assessed. Natural water availability can be represented by mean flows and minimum flows, knowledge of which is highly important for adequate planning and proper allocation of available water resources, so as to minimize conflicts between different users (Novaes, 2005). In this context, the results show that water availability would be reduced under the scenarios A2 and B2.

Time series of water balance components were analysed at a daily time-scale. Table 2 shows characteristic mean, minimum and maximum observed and simulated daily values, and extreme events given by fitting an Extreme Value Type I (Gumbel) distribution (Naghettini and Pinto, 2007), for a 100 year return period. To use this distribution, a series of annual maximum daily values was formed, and from the parameters of the fitted Gumbel distribution, the frequency of occurrence of any extreme event can be estimated, or the inverse of that frequency corresponds to the event's return period.

Table 2. Characteristics of daily time-series for climate scenarios (mm)
Water balance componentBaselineScenario B2Scenario A2
 Min.MeanMax.TR = 100Min.MeanMax.TR = 100Min.MeanMax.TR = 100
  1. Min., minimum; Max., maximum; TR, return period.

Bare soil
Surface runoff0.00.172.675.10.00.1118.0124.70.00.2190.8165.0
Surface runoff0.

Mean daily rainfalls are higher in the baseline period than under scenarios A2 and B2. However the maximum daily rainfall under A2 is about 1.5 times greater than under B2, and is double the baseline value. Similar values are found for rainfalls with 100 year return periods. For evapotranspiration and drainage, values in the water balance do not show major differences between the summary values calculated from their respective time series. Thus the maximum daily rainfall values are reflected in surface runoff.

4. Conclusions

The results from this study show that the B2 and A2 scenarios have mean annual rainfall and evapotranspiration that are of the same order of magnitude as those in the baseline period. Mean annual drainage from both scenarios A2 and B2 was less than that for the baseline period. Maximum daily rainfalls under A2 are about 1.5 times greater than those for the B2 scenario and are double those of the baseline. Similar values are found for rainfalls with 100 year return periods. Regarding surface runoff, both scenarios A2 and B2 generate depths of surface runoff greater than those in the baseline period, with those under A2 being generally greater than those under B2 throughout the simulated period. Mean values of surface runoff are about 118% higher for B2, and about 268% higher, for A2, relative to baseline.

Thus, it is seen that the climatic scenarios with maximum and minimum carbon emissions (A2 and B2, respectively) do not significantly change the evapotranspiration and drainage flow components of water balance; characteristic features of their time series remain similar. However, maximum daily rainfalls give rise to increased surface runoff. Analysis of their frequency distributions shows that maximum daily rainfalls are higher, in both A2 and B2 scenarios, than in the baseline period. These more intense rainfalls give rise to greater surface runoff when rainfall intensity exceeds the capacity of water to infiltrate the soil. Increased surface flow implies a reduction in the infiltrated water which supplies evapotranspiration and drainage of water through the soil profile.

It is stressed that the results obtained are influenced by the time-scale used in the simulation models. In this study, the daily time interval used in simulations gave increased surface runoff, in contrast with results reported earlier where monthly or annual time-scales were used, and which showed effects on evapotranspiration.


We are grateful to MCT/FINEP/AÇÃO TRANSVERSAL—Climate and Weather Forecasting 04/2008, collaboration 1406/08, project 01.08.0568.00 for funding the research and to Dr José A. Marengo, of INPE, for providing PRECIS data and to CAPES for a grant to attend a Master's course.