Assessment of rainfall interception, soil moisture dynamics and seasonal headwaters in a micro-catchment of Western ghats

ABSTRACT The aim of this study is to assess the canopy rainfall interception, surface detention storage, and soil moisture fluctuation in a micro-catchment of a coffee plantation. Field measurements were carried out using 15 funnel-type ordinary rain gauges in an experimental area of 30 m × 30 m to measure the rainfall passing through the upper and lower canopy to estimate interception and throughfall. Rain gauges were placed at different locations to account for variability in net rainfall falling over the catchment. Throughfall measurements were compared with Gross rainfall measurements taken in an open space near to the plot. The overall ratio of the average of net ground precipitation to gross rainfall measured at the open space for the season was 0.82. Analysis for different rainfall intensities showed that lower-intensity rainfall contributes a higher fraction of the rainfall to interception loss. Site-specific field-based equations have been developed to convert the rainfall into throughfall for the agrarian basins of coffee plantations. Soil moisture depletion data revealed that the upper layer of the soil responds to rainfall and atmospheric demand quickly, compared to the deeper soil layers. Thick soil layers require a longer duration to reach field capacities and contribute significant amounts of the flow in seasonal/low-order streams. The discharge measurements after two months of continuous rainfall revealed that soils in plantation fields reach saturation during the monsoon season and release water to maintain flow in non-perennial streams after the monsoon season ends.


Introduction
The vegetation of the Western Ghats region in southern India is characterized by dense forest canopies supplied with abundant rainfall during the monsoon season.The annual rainfall ranges from 1500 mm to 6000 mm (Venkatesh, Nayak, Thomas, Jain, & Tyagi, 2021).Monsoon rainfall occurs primarily from June to September as the southwest monsoon, accounting for more than 80% of annual rainfall.High infiltration rates and subsurface delayed flow dominate the hydrological processes of the Western Ghats.Variations in topography, surface vegetation, and soil characteristics are significant for regional processes that influence soil moisture, evapotranspiration (ET), precipitation, and floods (Wood, Lettenmaier, & Zartarian, 1992).The soil water content fluctuates more near the surface than at deeper levels in response to rain events.Identical outcomes were also discovered during the summer monsoon season (Kurc & Small, 2004).In a humid forest, Sun, Onda, Kato, Otsuki, and Gomi (2014) showed that ET is controlled by soil water content up to a depth of 15 cm of soil.Evergreen trees have higher evaporation/transpiration losses than deciduous trees, which function as deep-rooted "water pumps" (Bruijnzeel, 1990;Hamilton & King, 1983).
Rainfall is the major input in the water balance in the areas of thick vegetation the initial losses are often ignored in the water balance studies.Forest catchment evaporates more water due to thicker canopies and not all the rainfall reaches the ground.The canopy holds a large amount of incoming rainfall and eventually allows the water to evaporate back into the atmosphere.Interception is the first substantial loss in wooded catchments compared to other types of catchments (Calder, 1977).Over twenty-five percent of the annual rainfall is intercepted by forest stands (Dingman, 2002) and contributes to climate change studies (Arnell, 2002).More intensive research on interception was conducted utilizing physically based models, namely Gash (1979) and Rutter, Kershaw, Robins, and Morton (1971).Interception studies have been carried out for a variety of vegetation types.However, there have been relatively few studies related to interception loss from the forests of the Western Ghats, particularly in the coffee plantation.Langkamp, Farnell, and Dalling (1982) investigated interception measures from Acacia planted plots and discovered that interception and stemflow (SF) were 11% and 16%, respectively.The high SF value was due to the upward-angled branches of the Acacia plantation.Forest interception accounts for 10-30% of rainfall, and in some cases, it may account for 50% of rainfall (Calder, 1990;Liu, 1997).Net rainfall and canopy interception are always used in hydrologic cycle modeling at the global, regional, and watershed scales (Ram´ırez & Senarath, 2000).
Indian coffee is grown in an area of 4150 km 2 .The states of Karnataka (57%) and Kerala (24%) account for approximately 81% of the plantation land in the Western Ghats region.The states of Karnataka and Kerala compose a significant portion of the Western Ghats and serve as the principal headwater source for runoff generation in the area.The main river systems that originate in this area have an impact on the economy of Karnataka state, including the Krishna, Kaveri, and Nethravathi.As a result, it is critical to investigate the varied hydrological properties.The rainfall-runoff response of a catchment forms the most fundamental aspect of hydrological modeling.In turn, the catchment response depends on important hydrological parameters such as rainfall, runoff, evapotranspiration, groundwater, surface water, and rainfall interception in thick vegetative areas.The most fundamental aspect of hydrological modeling is a catchment's rainfall-runoff response.The catchment response is then determined by essential hydrological parameters such as rainfall, runoff, evapotranspiration, groundwater, surface water, and rainfall interception in thickly vegetated areas.The accuracy of the model characterizing the catchment is dependent on the accuracy of the hydrological inputs.Headwaters originate in the upper portion of the hills of the Western Ghats and contribute significantly to river flow.The headwater flow of a 16-hectare catchment flowing into a surface storage (irrigation) tank was measured at the tank's outlet before it entered the stream.There are numerous storage tanks around the area.Tank discharge is seasonal, with flow confined to the monsoon season and lasting for several months after it ceases during the monsoon season.Many higher order streams are perennial streams whereas lower orders are seasonal.Tank outflow contribution is the major part of the flood during the monsoon season.Measurements under consideration are representatives of the major land use type in the regions of Western Ghats.The study area often consists of two thick canopies at two different levels that intercept large quantities of incoming rainfall, especially small intensities of rainfall.In the hydrological modeling often the rainfall interception depths are assumed which can affect the accuracy of hydrological parameters estimation.Therefore, interception depths were measured for 63 continuous daily rainfall events during the monsoon season.

