Natural Hazards

https://doi.org/10.1007/s11069-019-03670-9

ORIGINAL PAPER

Rainfall erosivity and erosivity density in Eastern Ghats

Highland of east India

Ch. Jyotiprava Dash1 · N. K. Das1 · Partha Pratim Adhikary1 

Received: 14 December 2016 / Accepted: 16 July 2019

© Springer Nature B.V. 2019

Abstract
The rainfall erosivity (R-factor in USLE) is the long-term average of the sum of the product
of rainfall kinetic energy and its maximum 30-min intensity. Therefore, at most 30-min
time intervals pluviograph records are required to calculate R-factor. But, such high-reso-
lution data are scarce in many parts of the world and require lengthy processing period. In
this study, R-factor was correlated with daily, monthly and annual rainfall, and its spatial
variability in Eastern Ghats Highland of east India was mapped. The result showed that
power regression models predicted satisfactorily the daily, monthly and annual R-factor,
of which annual R-factor model performed best (model efficiency 0.93). Mean monsoon
season R-factor was 15.6 and 10.0 times higher than the pre- and post-monsoon sea-
son R-factor, and thus remained highly critical with respect to erosion. Annual R-factor
values ranged from 3040 to 10,127  MJ  mm  ha−1  h−1  year−1, with standard deviation of
1981  MJ  mm  ha−1  h−1  year−1. Rainfall intensity was positively correlated with erosivity
density, and numerical value of rainfall intensity was almost double of the erosivity density
value. The combination of rainfall and erosivity density was used to identify flood, erosion
and landslide-prone areas. The developed iso-erosivity, erosivity density and risk maps can
be opted as a tool for policy makers to take suitable measures against natural hazards in
Eastern Ghats Highland of east India and elsewhere with similar rainfall characteristics.

Keywords  Erosivity · Erosivity density · Spatiotemporal variation · Soil erosion · Eastern

Ghats

1 Introduction

Soil erosion is a global problem, reducing the long-term viability of agriculture in many
parts of the world, and thus poses a significant threat to the world’s food security in the
context of an increasing global population (Lal 1998). Not only it declines the agricul-
tural productivity, it also exerts negative impact on surrounding ecosystem such as increase
in landslide activity, reservoir sedimentation, contaminant diffusion and other ecosystem

* Partha Pratim Adhikary

partha.adhikary@icar.gov.in; ppadhikary@gmail.com
1
ICAR-Indian Institute of Soil and Water Conservation, Research Centre, Sunabeda, Koraput,
Odisha 763002, India

13
Vol.:(0123456789)
 Natural Hazards

disturbances. Dabral et  al. (2008) reported that 75 billion tons of soil per year is being
removed due to erosion worldwide, most of which come from cultivated land, causing
20 M ha land unproductive. Hence, it is necessary to quantify the magnitude and spatial
extent of soil erosion for undertaking effective management strategies.
Since soil erosion spans a wide range of spatial scales, and the complexity of the vari-
ables makes erosion measurement difficult at large scales, soil erosion models play an
important role in estimating or predicting erosion. Among various soil erosion models,
the Universal Soil Loss Equation (USLE) (Wischmeier and Smith 1978) or Revised USLE
(RUSLE) (Renard et al. 1997), which predicts the average annual soil loss, has widely been
used at large spatial scales (Kinnell 2010; Farhan and Nawaiseh 2015; Panagos et al. 2015,
2016; Wu et al. 2016). Although all factors used within USLE or RUSLE are important for
generating erosion, the erosive force of rainfall (rainfall erosivity) is particularly strong,
as high volume and intensity of rainfall are the driving force for soil erosion processes,
through detachment and transport of soil particles and creation of runoff (Nyssen et  al.
2005).
The R-factor describes the effect of rainfall on sheet and rill erosion by measuring the
kinetic energy and intensity of erosive rainfall (Wischmeier and Smith 1978). The R-factor
not only used as an input parameter for soil erosion models, but also used for sediment
yield and water quality modeling. Hence, accurate assessment of R-factor is important for
a given area, which requires long-term pluviograph data (at least 20  years data, Renard
et al. 1997; Lee and Heo 2011). However, lack of such long-term pluviograph data is the
main concern in many places of the world. Even if sufficient pluviograph data are avail-
able, the calculation of R-factor is difficult because of its complicated and lengthy pro-
cessing procedure (Kinnell 2010; Lee and Heo 2011). To overcome this problem, develop-
ment of some simplified functions/models can be possible that correlate the R-factor with
more readily available rainfall data (daily, monthly and annual). Simple models have been
used to predict R-factor from daily (Raghunath and Erasmus 1971; Yu and Rosewell 1996;
Angulo-Martinez and Begueria 2009; Yang and Lu 2015), monthly (Renard and Frei-
mund 1994; Loureiro and Coutinho 2001; Tiwari et  al. 2015) and annual (Diodato and
Bellocchi 2010; Lee and Heo 2011) rainfall data. In these models, R-factor is calculated
as a linear (Rambabu et  al. 1978; Lo et  al. 1985; Sudhishri and Patnaik 2004), logarith-
mic (Wischmeier and Smith 1965), exponential (Richardson 1983), polynomial (Renard
and Freimund 1994) or power (Renard and Freimund 1994; Diodato 2004; Elangovan and
Seetharaman 2011) function of rainfall.
Annual R-factor maps were developed for India taking 44 stations by Rambabu et  al.
(1978) and 52 stations by Tiwari et al. (2015). As rainfall is erratic in nature and accord-
ingly R-factor varies, location-specific erosivity maps required to be developed to under-
take specific soil and water conservation measures. The average R-factor for the years
2004–2008 was found to be 1515 MJ mm ha−1 h−1 year−1 for Kerala, India (Prasannaku-
mar et al. 2012). However, at nearby locations in the same state, the average R-factor for
2005–2008 was observed to be 32 MJ mm ha−1 h−1 year−1 (Prasannakumar et al. 2011).
Similar variation was also noticed in Arunachal Pradesh; average annual R-factor was
observed as 1895 MJ mm ha−1 h−1 year−1 (Dabral et al. 2008), but the estimated R-factor
was 9748 MJ mm ha−1 h−1 year−1 with standard deviation of 1602 MJ mm ha−1 h−1 year−1
(Rawat et  al. 2013). Sudhishri and Patnaik (2004) reported mean seasonal and annual
R-factor for Eastern Ghats High Zone of Odisha taking 10 gauging stations data recorded
for 7 years (1995–2001). Their estimates of seasonal and annual R-factor ranged from 388
to 945 and 619 to 1062 m t ha−1 cm−1, respectively. However, their results were limited in
scope due to brief periods of data availability, which were less than the recommended data

