Hydrology of Warm Humid Regions (Proceedings of the Yokohama Symposium, July 1993). lAHSPu%.no.216, 1993. V 43

Hydrological processes in the Serra do Mar, Sâo Paulo, Brazil

MOTOHISA FUJIEDA, TETSUYA KUDOH & YUKIO MASHIMA Forestry and Forest Products Research Institute, Ibaraki, Tsukuba, 305, Japan

Abstract This paper deals with some results of a watershed experiment at the Cunha Forest Hydrologie Laboratory in the Serra do Mar, Sâo Paulo, Brazil. Streamgauging from small forested watersheds covered with the Mata Atlantica commenced in 1982. Measurement of crown interception and surface runoff from a hillslope were also conducted to quantify individual components of hydrological processes. Streamflow as well as crown interception and surface runoff data were analysed to better understand the hydrological processes and the effect of forest cover on these processes in headwater areas.

INTRODUCTION

The forested area of Sâo Paulo State, the Federative Republic of Brazil has decreased from 81.3% to 8.3% with an expansion of agriculture in the past century. Natural forests are only found in the Serra do Mar (Institute Florestal, 1980). The earliest developed coffee plantations are located in Paraiba Valley, in the northeast portion of the state. Extensive farming in this region has caused environmental problems such as soil erosion, sedimentation in reservoirs and flood. The state government is implementing soil and water conservation projects by establishing forests and by protecting natural forests in the Serra do Mar. At the same time, a research project that supplies scientific data for the policy of forest management is strongly required. Under these circumstances the forest hydrology research project was initiated in 1979 by the Forestry Institute of Sâo Paulo State and the Forestry and Forest Products Research Institute of Japan through assistance by the Japan International Cooperation Agency (JICA, 1986). The project established the Cunha Forest Hydrologie Laboratory, located in the headwaters of the Paraiba River, for the purpose of evaluating the hydrological processes and water resources in the subtropical forested basin. The laboratory consists of two gauging stations, three lysimeters, three experimental plots for surface runoff and sediment yield and a meteorological station. The object of this report is to summarize some results from a watershed experiment with the purpose of understanding the hydrological processes and the effect of forest vegetation on these processes in headwater areas.

SITE DESCRIPTION

The laboratory is located in the Serra do Mar of Sâo Paulo State, latitude 23°13'S, Motohisa Fujieda et al.

Fig. 1 Location of the Cunha Forest Hydrologie Laboratory. longitude 45°01'W (Fig. 1). Under Koppen's system, the climate is classified as Cwa type, which is a humid subtropical climate that has precipitation in all seasons with maximum in the summer. The region is covered with the Mata Atlantica, a mountain rainforest that is found from 800-1500 m altitudes in the Serra do Mar. The Mata Atlantica as well as the Mata Amazonica (Amazon forest) is a basic forest type in Brazil. The mean annual rainfall at the meteorological station (1045 m) is 2391 mm, and 71% of annual rainfall occurs during the rainy season (October to March). The mean annual temperature is 16.5°C, however, the average maximum and minimum temperature in winter are 20.6°C and 6.6°C, and in summer 26.3°C and 16.0°C, respectively. The bedrock in the region is gneiss and crystalline schist of Precambrian age. Hillslope soils are oxisols with thin (up to 30 cm) A-horizons containing organic matter and deeper (70 to 200 cm) B-horizons with high levels of sesquioxides saprolite underlying the soil mantle. The soil texture is classified as sandy clay. The hydraulic conductivity values of the A and B-horizon range from 10"2 to 10~3 (cm s"1) and from 10"3 to 10"5 (cm/sec), respectively. Table 1 shows some physiographic characteristics of experimental watersheds.

Table 1 Some physiographic characteristics.

Watershed B D Drainage area (ha) 36.68 56.04 Elevation (m) 1025-1199 1048-1222 Mean slope (%) 25.1 18.4 Length of channel (m) 920 1260 Drainage density 0.433 0.353 Hydrological processes in the Serra do Mar, Sâo Paulo 45

STUDY METHOD

A forest watershed experiment is required to quantify individual components of the hydrological cycle before watershed treatments (watershed standardization). The meteorological measurements and hydrological measurements such as crown interception, surface runoff from a hillslope and streamfiow were implemented for the standardization. Details of experimental facilities and hydrological measurements at the laboratory are given by JICA (1986). The separation of storm flow from storm hydrograph (basefiow separation) were conducted by drawing a straight line from the rising point to the characteristic point of falling limb (Chow, 1964). The basefiow separation was conducted for every storm hydrograph during the period. Monthly storm flow is obtained by adding storm flow estimated by the basefiow separation. Monthly basefiow is obtained by subtracting monthly streamfiow from monthly storm flow. The monthly storm flow ratio is calculated by dividing monthly storm flow by monthly streamfiow. The beginning of the water year is October and the 1983 water year is defined as the period 1 October 1982 through 30 September 1983.