Study area
The study area is located in the Central Coffee Research Institute, Balehonnuru in the Indian state of Karnataka, at 13°22'N and 75°28'E, at an altitude of 900 m above mean sea level.The catchment outlet is 832 meters above sea level and covers a surface area of 16 hectares.The tank outlet joins the perennial stream downstream.When the tank is full, the water spread area of the tank is roughly 5700 m 2 .
The discharge measurements at the tank outlet and rainfall interception were carried out on fully-grown coffee plants from early August to mid-November 2018.Following soil moisture measurements were taken at site 1 from mid-September 2018 to mid-February 2019, and from early November 2018 to May 2019.The average annual rainfall of the region is 2589 mm.Most of the rainfall occurs during the monsoon season from June to September and reduces during post-monsoon from October to November.Few spells of convective rainfall occur during the nonmonsoon season.Winter occurs from December to February followed by summer, or pre-monsoon, from March to May.The mean maximum and minimum temperatures are 27.5°C and 16.5°C, respectively.The catchment is composed of sandy clay loam soil, with infiltration rates of approximately 10 cm/hr measured by double-ring infiltration experiments at two places.Coffee, Black Pepper (Piper nigrum), and Arecanut (Areca catechu) are the dominant crops in the region.For a few centuries, coffee plantations have been the most common plantation crop in the southern region of Western Ghats.In many places the forested trees are undisturbed and the forest floor is cultivated with the coffee plantation.Forest trees protect coffee plants from high-intensity solar radiation during summer.The rate of interception is controlled by rainfall intensity, leaf area index (LAI), and meteorological conditions during and after rainfall (Muzylo et al., 2009).