13
Natural Hazards

requirement (over 20 years). Elangovan and Seetharaman (2011) reported the annual aver-
age R-factor for the period 2005 to 2009 ranged from 704 to 2849 MJ mm ha−1 h−1 year−1
for Krishanagiri watershed region of Tamil Nadu, India. Similarly Praveen and Kumar
(2011) reported R-factor varied from 508 to 584 MJ mm ha−1 h−1 year−1 in Upper South
Koel Basin, Jharkhand, India. Heterogeneity in rainfall and subsequent R-factor across
India encourages for assessing R-factor for different locations. Therefore, the objectives of
this study have been framed as (1) to develop a suitable R-factor model based on rainfall
data and (2) to assess the spatiotemporal pattern of rainfall erosivity and erosivity density
in Koraput district of Odisha, which comes under Eastern Ghats Highland region of east
India.

2 Materials and methods

2.1 Study area

Eastern Ghats region (19.8 Mha) of India, which comprises of hillocks, sloping lands and
fertile flat lands, is severely affected by soil erosion (Lenka et al. 2012). Koraput district of
Odisha is one of those areas of Eastern Ghats Highland zone, where soil erosion is a seri-
ous problem due to rolling topography, high-intensity rainfall and deforestation (Sudhishri
and Patnaik 2004; Lenka et  al. 2012; Adhikary et  al. 2015, 2017). The study area lies
between 81°05′ to 83°5′ East longitude and 18°4′ to 19°5′ North latitude (Fig. 1), covering
an area of 8379 km2. It is situated at elevation varying between 127 m and 1655 m above
mean sea level (msl). Around 8% of the study area lies below 500 m above msl, 27% above
900  m above msl, and 65% between 500 and 900  m above msl. Most of the plain lands
are located in western and northwestern parts and hilly regions in the east, southeast and
central parts of the study area. The district is divided into 14 administrative blocks, namely
Bandhugaon, Boipariguda, Boriguma, Dasamantapur, Jeypore, Koraput, Kotapad, Kundra,
Lamataput, Laxmipur, Nandapur, Narayanpatna, Pottangi and Semiliguda. The climate
is tropical with mean annual maximum and minimum temperatures of 35.8 and 7.6  °C,
respectively (Adhikary et al. 2015). Monsoon season starts in the second week of June and
continues up to September. The mean annual rainfall is 1452.2 mm occurring in 70 days.
The monsoon season which occurs during June to September contributes more than 80%
of the annual rainfall, whereas annual erosive rainfall accounts for about 75% of the total
annual rainfall in the study area. However, the area experiences a special monsoon feature
characterizing intense and short-duration storms, and helps in generating enormous runoff
and heavy soil loss (Lenka et al. 2012; Adhikary et al. 2017, 2018). The wettest months are
July (332.5 mm) and August (311.0 mm), whereas the driest ones are December (4.4 mm),
January (8.0 mm) and February (11.6 mm) (Fig. 1).

2.2 Data acquisition and model development

In the study area, among 15 meteorological stations, only one station has self-recording
rain gauge (ICAR-Indian Institute of Soil and Water Conservation (IISWC), Research
farm), and all others are occupied with non-recording rain gauges (Table  1). The self-
recording rain gauge rainfall data of the meteorological observatory of IISWC, Research
Centre, Sunabeda, Koraput, Odisha, India, were used for the computation of R-factor
for a period of 22  years (1994–2015). All the rainfall events greater than 12.6  mm were

13
 Natural Hazards

Fig. 1  Location of meteorological stations along with the mean rainfall and erosive rainfall events in the
study area (Inset: mean monthly rainfall (1994–2015))

considered as erosive events (Wischmeier and Smith 1978) and were separated when there
was no rainfall for more than 6 h. The duration of each erosive event on the recording rain
gauge chart was divided into different time intervals depending on the slope of the chart,
and the rainfall intensity for that time interval was calculated. Then, the energy for each
interval (Ei), was calculated using the model developed by Brown and Foster (1987) as this

13
Natural Hazards

Table 1  Characteristics of Station name Latitude Longitude Elevation (m)

gauging stations used in this
study
Bandhugaon 18°56′00.02″ 83°15′26.25″ 237
Boipariguda 18°45′03.98″ 82°25′59.86″ 778
Boriguma 19°03′06.18″ 82°32′55.01″ 593
Dasamantapur 19°02′44.45″ 82°55′16.49″ 889
Jeypore 18°51′38.21″ 82°33′03.77″ 584
Koraput 18°48′48.55″ 82°42′44.39″ 887
Kotapad 19°08′25.55″ 82°19′05.01″ 557
Kundra 18°55′19.34″ 82°23′28.64″ 589
Lamataput 18°37′45.43″ 82°35′49.24″ 906
Laxmipur 18°59′27.72″ 83°06′53.98″ 902
Nandapur 18°33′50.51″ 82°44′14.67″ 985
Narayanpatna 18°52′36.89″ 83°10′08.11″ 376
Pottangi 18°34′06.68″ 82°58′32.50″ 1033
Semiliguda 18°42′40.21″ 82°51′02.80″ 913
Research farm 18°41′36.74″ 82°48′08.01″ 910
(Recording rain
gauge)

was found best for this region when compared among six well-known models (Dash et al.
2018).
Er = 0.29[1 − 0.72 exp(−0.05ir )] (1)
where Er is the energy per unit rainfall (MJ mm−1 ­ha−1) and ir is the rainfall intensity for
each interval (mm  h−1). Then, the values of Er were multiplied by the rainfall amounts
for each interval to calculate the rainfall energy for that interval. By summing these
values, the total energy (E) of the rainfall (MJ  ha−1) was calculated. Further, the maxi-
mum rainfall intensity for a 30-min period of the rainfall event (I30, mm  h−1) was deter-
mined. Eventwise erosivity factor (Re, MJ  mm  ha−1  h−1) was calculated by multiplying
E by I30. The daily erosivity factor (Rd) (MJ mm ha−1 h−1), monthly erosivity factor (Rm)
(MJ mm ha−1 h−1 month−1), seasonal erosivity factor (Rs) (MJ mm ha−1 h−1 season−1) and
annual erosivity factor (Ra) (MJ mm ha−1 h−1 year−1) were also determined.
The daily, monthly and annual R-factors were calculated by using the following
equations:
n
∑
Rd = (Re ) (2)
i=1

l
∑
Rm = (Rd ) (3)
j=1

r
∑
Ra = (Rm ) (4)
k=1

13
 Natural Hazards

where n is the number of erosive events in a day, l is the number of days in a

month having erosive events and r is the number of months. The seasonal ero-
sivity factors were calculated for three seasons, namely pre-monsoon (Febru-
ary–March–April–May), monsoon (June–July–August–September) and post-monsoon
(October–November–December–January).
Further, relationships between daily, monthly and annual rainfall and R-factor were
developed using linear (R = aP + b), power (R = aPb) and polynomial (R = aP2 + bP + c)
functions, where P is the daily, monthly or annual rainfall, and a, b and c are constants. The
best-fit regression model was chosen based on highest coefficient of determination (R2)
value.