RESULTS AND DISCUSSION

Crown interception

Measurement of crown interception was conducted from 9 January 1983 to 20 January 1984, 51 storms were measured. The relationships between gross rainfall (Pg) and throughfall (P,), stemflow (Ps) and crown interception (7C) are expressed by the following equations:

Pt = -0.021 + 0.807 Pg (r = 0.9987)

Ps = -0.051 + 0.012 Pg (r = 0.8584)

Ic = 0.141 + 0.180 Pg (r = 0.9744)

Table 2 contains results of crown interception on several types of tropical forest in Brazil (Cicco et al., 1988). Crown interception varies from 12.4 to 27.3% depending

Table 2 Crown interception in the tropical forest, Brazil.

Place Forest type P, Ps Ie Manaus, AZ (1982) Floresta de Terra Firme 81.8 18.2 Manaus, AZ (1982) Floresta de Terra Firme 77.7 0.3 22.0 Vicosa, MG (1983) Floresta Natural Secundaria 87.4 0.2 12.4 Rio de Janeiro, RJ (1986) Reflorestament 83.0 17.0 Agudas, SP (1983) Cerradao 72.7 27.3 Sâo Moronel, SP (1985) Cerradao 80.5 2.9 16.6 Cunha, SP (1988) Floresta Natural Secundaria 80.7 1.1 18.2 AZ = Amazonas; MG = Minas Gérais; RJ = Rio de Janeiro; SP = Sâo Paulo State. 46 Motohisa Fujieda et al. on forest type. Crown interception at Cunha, 18.2% of annual rainfall, is intermediate with respect to these reported values and is nearly equal to that of the Floresta de Terra Firme (a kind of Amazon forest). Since measurements at Cunha were conducted in natural forests, the difference in catch among 16 throughfall gauges varies highly. The total throughfall caught by each of gauges during the period ranges from 68 to 98% of gross rainfall (mean 0.828; S.D. 0.096). Hewlett (1982), however, points out that about 10 throughfall gauges are required to equal the mean accuracy of a single gross raingauge in the open. Therefore, the results measured at the experiment plot are believed to be of similar accuracy compared with the results shown in Table 2.

Surface runoff from a hillslope

Surface runoff from an experimental plot (area 600 m2; slope 16%) was measured from November 1982 to October 1983. The total surface runoff during the period is only 0.61% of total rainfall. Monthly surface runoff ratio varies from 0.0 to 1.55% depending on rainfall and soil moisture content. The highest surface runoff ratio (4.0%) occurred in the rainy season under a high rainfall intensity. The minimum rainfall intensity required for the generation of surface runoff on the plot may be more than 10 mm h"1. Leite (1985) measured overland-flow and interflow from Alfisol soil planted with cacao in Bahia, Brazil. He reported that the highest overland-flow volumes represented 24% of the rainfall, however, the average overland-flow is 1.1% of annual rainfall. Averaged slope in his plots was 23% to 26%. These results indicate that the surface runoff ratio of Cunha is about half of that at Bahia. The difference mainly results from the differences of the physical properties of soils, such as soil texture and permeability. The result of Cunha suggests that surface runoff in the undisturbed forest of the experimental watershed rarely occurs and that almost all the water which has fallen to the forest floor enters the soil.

Streamflow characteristic

The streamgauging at D-watershed has been in operation since March 1982, and daily streamflow records for eight water years (1982-1990) have been analysed. We would first like to describe the streamflow regime of the watersheds. Figure 2 shows the flow duration curves of average daily streamflow during the period, including B-watershed for three water years (1988-1990). The streamflow of D-watershed shows a rapid decrease till 10% of flow, and then describes a gentle straight line on semilogarithmic paper. It is estimated that the dominant component of streamflow till 10% may be storm flow, and after that the baseflow component may become dominant. The mean daily streamflow for the period is 4.51 mm, which is equivalent to 34.5% of time flow. The medium daily streamflow, 50% of time flow, is 3.66 mm. The hydrological regime of B-watershed is found to be the same as that of the D-watershed. Figure 3 shows the relations between monthly rainfall and storm flow. Monthly storm flow gradually increases with amount of rainfall. The monthly storm flow ratios during the rainy season range from 0.018 to 0.579 (mean; 0.095) and those during the Hydrological processes in the Serra do Mar, Sâo Paulo 47

