Estimating Rainfall Erosivity of the Revised Universal Soil Loss Equation from daily rainfall depth in Krishanagiri Watershed region

of Tamil Nadu, India

Arun Babu Elangovan1+ and Ravichandran Seetharaman2 1 Assistant Professor and 2 Professor, Centre for Water Resources, Anna University, Chennai, India

Abstract. The Present paper contributes to the quantitative assessment of Revised Universal Soil Loss Equation’s Rainfall Erosivity (R) from daily rainfall depth in a data scarce watershed region. Rainfall depth for every 15 minutes from self recording rainguages was measured at four locations in the watershed and a simple model was established between rainfall erosivity and depth of rainfall. 163 events that contribute to soil erosion were identified and regression analysis was carried out using linear, logarithmic, exponential, power, polynomial and quadratic methods. It was found that a power function gave the highest coefficient of determination when compared with five other simple regression analysis of the Rainfall erosivity (MJ mm/ha-event) versus the depth of rainfall. This regression equation had a 0.706 coefficient of determination statistically, and hence it can be used to estimate the rainfall erosivity of the other meteorological stations. To examine the validity of the equation developed, a further test of the R was carried out for another station which has observed R. The difference between the estimated and the observed erosivity is the experimental error. Statistics of the experimental errors are reported as the Mean Absolute Error (34) and the Root Mean Square error (84) for the validation station. These results indicate that the new procedure can give reasonably accurate results for the entire Krishnagiri watershed. Rainfall erosivity of the remaining stations which measures only depth was estimated using the developed power function. Then the spatial pattern of R map was generated using different interpolation techniques in Arc GIS 9.0 which will aid in identifying the vulnerable areas of soil erosion.

Keywords: RUSLE, GIS, Rainfall Depth, Rainfall Erosivity

1. Introduction Rainfall Erosivity (R- factor) is the basic and important factor in the assessment of soil erosion in the

mathematical model, Universal Soil Loss Equation USLE and its revised form RUSLE. Erosivity is the potential capacity of the raindrops to cause detachment of the soil particles from its location and it depends on rainfall intensity its recurrence. Hence it is important to accurately estimate erosivity for quantitavite estimation of soil erosion. The R-factor is defined as the mean annual sum of individual storm erosion index values, EI30, where E is the total storm kinetic energy and I30 is the maximum rainfall intensity in 30 minutes. To compute storm EI30 values, continuous rainfall intensity data are needed. Wischmeier and Smith (1978) recommended that at least 20 years of pluviograph data be used to accommodate the natural climatic variations. Self recording rain gauges are usually expensive, so rainfall erosivity is typically known only at a limited number of locations. Hence an attempt is made in this work to establish a simplified relationship between erosivity and depth of rainfall as depth of rainfall can be measured at all locations.

2. Study area Krishnagiri watershed is in the northwest of Tamil Nadu state of India and located at the latitudes of

12.275 N -13.125 N and longitude of 77.625 E – 78.375E (Figure 1) in the tropical hot zone with a + Corresponding author. Tel.: + 91 44 2235 7469; fax: +91 44 2235 1075 E-mail address: e_arunbabu@yahoo.co.in / arunbabu@annauniv.edu

48

2011 International Conference on Environmental and Computer Science IPCBEE vol.19(2011) © (2011) IACSIT Press, Singapore

maximum temperature 34 C to 37 C and minimum temperature 22 C to 24 C and a yearly average precipitation of mm. It has a total area of 3000 km2 with elevation varies from 540 m to 1200m above mean sea level and elevation decreases from Northwest to southeast direction. Land use in this watershed includes agricultural plantation, builtup area, forest, waterbody, pasture land, industries and hills. Ponniyar river is the major river flows through the watershed from Northwest to southeast direction drains the water to krishnagiri dam which caters an ayacut of 3642 ha. Figure 2 shows subwatersheds namely upper ponniayar, Chinnar, Sulagiri Chinnar, Markandanadhi, Nachikuppam, Veppanapalli, Middle ponnaiyar and lower ponnaiyar.

Fig 1: Krishnagiri Watershed with raingauge locations Fig 2: Subwatersheds

2.1. Data Used Basic data required for this study include the depth of rainfall from the whole watershed and its

surrounding area. This watershed includes regions from three different states namely Tamilnadu, Karnataka and Andrapradesh. 24 raingauges in and around the watershed were identified from the listings of Indian Metereological Departmentt. Daily rainfall depths for the identified stations were collected from the respective state data centre for a period from 2005 to 2009. Veppanapalli and Sulagiri are the two locations were self recording rainguages were installed to monitor intensity of rainfall.