2.3 Model evaluation

The level of agreement between the model predicted and observed data of event, daily,
monthly and annual R-factors was assessed using three statistical procedures, viz. mean
absolute error (MAE), root mean square error (RMSE) and Nash–Sutcliffe modeling effi-
ciency (Ef) (Nash and Sutcliffe 1970), and calculated using the following equations:
n
∑
MAE = |(Oi − Pi )|∕n (5)
i=1

√
√ n
√∑
RMSE = √ (Oi − Pi )2 ∕n (6)
i=1

∑n
(Oi − Pi )2
Ef = 1 − ∑i=1
n (7)
̄ 2
(Oi − O)
i=1

where Pi is the predicted values, Oi is the observed values, Ō is the mean of observed val-
ues and n is the total number of observations. The RMSE value of 0 and Ef value of 1 rep-
resent a perfect prediction. The best developed models among daily, monthly and annual
R-factors were used to estimate the R-factor for other stations.

2.4 Calculation of erosivity density

As rainfall amounts and R-factor values vary over time and space, and sufficient length of
record is needed to obtain reliable average value of R-factor, to address this issue, the ero-
sivity density (ED) concept was introduced (Foster et al. 2008; Panagos et al. 2016). This
index is far more stable and does not require more length of rainfall record, a better choice
in case of missing data, and also relatively independent of elevation up to 3000 m (Foster
et al. 2008; Dabney et al. 2011). The ED (MJ ha−1 ­h−1) is expressed as the ratio of R-factor
to rainfall (Foster et al. 2008; Panagos et al. 2016), which measures the erosivity per unit
rainfall (mm). In this study, monthly, seasonal and annual EDs were determined.

13
Natural Hazards

2.5 Spatial interpolation of R‑factor and ED

The inverse distance weighing (IDW) is a deterministic interpolation technique which cre-
ates surfaces from sample points using mathematical functions based on the extent of simi-
larity. This simplicity of the method has made it popular (Adhikary and Dash 2017). In
IDW, the interpolated estimates are based on values at nearby location, and it gives weight
to data points such that their influence on prediction is reduced as distance from the point
increases (Adhikary and Dash 2017). The interpolated value z(x0) is expressed in Eq. 8.
∑n xi
i=1 h𝛽
ij
z(x0 ) = ∑n 1 (8)
i=1 h𝛽
ij

where xi is the ith data value, hij is the separation distance between sample data value and
the interpolated value, n is the total number of sample data values and β is the weighting
power. The estimation quality will be significantly governed by the choice of the weight-
ing power (Mueller et al. 2001; Adhikary and Dash 2017). The optimal weighting power
is dependent on the spatial structure of the data and is primarily influenced by the coeffi-
cient of variation (CV), skewness and kurtosis of the data (Mueller et al. 2001). The opti-
mal power function was assessed by testing of a series of powers ranging from 1.0 to 4.0
(Kravchenko and Bullock 1999). In this work, the weighing power was selected by a series
of small increments to determine the value that minimizes the prediction error. Three sta-
tistics, namely MAE, RMSE and mean absolute percentage errors, (MAPE) were used to
describe the prediction error.

3 Results and discussion

This part of the paper will deal with (1) the development and validation of R-factor models
for the study area based on rainfall data, (2) the mapping of spatial variation of R-factor
for the study area and (3) the assessment of seasonal and annual ED pattern over the study
area.

3.1 R‑factor as a function of rainfall

Five different regression models such as linear, exponential, power, logarithmic and poly-
nomial were used to develop a relationship between Re (MJ mm ha−1 h−1) and event erosive
rainfall (Pe, mm), and observed that the relationship between Re and Pe was fitted best for
power regression model (Re = 0.295P1.769e ) having coefficient of determination 0.659, and
is presented in Fig. 2a and Table 2. Similar relationship was also observed for Rd and daily
rainfall (Pd) (Fig.  2b), with a coefficient determination value of 0.650. In this study, the
value of b for Rd was found to be 1.850, which was within the range of b values reported
by other researchers (Sepaskhah and Sarkhosh 2005). Brown and Foster (1987) reported
the value of b as 2.0, while Cooley (1980) suggested the value of b should be ranged from
1.5 to 2.2 based on empirical approaches. Similarly, Richardson (1983) derived the value
of b equal to 1.81 for 11 locations of east of the Rocky Mountains in the USA. Bagarello

13
 Natural Hazards

Fig. 2  Regression models developed for a daily erosive rainfall and erosivity, b daily rainfall and erosivity,
c monthly rainfall and erosivity and d annual rainfall and erosivity

Table 2  Best-fit models for rainfall and R-factor and their prediction statistics
Relationship Type of equa- Model ME R2 RMSE Ef
between tion (MJ mm ha−1 h−1) (MJ mm ha−1 h−1)

Re and Pe Power Re = 0.295 × P1.769

e 0.09 0.659 168.5 0.69
Rd and Pd Power Rd = 0.208 × P1.850
d 5.62 0.650 176.3 0.61
Rm and Pm Power Re = 2.686 × P1.119
m 9.02 0.815 290.1 0.86
Ra and Pa Power Ra = 0.008 × P1.887
a 10.7 0.841 387.9 0.93

ME, mean error; RMSE, root mean square error; Ef, Nash–Sutcliffe model efficiency; Re, eventwise R-fac-
tor; Rd, daily R-factor; Rm, monthly R-factor, Ra, annual R-factor, Pe, event rainfall; Pd, daily rainfall; Pm,
monthly rainfall; Pa, annual rainfall

and D’Asaro (1994) for 32 Sicilian locations and 3 locations in the continental south of
Italy obtained values of b ranged between 1.22 and 2.08 and proposed an average value of
1.54 (Sepaskhah and Sarkhosh 2005). The relationship between the Rm and Pm followed
the same trend that is power regression model (Rm = 2.686*P1.119m ). Sudhishri and Patnaik
(2004) developed a linear function for monthly rainfall and erosivity (R2 = 0.91). In this
study, it was observed that relationship between Ra and Pa also followed power regression
model (Ra = 0.008*P1.887
a ), with b value of 1.887. Renard and Freimund (1994) suggested
estimating annual R-factor, using annual rainfall by the power regression model in the con-
tinental United States, with b value of 1.61, when rainfall is less than 850 mm. Similarly,
Jung et al. (1983) noticed a positive relationship between Pa and Ra-factor at Suwon, in the
midwestern region of Korea, over 17 years (1964–1980), and they obtained power regres-
sion model (b = 1.4947), with the coefficient of determination of 0.750. Sanchez-Moreno
et al. (2014) reported power function was the best fit between R-factor and annual rainfall
for Santiago Island, Cape Verde (b = 1.68 and R2 = 0.92).