lOO.Or

50.0

Mean Median D-Watershed 4.51 3.66 Watershed 3.83 2.92

50 100 Percent of Tiae Streaaflow Equalled or Exceeded Fig. 2 Flow frequency distribution for two watersheds. dry season range from 0.00 to 0.128 (mean; 0.044). The ratios during the rainy season shows large variation, however, the ratios less than 300 mm monthly rainfall are almost below 0.10. According to the variable source area conception, storm flow is usually produced from the source area, which are channel and spring heads and seepage from hillslopes and swamps that are quickly saturated by rainfall (Hibbert & Troendle, 1987). At D-watershed, the wetlands along streams comprise about 2.5 ha, equivalent to 4.5% of the watershed area. Since the area is usually saturated, it will become the source area after rainfall. Thus, the monthly storm flow ratio under the small monthly rainfall will be almost the same as the ratio of saturated area to drainage area. This means rain falling on hillslopes and riparian except wetlands enters the soil and almost all of it may become soil moisture and groundwater recharge. Because of reduced saturated areas in the dry season, the mean ratio of the dry season

0 100 200 300 400 500 Monthly Rainfal1 (mm) Fig. 3 The relations between monthly rainfall and storm flow at D-watershed. 48 Motohisa Fujieda et al.

decreases to half that of the rainy season. Figure 4 shows the relations between average monthly rainfall and streamflow during period. The maximum monthly streamflow appears during the January—March recharge period, and the minimum monthly streamflow appears in late winter or early spring, however, the most noticeable feature is that monthly streamflow during the May—August dry season is much more than monthly rainfall during this period. The seasonal changes follow a counter-clockwise loop from October to September. The loop is caused by an effect of basin storage within the watershed. If the storage capacity of a basin was small Fig. 4 would describe a straight line similar to the relations of monthly rainfall and storm flow shown in Fig. 3. The result indicates that a part of the rainfall in the rainy season may be stored in the regolith as soil moisture and groundwater storage and later supplied to streams as baseflow in the dry season. This temporal release of baseflow (sustained flow of streams) is one of most important hydrological features controlled by such basin characteristics as physical properties of soil, depth of soil mantle and vegetative cover. Especially, the vegetative cover protects the soil surface from a shower in the rainy season and maintains infiltration capacity of soil. The magnitude of the loop may depend chiefly on a scale of riparian area that groundwater and soil moisture are stored. It is estimated that the riparian areas may be the place of storm flow production as well as that of groundwater storage at basin. They are typically located at elevations of 1000-1100 m in the mountain plateaus of Paraibuna basin. Their evaluation would help to understand the hydrological characteristics in the Serra do Mar.

300 CUNHA ® D-watershed • B-watershed

to (U

0 100 200 300 400 Monthly Rainfal1(mm) Fig. 4 The relations between mean monthly rainfall and streamflow.

Hydrological processes

Table 3 shows annual rainfall, streamflow and losses during the period. Mean annual storm flow is only 11.1% of mean annual rainfall. Mean annual baseflow is 59.0% of the rainfall and accounts for almost 84% of the total streamflow. Hewlett (1982) suggests that baseflow in upland streams with healthy forest cover contribute about Hydrological processes in the Serra do Mar, Sào Paulo 49

Table 3 Annual water budget at D-watershed.

Water year Annual rainfall Annual storm Annual baseflow Annual streamflow Annual losses (mm) flow (mm) (mm) (mm) (mm)

1983 2583.5 197.8 1630.5 1828.3 758.5 1984 1852.0 148.8 1197.2 1346.0 506.0 1985 3113.3 748.9 1973.5 2722.4 390.9 1986 2378.6 184.9 998.2 1183.1 1195.5 1987 2602.3 230.6 1734.6 1965.2 637.1 1988 2191.2 221.9 1370.4 1592.3 598.9 1989 2466.1 199.3 1326.2 1525.5 940.6 1990 1867.3 187.9 1010.9 1198.8 668.5