Materials & methods
The experimental setup and plantation pattern are briefly discussed in this section.Forest trees and coffee plants form two canopies at different heights in coffee fields.Along with the upper forest tree canopies, matured coffee plants cover the majority of the forest floor.Two types of coffee plants namely coffee Arabica and Coffee Robusta are grown in the region.About 2500 Arabica plants and 1300 Robusta plants are grown per hectare.The pattern of the plantation is presented in Figure 1.Plants are evenly spaced apart to form square or rectangular planting patterns, with the plants spreading conically to almost semi-circularly.Plants that have reached maturity or shrubs with nearly conical shapes that are regularly spaced often leave an empty space between them.Rain gauges were set on these bare patches, between the coffee plants and beneath the canopy of the shade trees, to measure the wide range of interception inside the plot (two canopy layers) and beneath the canopy of the shade trees (single layer).Rainfall interception was assessed in coffee Robusta in a 30 m × 30 m area in the current study.Throughfall (TF) for coffee plants beneath forest trees was measured during 63 daily rainfall events during the peak of the monsoon season, with changing daily rainfall intensity.From the start of rainfall interception measurement to the end of the monsoon, 63 continuous daily rainfall events were considered.Throughout the monsoon season, the leaf area index of fully grown coffee plants remains almost unchanged.The focus of the study is under constant LAI to capture interception loss for varying (2 to 100 mm) rainfall depths.The intensity of rainfall plays a crucial role in interception loss.Higher rainfall intensity can saturate the foliage quickly, leading to an increased interception.By considering continuous rainfall events, it is possible to capture variations in rainfall intensity, which allows for a more comprehensive assessment of interception loss (Návar, 2020).Fifteen funnel-type rain gauges were used to capture TF.Gross rainfall was measured using a single rain gauge.They were positioned on the ground and raised to a height of about 30 cm to avoid the raindrop backsplash.The layout of the rain gauge setup is shown in Figure 2.
A portion of incoming rainfall that is intercepted, stored, and evaporated from the leaves and stems of vegetation is called rainfall interception loss.Excess rainfall passing through the canopy after canopy  storage reaches the ground is called throughfall.Gross rainfall is the precipitation that is falling over the canopy without any obstructions.Gross rainfall (GRF) is measured by placing the rain gauge above the canopy on a tower.But it is often expensive to set up a tower; therefore, it is measured on the ground in an open space near the interception measurement site.The ratio of TF to GRF represents the amount of precipitation reaching the ground surface passing through vegetation.The overall average of the ratios TF/GRF will give the average interception in the annual water balance estimates.
Stemflow is a film of water that flows over the surface of a tree stem before reaching the ground.Stemflow is measured using a semi-circular pipe wrapped around the stem of the tree which channelizes the water flowing over the tree to the container.In the process, water is evaporated from the stem and some of the small-intensity stemflow will not reach the ground surface.The sum of leaf interception loss and   stemflow gives the total canopy interception loss from the GRF.The interception equation is given by Where, I is the interception, P is gross precipitation, TF is the throughfall and SF is the stemflow.
The raingauge measures TF only at a point, catchdrains of 1 m × 0.24 m are used to measure the TF in a unit area.TF from the catchdrain is collected in a container on a daily basis.The meteorological station is situated at a distance of 400 m from the interception measurement site.TF passing through the two canopies namely the shading tree and the under-shade coffee plant canopy was measured using fifteen funnel-type ordinary rain gauges under the vegetation and the gross rainfall is measured outside the throughfall measuring area in an open space at a distance of about 50 m from the outer edge of the site.Through fall was measured only in the monsoon season.Since 80% of the year's precipitation falls during the monsoon season.The season consists of low-intensity long-duration rain (Putty, Javeed, Shishir Kumar, & Ashwini, 2021).On a daily scale, it almost completely includes all rainfall intensities from 2 to 100 mm as shown in Figure 3(b).Hence, it is assumed that measurements from a single season are sufficient.
The daily rainfall event of gross rainfall and throughfall was measured at the study site at 9 a.m.Statistical analysis was performed to determine the correlation between GRF and TF for different rainfall intensities.The rainfall events were categorized into four classes, namely daily rainfall depths of less than 10 mm (n = 13), 10 to 25 mm (n = 20), 25 to 50 mm (n = 15) and greater than 50 mm (n = 15).Crockford and Richardson (2000) discovered that defining interception mechanisms on an event basis is more difficult.In the presence of leaf area index (LAI) data, it is possible to model the rainfall interception events using Gash (1979) model and estimate the evaporation rates due to rainfall interception.But Gash's model is data intensive.Empirical equations generated from field observations gave reliable site-specific estimates of rainfall interception in the current study.Empirical equations are developed for each rainfall class interval to predict the net rainfall from measured gross rainfall.These empirical equations provide better estimations in the absence of field data.Figure 4 shows the empirical equations.