13
Natural Hazards

Table 3  Prediction errors of MAE RMSE MAPE

R-factor and erosivity density by
IDW method used in the study Erosivity
area
 Pre-monsoon 41.84a 46.40a 9.45c
 Monsoon 291.86a 366.51a 4.71c
 Post-monsoon 46.23a 54.09a 7.63c
 Annual 381.02a 524.72a 5.98c
Erosivity density
 Pre-monsoon 0.53b 0.66b 13.97c
 Monsoon 0.61b 0.75b 11.22c
 Post-monsoon 0.29b 0.32b 6.21c
 Annual 0.45b 0.54b 10.34c

MAE, mean absolute error; RMSE, root mean square error; MAPE,
mean absolute percentage error; a, MJ  mm  ha−1  h−1  year−1; b,
MJ ha−1 h−1; c, %

In this study, the models developed for determining daily, monthly and annual R-factors
were compared. Selection of the best-fit model was based on three statistical procedures,
viz. MAE, RMSE and Ef. They were calculated using Eqs.  5, 6 and 7, respectively, and
the values are presented in Table 2. It was observed that among the developed four erosiv-
ity models, annual model best represented the R-factor using annual rainfall, having high-
est R2 (0.841) and Ef value (0.93). This study agrees with Angulo-Martínez and Beguería
(2009) that daily or monthly erosivity model has inherent limitations in the use of daily
or monthly weather records for estimating the R-factor in the USLE/RUSLE. Moreover,
Selker et al. (1990) also emphasized on poor representation of R-factor by daily or monthly
rainfall parameters. Apart from this, many studies reported strong correlations between
annual R-factor and annual rainfall for many locations around the world (Lo et al. 1985;
Sepaskhah and Sarkhosh 2005; Angulo-Martínez et al. 2009; Diodato and Bellocchi 2010;
Bonilla and Vidal 2011; Lee and Heo 2011).
Though annual rainfall is a good indicator of annual R-factor, monthly and seasonal
erosivity is required for making suitable management strategies in the erosion-prone areas.
Therefore, in this study both monthly and seasonal erosivity models were used to compute
R-factor at other locations and mapping of the Rs-factor and Ra-factor in the study area was
also carried out.

3.2 Validation of IDW method

To evaluate the accuracy of IDW method in spatial interpolation of R-factor and ED,
three prediction errors, namely MAE, RMSE and MAPE, were calculated and are pre-
sented in Table 3. It was observed that MAE and RMSE values for annual R-factor were
381.0 and 524.7 MJ mm ha−1 h−1 year−1, respectively, and those were 5.9 and 8.2% of
the average R-factor value, respectively. Similarly, MAE and RMSE values for mon-
soon R-factor were 291.8 and 366.5 MJ mm ha−1 h−1 year−1, respectively. Mean abso-
lute percentage error values for R-factor interpolation ranged 4.71–9.45%. MAE and
RMSE values for annual ED were 0.45 MJ ha−1 h−1 and 0.54 MJ ha−1 h−1, respectively.
MAPE values for ED were slightly higher than that of R-factor. However, MAPE values
for all the cases were less than 15%, indicating fair prediction by IDW method. In IDW

13
 Natural Hazards

Fig. 3  Mean annual rainfall in the study area

interpolation technique, the MAPE values of < 15% are considered fair and can be used
for all practical purposes (Adhikary and Dash 2017). All the data indicating that the
interpolation error was less than 10% and much lower than their respective standard
deviation values thus can be used for the prediction of surface map.

3.3 Mapping of annual rainfall, seasonal and annual R‑factors

It was observed that among the developed four erosivity models, annual R-factor model
best represented the R-factor using annual rainfall, having highest R2 value (0.841), and
low MAE (10.7) and RMSE (387.9) values. The mean annual rainfall map of the study
area is presented in Fig. 3. It was observed that the annual rainfall ranged between 980
and 1843 mm. Low rainfall occurs in the northeast part of the area, whereas higher rain-
fall occurs in southwest part of the study area. The monthly R-factor values for differ-
ent meteorological stations of the study area are presented in Table 4. The highest mean
monthly R-factor was noticed in July (2778 MJ mm ha−1 h−1 month−1) followed by August
(2770 MJ mm ha−1 h−1 month−1) and September (1816 MJ mm ha−1 h−1 month−1), while
the lowest value was found in February (0.2 MJ mm ha−1 h−1 month−1) followed by Janu-
ary (1.2 MJ mm ha−1 h−1 month−1), December (2.5 MJ mm ha−1 h−1 month−1) and March
(9.0 MJ mm ha−1 h−1 month−1). The monthly R-factor values were summed to calculate
the seasonal R-factor. The R-factor during pre-monsoon season ranged 184.8–576.4 MJ 
mm ha−1 h−1 month−1, while during monsoon, its value ranged 4113–8669 MJ mm ha−1 
h−1  month−1 (Fig.  4a, b). It was observed that the mean monsoon season R-factor was
almost 15.6 and 10.0 times higher than pre-monsoon and post-monsoon season R-factor,
respectively (Table 5). Similarly, mean R-factor was nearly 1.5 times higher during post-
monsoon season than pre-monsoon season. The higher R-factor during post-monsoon
season in the study area is mainly due to cyclonic storm occurred in the Bay of Bengal

13
 Natural Hazards

Table 4  Monthly R-factor for different meteorological stations in the study area

Station name January February March April May June July August September October November December