2.2. Methods Out of 24 stations in the watershed, four stations are installed with self recording gauges from which R

factor can be estimated as R = EI30 where E is the total energy for a storm and I30 is the maximum 30 minute intensity. Total energy for a storm is computed from

where e = unit energy (energy per unit of rainfall), ΔV = rainfall amount for the kth period k = an index for periods during a rain storm where intensity can be considered to be constant m = number of periods. Unit energy is computed from : e = 0.29[1− 0.72 exp(− 0.082i)]

where unit energy e has units of MJ/(ha·mm) and i = rainfall intensity (mm/h). The next step is to determine the maximum 30-minute intensity I30. Maximum 30-minute intensity is the

average intensity for the continuous 30 minutes with the maximum rainfall.

49

The Rainfall intensity obtained from the automatic rainguage from three stations namely veppanapalli, soolagiri and KRP dam was used to calibrate a simple model. The developed model was validated from the observed R from melumalai station. Spatial variation of R was developed using the various interpolation techniques available in GIS environment. This can help the planners and soil conservationist to develop strategies for conservation measures within the watershed.

3. Results and Discussion The results obtained from this work are reported in the form of tables and graphs. Fig 5.1 shows the

variation of rainfall over the entire watershed from the year 2005 to 2009. The graph clearly shows that regions in Karnataka receive high rainfall over the regions in Tamil Nadu. These stations are located in the upstream side of the watershed on a higher elevation.

Fig 3: Spatial and temporal variation of rainfall depth

A B

Fig 4: A. Regression relationship between rainfall erosivity and Rainfall depth B. Scatter plot between EI30 observed and predicted at the validating station

Rainfall charts from the stations veppanapalli, KRP dam and soolagiri were analysed for identifying events of rainfall that exceeds a depth of 12.6mm. As the energy for the droplet of water to cause erosion is available only when the depth exceeds 12.6mm. There are 163 daily rainfall days which has the capability to erode the soil. Each event was taken for estimating the Rainfall erosivity factor. The recorded chart produced rainfall depth for every 15 min from 8:30 AM to next day 8:30 AM. The depth for every 30 min was tabulated and the intensity for every 30 min was calculated. Unit energy for the rainfall was estimated using

50

the intensity. The product of Energy and the maximum 30 minute intensity gives the R factor for the day considered. Monthly R factor was estimated by the sum of the daily R factor and the annual R factor was estimated by the sum of monthly rainfall R. Regression analysis was done for the depth and R factor using linear, logarithmic, exponential, power, polynomial, quadratic methods. From the regression analysis a power function gave the highest coefficient of determination when compared with five other simple regression analysis of the R-factor (MJ mm/ha-event) versus the depth of rainfall. The regression equation had a 0.706 coefficient of determination statistically (Fig.4 A), so it can be used to estimate the rainfall erosivity of the other meteorological stations.The simple power function developed to estimate the R factor from the depth of rainfall is given as R = 0.193 P 1.895 where R is the rainfall erosivity factor and P is the depth of rainfall.

In order to examine whether equation calibrated by using three rainfall station point data, can be applied to determine the R factor for other stations in the watershed. A further test of the EI30 model was carried out using data from melumalai station. This station has an observed data for 44 events of rainfall that exceeds 12.6 mm depth. The calibrated equation was used to predict the R factor from the depth of rainfall. Fig 4 B shows the 1:1 relationships between measured and estimated values. The difference between the estimated and the measured actual monthly erosivity index (EI30-aj ) is the experimental error

From the errors the Mean Absolute Error and Root Mean Square Errors can be computed as follows

Mean Absolute Error

Root Mean Square Error

Statistics of the experimental errors are reported as the Mean Absolute Error for the validation station

was 34 and the Root Mean Square was reported as 84. Table 5.1 clearly gives an idea that the errors for the predicted value was negative as the developed equation under predict. This can be rectified when more variables were added to the equation. These tests indicate that the new procedure can give reasonably accurate results for the Krishnagiri watershed.