Mean 2381.8 265.00 1405.2 1670.2 711.6 Percent 100.0 11.1 59.0 70.1 29.9

85 % of total streamflow. Mean annual évapotranspiration is estimated to be about 30% of the rainfall, 712.0 mm year"1 in water depth equivalent, by the hydrological budget. On the other hand, a potential évapotranspiration at the laboratory calculated by the HAMON method ranges from 700 to 800 mm year"1. Mean annual évapotranspiration calculated by the hydrological budget indicates it is the same as potential estimated évapotranspiration. This result suggests that hydrological measurements have been conducted with high accuracy. The summary of hydrological processes at D-watershed is as follows. About 18% of annual rainfall is intercepted by forest cover and returns to the atmosphere. The rainfall arriving on the forest floor infiltrates into the surface soil and remains in the soil to feed baseflow or évapotranspiration. Surface runoff is a rare occurrence on forested hillslopes even grassed ones, and constitutes only 0.6% of annual rainfall. Storm flow is generated from wetland source areas adjacent to streams and from seepage from hillslope; however, the total volume of storm flow is only 11 % of annual rainfall. About 61% of annual rainfall is stored in the soil mantle and flows via subsurface routes to streams as baseflow throughout the year. Evapotranspiration from the soil has been estimated as 12% of annual rainfall by the hydrological budget. Based on the above results, the hydrological processes of D-watershed are schematically illustrated in Fig. 5. The figure indicates that the dominant water movements in the watershed are the processes of rainfall, interception storage, soil moisture storage, groundwater storage and baseflow. It is assumed that the capacity of soil moisture and groundwater storage at the watershed depends on the characteristic of riparian areas, such as expanse and depth of sediment. The difference in riparian areas between D and B-watersheds might account for the differences in the components of streamflow shown in Tables 3 and 4. The baseflow of D-watershed with a large riparian area and deeper soils is more sustained than that of B-watershed with a small riparian area and shallow soils.

CONCLUSION

The results of hydrological measurements at the laboratory suggest that much water falling on the Serra do Mar may be stored in the deep regolith and later yield to small 50 Motohisa Fujieda et al.

Rainfall 100 X

Riparian flillslope

Interception Storage — Interception 17 X

Steifloi Throughfall 1 X 77 X Evapotrans­ Net Rainfall piration 78 30 X Surface Runoff 5 X Soil Moisture Evapotranspi ration Storage froi Soil 13 X Subsurface Runoff

<{ 6 X )—J Groundwater Storage

Direct Runoff Base Runoff 11 X 59 X 1

Streaaflow 70 X

Fig. 5 The hydrological processes at D-watershed. streams throughout the year. This means the region is a valuable area of the headwaters of the Sâo Paulo metropolitan area and Paraiba Valley. Therefore, the conservation of the soil surface and vegetative cover of the headwaters by means of expanding the protection areas and forests in the Serra do Mar would be an

Table 4 Annual water budget at B-watershed.

Water year Annual rainfall Annual storm Annual baseflow Annual streamflow Annual losses (mm) flow (mm) (mm) (mm) (mm)

1988 2000.1 342.8 1125.6 1468.4 531.7 1989 2165.8 376.9 1136.8 1513.7 652.1 1990 1756.3 358.0 836.7 1194.9 561.6

Mean 1974.1 359.2 1033.0 1392.2 581.9 Percent 100.0 18.2 52.3 70.5 29.5 Hydrological processes in the Serra do Mar, Sâo Paulo 51

appropriate way of dealing with the problems of soil and water conservation in the valley.

Acknowledgements The authors wish to thank the Department of Forestry Development Cooperation of the Japan International Cooperation Agency for their support and encouragement, and to express their appreciation to the Forestry Institute of Sâo Paulo State and the Forestry and Forest Products Research Institute of Japan.

REFERENCES

Chow, V. T. (1964) Handbook of Applied Hydrology, 14.8-14.13. McGraw-Hill, New York. Cicco, V., Arcova, F. C. & Fujieda, M. (1988) Interceptacao das chuvas por floresta natural secundaria de Mata Atlantica, Sâo Paulo. Silvic. S. Paulo 20/22, 25-30. Hewlett, J. D. (1982) Principles of Forest Hydrology. The University of Georgia Press, Athens, Georgia, USA. PP. Hibbert, A. R. & Troendle, C. A. (1987) Streamflow by variable source area. Chapter 8, in: Forest Hydrology and Ecology at Coweeta. Springer-Verlag, New York. Institute Florestal (1980) Institute Florestal de Sâo Paulo. Brazil. JICA (1986) Synthetic report of the Japanese technical cooperation project for the forestry research in Sâo Paulo, Brazil. Leite, J. O. (1985) Interflow, overland-flow and leaching of natural nutrients on an Alfisol slope of southern Bahia, Brazil. J. Hydrol. 80, 77-79.