MJ mm ha−1 h−1 month−1
Bandhugaon 1.5 1.0 9.0 31.0 143.9 788.7 1077.1 1240.7 1006.0 502.2 43.3 2.5
Boipariguda 25.7 9.9 44.7 192.8 191.5 1294.5 2788.2 2770.3 1815.8 550.0 110.9 27.3
Boriguma 1.2 2.7 12.5 133.2 265.2 1421.5 1671.0 1847.2 1233.2 396.1 48.3 27.4
Dasamantapur 38.2 22.9 23.8 111.5 161.3 1050.0 2265.1 2504.2 1381.5 513.0 29.6 14.1
Jeypore 29.6 12.8 75.5 180.3 307.9 1375.6 2529.3 2620.5 1570.9 427.2 99.5 22.9
Koraput 15.0 19.4 65.1 183.1 305.2 1258.5 2193.6 2496.1 1440.3 544.7 64.6 18.1
Kotapad 6.6 9.9 49.4 164.2 188.4 1123.0 2015.8 2296.1 1506.7 548.4 51.8 14.2
Kundra 7.6 10.1 40.0 150.4 131.6 1200.9 1782.9 1954.1 1122.5 385.8 20.2 16.3
Lamataput 2.9 0.6 20.6 67.3 99.7 759.1 1334.4 1582.3 744.0 319.1 42.2 19.7
Laxmipur 27.3 4.5 22.8 216.8 264.6 1055.0 1415.2 1985.4 1246.3 573.3 49.0 18.5
Nandapur 40.9 20.7 39.0 144.4 214.7 1251.4 2047.8 2154.7 1360.8 630.1 79.9 29.1
Narayanpatna 21.9 1.0 19.9 87.4 267.1 905.5 1198.2 1514.0 995.1 656.7 186.1 26.2
Pottangi 4.5 0.2 105.4 149.0 213.2 1148.8 1702.0 1998.5 1458.9 729.1 75.3 46.6
Semiliguda 48.2 9.4 48.2 148.0 205.3 1034.4 1848.3 2037.3 1347.7 532.0 80.5 16.8
Research farm 13.3 26.4 59.8 87.6 159.6 730.3 1469.9 1231.5 778.9 234.7 82.0 5.1

13
 Natural Hazards

Fig. 4  Seasonal and annual R-factor map of Koraput district a pre-monsoon, b monsoon, c post-monsoon
and d annual

Table 5  Seasonal and annual R-factor for different meteorological stations

Station name Pre-monsoon (Febru- Monsoon (June– Post-monsoon Annual R-factor
ary–March–April– July–August–Sep- (October–November–
May) tember) December–January)
MJ mm ha−1 h−1 season−1 MJ
mm ha−1 h−1 year−1

Bandhugaon 184.8 4112.5 549.5 3039.7

Boipariguda 438.9 8668.8 713.9 10,127.0
Boriguma 413.6 6172.8 473.0 5786.8
Dasamantapur 319.6 7200.7 594.9 7262.4
Jeypore 576.4 8096.3 579.2 9215.6
Koraput 572.7 7388.5 642.4 8162.6
Kotapad 411.9 6941.5 621.0 7117.6
Kundra 332.1 6060.4 430.0 5427.9
Lamataput 188.2 4419.8 383.9 3172.6
Laxmipur 508.7 5702.0 668.1 5620.4
Nandapur 418.9 6814.8 780.0 7265.2
Narayanpatna 375.4 4612.8 891.0 4355.9
Pottangi 467.8 6308.2 855.5 6719.1
Semiliguda 410.9 6267.7 677.5 6277.6
Research farm 333.4 4210.5 335.2 6083.6
Mean 396.9 6198.5 613.0 6375.6

13
Natural Hazards

Fig. 5  Relationship between rainfall intensity and erosivity density

during October and November. However, pre- and post-monsoon R-factor did not dif-
fer significantly, indicating no or less seasonal variability in mean R-factor for both pre-
monsoon and post-monsoon seasons.
The mean annual R-factor in the study area was 6376  MJ  mm  ha−1  h−1  year−1 with
standard deviation of 1981  MJ  mm  ha−1  h−1  year−1 (Table  5). The range of R-factor in
Koraput district was 3040–10,127 MJ mm ha−1 h−1 year−1 (Fig. 4d). The highest annual
R-factor was observed in Boipariguda block (10,127 MJ mm ha−1 h−1 year−1), whereas the
lowest R-factor was observed at Bandhugaon (3040 MJ mm ha−1 h−1 year−1).

3.4 Erosivity density as a function of rainfall intensity and its classification

The ED values were calculated by dividing R-factor by the corresponding rainfall depth
(Foster et al. 2008; Panagos et al. 2016). Smaller values of ED indicated that the R-factor
was mainly influenced by the rainfall amount, while high values of ED explained that rain-
fall event of high intensity was occurred (Panagos et al. 2016). In the study area, relation-
ship between rainfall intensity and ED is presented in Fig. 5, and it is observed that ED was
significantly and positively correlated with rainfall intensity, and followed power regres-
sion equation. In the study area, 85% of rainfall intensity values found to be concentrated in
the range of 5–10 mm h−1, whereas majority of ED values (65%) observed to be less than
5 MJ ha−1 h−1.
Dabney et al. (2011) highlighted the significance of ED in predicting soil erosion and
reported that high monthly ED values were associated with the prediction of high runoff.
Therefore, the regions with high ED are often exposed to risk of flooding and even water
scarcity because of infrequent but very intense erosive rainstorms (Panagos et  al. 2016).
As high ED values are mainly influenced by rainfall intensity, rather rainfall amount, in
this study an attempt was made to classify ED values (very low, low, medium, high and
very high) based on rainfall intensity classification given by Girmay et al. (2009) for tropi-
cal region, which is more prevalent in the study area and presented in Table  6. For the
study area, ED value was found to be less than 3 MJ ha−1 h−1, when rainfall intensity was

13
 Natural Hazards

Table 6  Classification of Category Rainfall intensity (mm h−1) Erosiv-

erosivity density for tropical ity density
rainfall (MJ ha−1 h−1)

Very low < 6 < 3

Low 6–12 3–6
Medium 12–18 6–9
High 18–30 9–15
Very high > 30 > 15

Fig. 6  Variation of monthly rainfall, R-factor and erosivity density in the study area

very low (< 6 mm h−1). Similarly when rainfall intensities were high (18–30 mm h−1) and
very high (> 30  mm  h−1), ED values were observed to be more than 9  MJ  ha−1  h−1 and
15  MJ  ha−1  h−1, respectively. Therefore, it can be inferred that ED values of more than
3 MJ ha−1 h−1, which was considered high for temperate region (Dabney et al. 2011), that
same value might not be high for tropical region, particularly dominated by monsoon.