Table 1: EI30 Measured, predicted and the error for the validation station

Rainfall events

EI30 Measured

EI30 Predicted Error

1 7.08 24.92 -17.84 2 8.63 26.76 -18.13 3 17.94 49.12 -31.19 4 18.43 24.92 -6.49 5 20.14 26.76 -6.63 6 20.14 56.90 -36.76 7 23.65 42.82 -19.17 8 23.80 30.64 -6.84 9 23.80 30.64 -6.84 10 25.76 32.68 -6.92 11 25.76 32.68 -6.92 12 29.92 36.93 -7.01 13 32.13 39.15 -7.02 14 33.28 127.75 -94.47 15 34.42 41.42 -7.00 16 36.79 43.76 -6.97 17 36.79 43.76 -6.97

18 39.25 46.16 -6.91 19 40.61 86.03 -45.42 20 47.71 66.36 -18.66 21 49.90 56.36 -6.47 22 52.77 59.07 -6.30

Rainfall events

EI30 Measured

EI30 Predicted Error

23 52.77 59.07 -6.30 24 52.77 59.07 -6.30 25 55.72 61.82 -6.11 26 57.47 86.03 -28.56 27 68.33 73.46 -5.13 28 85.90 89.32 -3.42 29 104.92 137.35 -32.42 30 105.44 106.64 -1.20 31 105.44 106.64 -1.20 32 109.58 110.28 -0.69 33 118.10 117.73 0.37 34 159.83 416.99 -257.16

51

35 180.85 171.69 9.16 36 202.70 190.22 12.48 37 225.71 209.63 16.08 38 249.85 284.57 -34.72 39 249.90 229.94 19.96

40 249.90 229.94 19.96 41 256.13 235.16 20.97 42 275.24 251.13 24.11 43 281.68 638.56 -356.88 44 320.47 613.59 -293.12

The spatial distribution of annual average erosivity for the period 2005 to 2009 is shown in Figure 5. The R factor varies from 704 to 2849 MJ mm/ha h year. Higher the erosivity higher is the chance for erosion. The spatial pattern clearly shows the areas vulnerable for soil erosion based on erosivity. All the interpolation methods clearly show the higher erosive power of rainfall in the southeast portion of the watershed. The amount of soil eroded from the watershed depends on many factors like the soil type, slope, and vegetation cover. This gives an insight into the rainfall erosive capability of the krishnagiri watershed

Fig 5. Spatial distribution of Rainfall Erosivity by different interpolation methods

4. Conclusion A Simplified relationship for estimating the erosivity in krishnagiri watershed area is proposed. Data

from three locations were used to develop the relationship and one additional station across the watershed is used to validate the developed relationship. It is established that the relationship between the erosivity and rainfall depth can be expressed in a potential form. The regression model and the erosivity map here provided represent a helpful mean for soil erosion assessment and mapping, both at watershed and at regional scale.

5. References [1] A. Abu Hammad, H. Lundekvam and T. Børresen , Adaptation of RUSLE in the Eastern Part of the Mediterranean

Region, Journal of Environmental Management, 2007, Vol.34, No.6, pp.829–841.

[2] B. K. Ramprasad, B. P. Kothyari, and R. K. Pande, Evaluation of rainfall erosivity in Bheta Gad catchment, Kumaun Hills of Uttar Pradesh, Central Himalayas, The Environmentalist, 2000,Vol 20, pp 301- 308.

[3] A. Ugur Ozcan, Gunay Erpul, Mustafa Basaran and H. Emrah Erdogan, Use of USLE/GIS technology integrated with geostatistics to assess soil erosion risk in different land uses of Indagi Mountain Pass ¸Ankiri, Turkey, Journal of Environmental Geology , 2008, Vol 53, pp 1731–1741.

[4] Carolina Munka, Gabriela Cruz, Ruben Mario Caffera, Long term variation in rainfall erosivity in Uruguay: a preliminary Fournier approach, GeoJournal , 2007, Vol 70, pp 257–262.

[5] Indrani Pal and Abir Al-Tabbaa, Suitability of different erosivity models used in RUSLE2 for the South West Indian region, The Environmentalist, 2009, Vol 29, pp 405–410.

[6] Mingxin MEN , Zhen rong Yu, Hao XU, Study on the spatial pattern of rainfall erosivity based on geostatistics in Hebei Province, China, Front. Agric. China , 2008, Vol 2(3),pp 281–289.

[7] Nazzareno Diodato, Predicting RUSLE (Revised Universal Soil Loss Equation) Monthly Erosivity Index from Readily Available Rainfall Data in Mediterranean Area, The Environmentalist, 2005, Vol 25,pp 63–70.

52