Net rainfall calculated from interception readings finally reaches the ground.To better understand how pore water moves into the soil strata around the root zone at different depths, granular matrix soil moisture sensors were installed at various depths to detect soil moisture storage.The soil moisture measurements were made in soil matric potential (SMP) units.Two sites were chosen to study the relationship between soil moisture loss and decreasing seasonal streamflow, as well as to use the remaining soil moisture for evapotranspiration and photosynthesis.The measurements were taken between early September 2018 and mid-February 2019 at site 1 and early November 2018 to May 2019 at site 2, which are 400 meters apart.Site 1 has a plant density of 1300 plants per hectare in the under canopy and 50 trees per hectare with a diameter of 1.5 m at breast height (DBH) and an average height of 25 m in the upper canopy.Roots extend much beyond 2 m depth in the natural forest, with the majority of roots found at 1.5 m depth and reported to be more densely clustered between 0.3 and 1.0 m (Kallarackal & Somen, 2008;Krishnaswamy et al., 2012).The nondestructive method resulted in a leaf area index of 2.3 for a coffee plant at the start of soil moisture measurement.Site 2 also has a plant density of 1300 plants per hectare in the under canopy and 160 trees per hectare in the upper canopy with about 0.5 m diameter at DBH.The upper canopy typically stands at a height of around 15 m.The site consists of predominantly Arecanut (Areca catechu) trees as shade trees.Blossom irrigation was applied in early January, resulting in an increase in soil moisture that remained for another three months at site 2, but site 1 was not irrigated.As a result, site 1 recorded a maximum measurable SMP of 200 earlier to site 2 (Figure 5) The runoff from the catchment was drained into a storage tank and streamflow was measured at a single outlet of the storage tank.Tank gets the flow from multiple inlets and drains through a single outlet.The area of the tank was found to be approximately 5780 m 2 when it is full.From 1 June to 4 August, which is considered the beginning of the monsoon season, the catchment received 1888 mm of rain over 63 rainy days.As a result, when stream flow measurement began, it was discovered that the study area was saturated.In the forest agrarian basins, the soils possess high infiltration rates, due to which the surface runoff is often absent except on the roads and pathways of the plantation.The runoff generally occurs in the form of interflow or pipe flow (Yadupathi Putty, 2009) before it reaches the stream.The presence of large-sized pores increases the permeability of forest soil, as has been noted repeatedly.This decreases surface flow and enhances water infiltration into soil profiles (Kirkby, 1978).The soils of the regions have the capacity to hold a large amount of water and release it to the streams when they are in a saturated condition.Soil textural classification was performed by sieve analysis.A pressure plate apparatus instrument was used to determine the water holding capacity of the soil taken at different depths at two different locations of the catchment.Comparing the saturated soil moisture depletion with the recession in streamflow by tracking the potential for soil moisture storage and, in some conditions, designating the excess precipitation as quick-flow runoff/pipe flow once the volume of possible soil moisture storage has been filled (Koster & Suarez, 1996;Liang & Xie, 2001;Markstrom et al., 2015) and streamflow measurements provide the knowledge about the contribution of saturated or pore water to the delayed flow.The soils of the regions have the capacity to hold large quantities of water and release it to the streams accounting for delayed flow in the streams or water source to the pipe flow.In this study, the point at which the significant streamflow recession begins and the complete cessation of the flow and their corresponding soil matric potential are discussed.Characteristics of the soil are determined using a pressure plate apparatus instrument to determine the water-holding capacity of the soil taken at different depths at two different locations of the catchment.The field capacity and wilting point are considered to be at a suction pressure of 33kPa and 1500kPa respectively.The amount of suction over which crops permanently wilt is termed the wilting point (Kirkham, 2014).The water-holding capacity of the soil is shown in Table 1.Similar experiments were carried out to determine the soil water retention curves accurately (Shwetha & Varija, 2015).The peak discharge increases when the catchment receives continuous rainfall or the antecedent moisture condition is at saturation.