3.5 Monthly, seasonal and annual erosivity density

The mean monthly rainfall, Rm-factor and ED are presented in Fig.  6. The monthly ED
was found to be more than 3.0 MJ ha−1 h−1 for all the months of the year except Febru-
ary. For four months (November–December–January–March), the ED values ranged from
3.2 to 3.8 MJ ha−1 h−1, and this was because of the predominance of low-intensity rainfall
events. Starting from April to August, ED values showed an increasing trend. Both July
and August months were having highest ED (5.4  MJ  ha−1  h−1), followed by September
(5.2  MJ  ha−1  h−1) and June (5.1  MJ  ha−1  h−1). Months having high ED values indicated
that the rainfall was characterized by high-intensity events of short duration (Panagos et al.
2016), which was noticed to be true for Eastern Ghats region of India. Therefore, in terms
of erosivity, August, July, September and June (Monsoon season) are more critical months
in the study area due to higher R-factor, combined with high-intensity rainfall.
The spatial variability of seasonal as well as annual ED for the study area is pre-
sented in Fig.  7. The ED was highest for monsoon season, followed by post- and pre-
monsoon seasons. The pre-monsoon ED values ranged from 1.60 to 7.10 MJ ha−1 h−1,

13
Natural Hazards

Fig. 7  Seasonal and annual erosivity density in the study area a pre-monsoon, b monsoon, c post-monsoon
and d annual

with a mean value of 3.82 MJ ha−1 h−1. Similarly, the post-monsoon ED values ranged

2.50–7.40  MJ  ha−1  h−1, with a mean value of 4.67  MJ  ha−1  h−1. The ED during post-
monsoon was nearly 1.2 times higher than the pre-monsoon values. The higher post-
monsoon ED values were due to the occurrence of heavy rainfall during that time
because of the development of cyclonic storm in the Bay of Bengal. This was also
depicted from the post-monsoon R-factor map, as the eastern part of the study area
showed higher R-factor, and that area is near to Bay of Bengal. The ED values dur-
ing monsoon season ranged from 3.50 to 7.60  MJ  ha−1  h−1, with mean value of
5.41  MJ  ha−1  h−1, which was 1.42 and 1.16 times higher than that of pre- and post-
monsoon values, respectively.
Spatial patterns of ED were found to be behaved differently during all the three sea-
sons. The central part of the study area showed higher values during pre-monsoon season
than other parts of the study area, while the east to west gradient was more prominent in
post-monsoon season, though the pre-monsoon and post-monsoon seasons had almost sim-
ilar ED values, but showing different spatial patterns over the study area. The annual ED
ranged from 3.10 to 5.50 MJ ha−1 h−1 with a mean value of 4.36 MJ ha−1 h−1. The spatial
distribution pattern of annual ED was observed to be more or less similar to the monsoon
ED, as this period contributed 83% of the annual rainfall, and was the most erosive one.
For the study area, from the maps of the ED, it was observed that the intense rainfall has
taken place between June and September, implying possibility of high soil erosion.

13
 Natural Hazards

Fig. 8  Map showing the risk areas based on annual rainfall and erosivity density

3.6 Erosivity density and risk area identification

Erosivity density along with rainfall can identify the areas vulnerable to natural haz-
ards like flood, soil erosion, landslide, etc. ED and rainfall maps were reclassified and
overlaid to categorize nine combined classes that represent the four quartiles of each
parameter. As described by Panagos et  al. (2015), highest risk is associated with the
areas where high ED is accompanied by low annual rainfall. The areas where ED value
was higher than 5.0 MJ ha−1 h−1 (4th quartile) and rainfall value was less than 1255 mm
(1st quartile) belong to this class. In this case, highly erosive rainfall accompanied with
dry soils can increase the risk of flood (Diodato et  al. 2011). The areas with lowest
risk comprised of the ED value less than 3.5  MJ  ha−1  h−1 (1st quartile) and rainfall
more than 1600 mm (4th quartile). Eastern part of the study area comprising Bandhu-
gaon block and a pocket at the southern fringe is medium to  high-risk-prone areas in
the region (Fig. 8). A wider stretch of land from west to northern part of the study area
comprising part of Boipariguda, Jeypore and Boriguma blocks where rainfall is high
but evenly distributed is coming under low-risk category. The areas where ED and rain-
fall both are high may be associated with landslide and soil erosion risk. But the land
slope and geological conditions in the study area is more conducive for soil erosion than
landslide.

13
Natural Hazards

3.7 Rainfall erosivity and soil erosion

In the study area, annual rainfall ranged between 980 and 1843  mm and more intense
events were noticed in the monsoon season from June to September. It was observed
that, only from the cultivable part of the study area, 13,333 thousand tonnes of soil was
lost annually at the rate of 43.9  t  ha−1  year−1 (Naik et  al. 2015). Naik et  al. (2015) also
reported that cropping sequences like finger millet–fallow–fallow, rice–fallow–fallow
and niger–fallow–fallow contributed 20.4%, 19.9% and 14.5% of total soil loss, respec-
tively, from the study area. Madhu et  al. (2016) reported the potential soil loss of the
study area as 25–30  Mg  ha−1  year−1. In a field experiment in Eastern Ghats High Land
of Odisha, Adhikary et  al. (2017) found the soil loss from upland paddy cultivated
fields as 12.5  Mg  ha−1  year−1. In another field experiment in the study area, Jakhar
et  al. (2016) reported the annual average soil loss from finger millet cultivated fields as
9.5 Mg ha−1 year−1. Soil loss not only from the agricultural crops, but also from aromatic
grasses is alarming. Soil losses from citronella, lemon and palmarosa grasses in this area
are 3.38, 3.93 and 7.71 Mg ha−1 year−1 (Adhikary et al. 2018). As the study area is vulner-
able to soil erosion, the R-factor maps in combination with ED maps will be helpful in
identifying the time period and locations of intense and heavy rainfall events compared to
low-intensity rainfall events. Similarly, the temporal distribution of R-factor can be very
important for identifying the high-risk periods especially when the soil remains bare dur-
ing high-erosion months (Sheridan and Rosewell 2003; Panagos et al. 2016). Almost 85%
of the annual rainfall erosivity in Koraput district is accounted during monsoon season
(June–September). Hence, the seasonal land use and land management practices can be
combined with seasonal distribution of R-factor to accurately predict the seasonal soil ero-
sion rate. Moreover, the development of annual erosivity maps can be helped in the estima-
tion of annual soil loss from any area (Panagos et al. 2016). The developed R-factor and
ED maps can be used for identifying the areas prone to soil erosion for a specific time of
the year, which can be considered as a useful tool for the policy makers to prioritize the
soil conservation activities for effective remediation.

4 Conclusions

The use of empirical models based on daily, monthly or annual rainfall will help to give an
approximate estimation of R-factor, in the absence of the pluviograph data. The R-factor
maps and the subsequent spatial and temporal ED assessment are important because of soil
erosion risk assessment and subsequent application of suitable soil and water conserva-
tion measures. In the study area, the monsoon season alone contributed 85% of the annual
R-factor and thus can be considered as very sensitive in terms of soil erosion. The mean
annual R-factor in Koraput was 6376  MJ  mm  ha−1  h−1  year−1. The spatial variability of
annual R-factor was high in the study area with the southwestern part having the highest
values, while the northeastern part having relatively low R-factor. When rainfall intensi-
ties were high (18–30  mm  h−1) and very high (> 30  mm  h−1), ED values were observed
to be more than 9 MJ ha−1 h−1 and 15 MJ ha−1 h−1, respectively. The spatiotemporal maps
of R-factor in combination with ED maps can be useful for natural resources conservation
policy makers to identify the hotspots and periods of highest soil erosion risk, and may
allow for taking timely and suitable soil conservation measures.

13
 Natural Hazards

Acknowledgements The authors would like to acknowledge Dr. P.K. Mishra, Former  Director, ICAR-
IISWC, and Dr. M. Madhu, Head, ICAR-IISWC, RC, Sunabeda, Koraput, for providing access to the
Research Centre’s data.

References
Adhikary PP, Dash ChJ (2017) Comparison of deterministic and stochastic methods to predict spatial variation
of groundwater depth. Appl Water Sci 7(1):339–348
Adhikary PP, Madhu M, Dash ChJ, Sahoo DC, Jakhar P, Naik BS, Hombegowda HC, Naik GB, Dash B (2015)
Prioritization of traditional tribal field crops based on RWUE in Koraput district of Odisha. Ind J Tradit
Knowl 14(1):88–95
Adhikary PP, Hombegowda HC, Barman D, Jakhar P, Madhu M (2017) Soil erosion control and carbon seques-
tration in shifting cultivated degraded highlands of eastern India: performance of two contour hedgerow
systems. Agrofor Syst 91(4):757–771
Adhikary PP, Hombegowda HC, Barman D, Madhu M (2018) Soil and onsite nutrient conservation potential of
aromatic grasses at field scale under a shifting cultivated, degraded catchment in Eastern Ghats, India. Int
J Sediment Res 33(3):340–350
Angulo-Martinez M, Begueria S (2009) Estimating rainfall erosivity from daily precipitation records: a com-
parison among methods using data from the Ebro Basin (NE Spain). J Hydrol 379:111–121
Angulo-Martínez M, López-Vicente M, Vicente-Serrano SM, Beguería S (2009) Mapping rainfall erosivity
at a regional scale: a comparison of interpolation methods in the Ebro Basin (NE Spain). Hydrol Earth
Syst Sci 13(10):1907–1920
Bagarello V, D’Asaro F (1994) Estimating single storm erosion index. Trans Am Soc Agric Eng 33:785–791
Bonilla CA, Vidal KL (2011) Rainfall erosivity in Central Chile. J Hydrol 410(1–2):126–133
Brown LC, Foster GR (1987) Storm erosivity using idealized intensity distribution. Trans Am Soc Agric Eng
30:379–386
Cooley KR (1980) Erosivity values for individual design storms. ASCE J Irrig Drain Div 106(2):135–145
Dabney SM, Yoder DC, Vieira DAN, Bingner RL (2011) Enhancing RUSLE to include runoff-driven phenom-
ena. Hydrol Process 25(9):1373–1390
Dabral PP, Baithuri N, Pandey A (2008) Soil erosion assessment in a hilly catchment of North Eastern India
using USLE, GIS and remote sensing. Water Resour Manag 22(12):1783–1798
Dash ChJ, Adhikary PP, Das NK, Alam NM, Mandal U, Mishra PK (2018) Comparison of rainfall kinetic
energy–intensity relationships for Eastern Ghats Highland region of India. Nat Hazards 93:547–558
Diodato N (2004) Estimating RUSLE’s rainfall factor in the part of Italy with a Mediterranean rainfall regime.
Hydrol Earth Syst Sci 8(1):103–107
Diodato N, Bellocchi G (2010) MedREM, a rainfall erosivity model for the Mediterranean Region. J Hydrol
387(1):119–127
Diodato N, Bellocchi G, Romano N, Chirico GB (2011) How the aggressiveness of rainfalls in the Mediterra-
nean lands is enhanced by climate change. Clim Change 108(3):591–599
Elangovan AB, Seetharaman R (2011) Estimating rainfall erosivity of the Revised Universal Soil Loss Equation
from daily rainfall depth in Krishanagiri watershed region of Tamil Nadu, India. In: International confer-
ence on environmental and computer science IPCBEE vol. 19. IACSIT Press, Singapore
Farhan Y, Nawaiseh S (2015) Spatial assessment of soil erosion risk using RUSLE and GIS techniques. Environ
Earth Sci 74(6):4649–4669
Foster GR, Yoder DC, Weesies GA, McCool DK, McGregor KC, Bingner R (2008) Draft user’s guide, revised
15 universal soil loss equation version 2 (RUSLE-2), Washington, DC
Girmay G, Singh BR, Nyssen J, Borrosen T (2009) Runoff and sediment-associated nutrient losses under
different land uses in Tigray, Northern Ethiopia. J Hydrol 376 (1–2):70–80
Jakhar P, Dass A, Adhikary PP, Sudhishri S, Naik BS, Hombegowda HC, Madhu M, Lenka NK, Chaudhary
PR, Panda RK (2016) Multitier agroforestry system for integrated resource conservation on uplands of
Eastern Ghats region in India. Agrofor Syst 91(4):697–712
Jung P, Ko M, Im J, Um K, Choi D (1983) Rainfall erosion factor for estimating soil loss. Korean J Soil Sci
Fert 16(2):112–118
Kinnell PIA (2010) Event soil loss, runoff and the universal soil loss equation family of models: a review. J
Hydrol 385:384–397
Kravchenko A, Bullock DG (1999) A comparative study of interpolation methods for mapping soil properties.
Agron J 91:393–400
Lal R (1998) Soil erosion impact on agronomic productivity and environmental quality: critical review. Plant
Sci 17:319–464
Lee JH, Heo JH (2011) Evaluation of estimation methods for rainfall erosivity based on annual precipitation in
Korea. J Hydrol 409:30–48
13
Natural Hazards