Results and discussion
The analysis of the 15 rain gauges located under the canopy to measure rainfall interception shows that only six rain gauges (rain gauges 1, 3, 6, 9, 13 and 14) contributed to much of the total interception.The remaining nine rain gauges (rain gauges 2, 4, 5, 7, 8, 10, 11, 12, and 15) were found to contribute less to the overall interception.The highest interception of the rainfall was found at rain gauge number 6 which was placed well inside a thick coffee plant along the side of the stem of the plant.The coffee plant near rain gauge 6 was also covered under the shading tree.An interception of 64% with TF to GRF ratio was 0.36 at this rain gauge.The TF was decreased at all rainfall intensities.It is obvious that rainwater passing the two thick canopies intercepts a significant amount of rain and does not allow it to pass through.Only 756 mm of the 2126 mm GRF was allowed to reach the ground surface.Table 2 depicts the rainfall totals obtained from many rain gauges positioned at various points on the 30 m × 30 m study plot, as well as the TF depths and their variability for additional rain gauge positions 1 to 5 and 7 to 15.The ratios greater than one indicate that TF is higher than the gross rainfall.This is sometimes referred to as negative interception, and it was documented in studies conducted by Crockford and Richardson (2000), Langkamp, Farnell, and Dalling (1982), and Lloyd and Marques (1988) for a variety of forest types and climates.In contrast to findings in previous studies, where interception loss was reported to be 24.52% in comparable forest regions (Cleophas et al., 2022), the total average of TF found in the present study was 82%, meaning that 18% of the rainfall is intercepted.Figure 3(a) illustrates that low rainfall depths are used to saturate the canopy when the total TF to GRF ratio is less than 0.6, resulting in substantial interception loss and low throughfall.Most TF/GRF ratios are above 0.6 and below 1, which reflects a mix of varying interception depths and rainfall depths.Additionally, the overall average of 0.82 falls in this range.TF/GRF ratio close to 1 and greater than 1, indicates that the TF values are very close to the GRF.This indicates that interception losses are higher for lower rainfall depths.
Throughfall measured by catchdrain to GRF ratios was measured for the subsequent year in the same site, as shown in Figure 3(c).As can be seen from the same figure, the dense coffee plant intercepted more rainfall than the sparse coffee plant.The empirical equation for the relationship between the gross rainfall and net  rainfall for all the four classes are shown in Figure 4.
The same empirical equations are used to assess rainfall-runoff relationships once the net rainfall has been determined.Soil moisture sensors were used to study processes like field capacity and wilting point at various soil depths.The soil strata have reached their field capacity when the SMP is 33 kPa.Up until the field capacity is reached, saturated water significantly contributes to streamflow, a characteristic of a nonperennial stream in the Western Ghats area.After canopy interception, rain is seen falling over litter fall, which is known as litter fall interception.
Although litter fall interception has not been included in this study, the actual net rainfall that reaches the soil is significantly lower than the canopy interception estimates given in the present study.Net precipitation that reaches the soil improves soil water storage and contributes significantly to pipe-flow during the peak monsoon season.When the soil gets saturated, soil storage increases which directly affects streamflow.
Figure 5 demonstrates that in Site 1, pipe flow continues until a suction of 10 kPa is reached (16 November 2018,) contributing to seasonal streams.Further reduction in soil water content upto field capacity (33kPa) will contribute to evapotranspiration, groundwater recharge, and subsurface delayed flow into the perennial streams.Flow in the perennial streams after the catchment reaches the field capacity is mainly from the baseflow.Soil moisture measurements were made using Granular matrix-type soil moisture sensors from early September 2018 to mid-February 2019 in site 1 and early November 2018 to May 2019 in site 2.
Soil sensors have the capacity to record matric potentials up to 200 kPa.The variations in soil matric potentials in both plots are shown in Figure 5.In both sites, the results showed that the upper layers quickly respond to irrigation or rainfall as compared to the deeper layers of soil.At site 1, during the dry season, the non-irrigated plants demonstrated high-stress behavior as compared to the irrigated plants at site 2. To reach field capacity at site 1, the top 15 cm and subsequent deeper soil layers required 39 and 82 (average of 4 layers) days, respectively, but at site 2, the top layer and deeper layers took 27 and 49 days, respectively.According to Wahren, Feger, Schwarzel, and Munch (2009), similar results in forest soil took a longer duration to attain field capacity than in agricultural soils.At both sites, the top soil layer dries faster than deeper layers.The difference in vegetative growth between site 1 and site 2 May account for site 1's longer time to attain field capacity (82 days) than site 2(49 days).Although the under canopy (coffee plants) at both sites is similar in density, site 2's top canopy was found to be denser than site 1's.The shading tree/ forest tree canopy in site 2 utilizes more water from the soil storage as compared to site 1.Hence it can be concluded that the density of the vegetation plays an important role in regulating soil moisture storage.
Since the response to rainfall takes longer to reflect in runoff due to the large spreading area of the tank, the discharge measured at the tank outlet was averaged over daily intervals.According to the analysis of the streamflow (Table 3), it is speculated that the groundwater and soil water storage may be responsible for the  runoff exceeding rainfall.It is well known that during the peak of the monsoon, subsurface flow/pipe-flow is the predominant streamflow mechanism in the Western Ghats, which contributes to both perennial and seasonal streams.Additionally, in the watersheds, irrigation tanks act as surface storage and temporary flow regulators.Whether they are man-made or natural, the usefulness of the tanks as flood control structures depends on their size.Seasonal streams run out of water shortly after the monsoon season ends (Figure 6).