Lenka NK, Dass A, Sudhishri S, Patnaik US (2012) Soil carbon sequestration and erosion control potential
of hedgerows and grass filter strips in sloping agricultural lands of eastern India. Agric Ecosyst Environ
158:31–40
Lo A, El-Swaify SA, Dangler EW, Shinshiro L (1985) Effectiveness of ­EI30 as an erosivity index in Hawaii. In:
El-Swaify SA, Moldenhauer WC, Lo A (eds) Soil erosion and conservation. Soil Conservation Society of
America, Ankeny, pp 384–392
Loureiro ND, Coutinho MD (2001) A new procedure to estimate the RUSLE E ­ I30 index based on monthly rain-
fall data and applied to the Algarve region, Portugal. J Hydrol 250:12–18
Madhu M, Naik BS, Jakhar P, Hombegowda HC, Adhikary PP, Gore KP, Barman D, Naik GB (2016) Compre-
hensive impact assessment of resource conservation measures in watershed of eastern region of India. J
Environ Biol 37:391–398
Mueller TG, Pierce FJ, Schabenberger O, Warncke DD (2001) Map quality for site-specific fertility manage-
ment. Soil Sci Soc Am J 65(5):1547–1558
Naik BS, Paul JC, Panigrahi B, Sahoo BC (2015) Soil erosion assessment from farming lands of Eastern
Ghats region of Odisha. Ind J Soil Cons 43(1):33–37
Nash JE, Sutcliffe JE (1970) River flow forecasting through conceptual model. J Hydrol 10:282–290
Nyssen J, Vandenreyken H, Poesen J, Moeyersons J, Deckers J, Haile M, Salles C, Govers G (2005) Rainfall
erosivity and variability in the Northern Ethiopian Highlands. J Hydrol 311:172–187
Panagos P, Ballabio C, Borrelli P, Meusburger K, Klik A, Rousseva S, Tadić MP, Michaelides S, Hrabalíková
M, Olsen P, Aalto J, Lakatos M, Rymszewicz A, Dumitrescu A, Beguería S, Alewell C (2015) Rainfall
erosivity in Europe. Sci Total Environ 511:801–814
Panagos P, Ballabio C, Borrelli P, Meusburger K (2016) Spatio-temporal analysis of rainfall erosivity and ero-
sivity density in Greece. Catena 137:161–172
Prasannakumar V, Shiny R, Geetha N, Vijith H (2011) Spatial prediction of soil erosion risk by remote sens-
ing, GIS and RUSLE approach: a case study of Siruvani river watershed in Attapady valley, Kerala, India.
Environ Earth Sci 64(4):965–972
Prasannakumar V, Vijith H, Abinod S, Geetha N (2012) Estimation of soil erosion risk within a small mountain-
ous sub-watershed in Kerala, India, using revised universal soil loss equation (RUSLE) and geo-informa-
tion technology. Geosci Front 3(2):209–215
Praveen R, Kumar U (2011) Integrated approach of Universal Soil Loss Equation (USLE) and Geographical
Information System (GIS) for soil loss risk assessment in Upper South Koel Basin, Jharkhand. J Geo-
graphic Inf Syst. https​://doi.org/10.4236/jgis.2012.46061​
Raghunath B, Erasmus IE (1971) A method for estimating erosion potential from daily rainfall data. Indian For
97(3):121–125
Rambabu, Tejwani KG, Agarwal MC, Chandra S (1978) Rainfall erosion potential and iso-erodent map of
India. Bulletin No. 2. Central Soil and Water Conservation Research and Training Institute, Dehradun,
India
Rawat J, Joshi R, Mesia M (2013) Estimation of erosivity index and soil loss under different land uses in the
tropical foothills of Eastern Himalaya (India). Trop Ecol 54(1):47–58
Renard KG, Freimund JR (1994) Using monthly precipitation data to estimate the R-factor in the revised USLE.
J Hydrol 157:287–306
Renard KG, Foster GR, Weesies GA, McCool DK, Yoder DC (1997) Predicting soil erosion by water: a guide
to conservation planning with the Revised Universal Soil Loss Equation (RUSLE). USDA Agriculture
Hand book No. 703. Washington DC
Richardson CW (1983) Estimation of erosion index from daily rainfall amounts. Trans Am Soc Agric Eng
26:153–156
Sanchez-Moreno JF, Mannaerts CM, Jetten V (2014) Rainfall erosivity mapping for Santiago Island, Cape
Verde. Geoderma 217–218:74–82
Selker JS, Haith DA, Reynolds JE (1990) Calibration and testing of daily rainfall erosivity model. Trans Am
Soc Agric Eng 33:1612–1618
Sepaskhah AR, Sarkhosh P (2005) Estimating storm erosion index in southern region of IR Iran. Iran J Sci
Technol Trans B Eng 31(B2):237–248
Sheridan GJ, Rosewell CJ (2003) An improved Victorian erosivity map. Aust J Soil Res 41(1):141–149
Sudhishri S, Patnaik US (2004) Erosion index analysis for Eastern Ghat High Zone of Orissa. Indian J Dryland
Agric Res Dev 19:42–47
Tiwari H, Rai SP, Kumar D, Sharma N (2015) Rainfall erosivity factor for India using modified Fourier index. J
Appl Water Eng Res. https​://doi.org/10.1080/23249​676.2015.10640​38
Wischmeier WH, Smith DD (1965) Prediction rainfall erosion losses from cropland east of the Rocky
mountains: a guide for selection of practices for soil and water conservation. Agricultural Handbook
No. 282, U.S. Department of Agriculture, Washington, DC

13
 Natural Hazards

Wischmeier WH, Smith DD (1978) Predicting rainfall erosion. USDA Agriculture Handbook No. 537. Wash-
ington D.C
Wu L, Liu X, Ma X (2016) Spatiotemporal distribution of rainfall erosivity in the Yanhe River watershed of
hilly and gully region, Chinese Loess Plateau. Environ Earth Sci 75:315. https​://doi.org/10.1007/s1266​
5-015-5136-6
Yang F, Lu C (2015) Spatiotemporal variation and trends in rainfall erosivity in China’s dry land region during
1961–2012. Catena 133:362–372
Yu B, Rosewell CJ (1996) An assessment of a daily rainfall erosivity model for New South Wales. Aust J Soil
Res 34:139–152

Publisher’s Note  Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

13