Summary and conclusions
The proposed approach may be applied in comparable study sites and is essential for understanding the soil moisture dynamics and interception during rainfall.
The following conclusions can be drawn from the current study: • It has been found that rainfall interceptions are highest when the TF/GRF ratio is smaller than 0.6.In interception experiments, TF/GRF ratios greater than 1 known as negative interception are frequently reported.To calculate net rainfall from gross rainfall, site-specific empirical equations are constructed.More research is required at several sites in order to develop generic empirical equations at the regional level.Better throughfall estimations will be obtained with additional litter fall interception measurements along with canopy interception.• Headwater streams depend on the release of stored soil moisture to continue flowing after monsoon rains have stopped.SMP measurements indicated that when the soil layers reached the suction pressure of 10 kPa, the flow in the first-order stream ceased.Two upper canopies with varying densities take different amounts of time to reach field capacity, indicating that the upper canopy has a significant role in water utilization.
• Even though this study does not account for the effect of microclimatic conditions, it is useful for watershed management and climate change studies.Capturing the various possible interception rates within the study site is assumed to be the best representation of actual interception.Furthermore, irrigations can be scheduled based on soil moisture fluctuation data.

Figure 2 .
Figure 2. Layout of rainfall interception measurement site.

Figure 3a .
Figure 3a.Throughfall to gross rainfall ratios and daily rainfall depths.

Figure 3b .
Figure 3b.Gross rainfall depths arranged in ascending order during interception measurements.

Figure 4 .
Figure 4. Empirical equations for gross rainfall to throughfall/net rainfall ratios at different rainfall (daily) intensities.

Table 1 .
Volumetric water content at various soil layers.

Table 2 .
Throughfall to gross rainfall ratios and overall rainfall depths.

Table 3 .
Rainfall and stream flow discharge from irrigation tank outlet.