Rainfall erosivity assessment over a flooding basin, Kelani River basin, Sri Lanka

Soil erosion is a process of nature causing the displacement and conveyance of the uppermost layer of soil to low-lying regions (Karunaratne et al. 2022; Chen et al. 2023). This is a complicated phenomenon that is influenced by many kinds of interrelated factors. Rainfall and land use changes are two key factors of them because they represent a critical pathway for sediment mixing with rainwater and subsequent deposition processes (Xiao et al. 2015; John et al. 2021). The rainfall erosivity (RE) is one of the most crucial elements that describes the erosive processes and suggests conservation strategies when utilizing soil erosion simulations (Panagos et al. 2017a). The factor, referred to as RE, is incorporated into indicating the capability of rainfall to induce soil erosion (Zhu et al. 2021; de Sousa Teixeira et al. 2022). That is one significant reason to select RE in this study.

Studying RE is essential for understanding and predicting the impact of rainfall on soil erosion. By studying RE, it is able to develop models and equations that help estimate soil loss, identify erosion risks, and formulate effective soil conservation measures (Mamedov et al. 2011). Addressing erosion processes occurring at the bed surfaces and banks of natural rivers, as well as engineered waterways, is crucial for effective water resources management (Pandey et al. 2021). This knowledge is crucial in various fields such as agriculture, land management, and environmental planning (Lee & Heo 2011). For farmers and land managers, understanding RE allows them to make informed decisions about how to manage their land in order to mitigate erosion risks (Panagos et al. 2017a; Nabi et al. 2020). Furthermore, studying RE is important for addressing the larger issue of sediment delivery in river catchments. As land clearance, intensification of land use, and population growth increase, sediment delivery in river catchments becomes a significant problem (Li & Ye 2018). Understanding the erosive force of rainfall is necessary to develop effective strategies for sediment control and watershed management (Nazuhan et al. 2018). Furthermore, this analysis can offer insights into the implications of climate change and variability. By examining patterns and trends in RE, experts can gain a deeper understanding of how alterations in precipitation intensity and distribution might influence soil erosion rates across diverse regions. This understanding is essential for devising adaptation strategies and implementing effective soil conservation measures to mitigate the adverse effects of erosion amid a changing climate. Future climate projections project potential threats of soil erosion and its consequences, including land degradation, diminished agricultural yields, and the depletion of soil carbon. Moreover, soil erosion compromises the soil's capacity to mitigate global warming by sequestering carbon (Borrelli et al. 2020; Panagos et al. 2022). The deposits of sediment load in reservoirs downstream have identified significant challenges with the management of water resources and hydraulic structures. That is the importance of studying RE of a particular region. Therefore, this particular research broadly studies one of Sri Lanka's major river basins to showcase this problem.

Various assessment methods are employed to measure RE and its potential impact on soil erosion. One common approach involves using indices such as the R factor in the Universal Soil Loss Equation (USLE) or the RUSLE model. These indices consider factors like rainfall amount, intensity, and distribution to estimate erosive power. Rainfall simulators represent another assessment type, utilizing artificial rainfall systems to study erosive effects under controlled conditions. Additionally, remote sensing techniques analyze rainfall patterns and characteristics over an area to assess RE. These assessment methods are vital for understanding and managing the influence of rainfall on soil erosion.

Other assessment types of RE include the following: (1) The use of RE maps derived from long-term rainfall data to identify areas with high erosivity potential (Xiao et al. 2015; Zhao et al. 2023). (2) Development of regression models or equations that predict RE based on easily obtainable rainfall data, such as average annual precipitation or monthly rainfall patterns (Panagos et al. 2017b; Fenta et al. 2024). (3) Analysis of historical rainfall data and its correlation with erosion measurements to determine the erosive power of different rainfall events (Brychta & Janeček 2017; Di Lena et al. 2021). (4) Several hydrological models, such as the Soil and Water Assessment Tool (SWAT) or the Soil Conservation Service Curve Number (SCS-CN) method, integrate RE assessments into their soil erosion predictions (Gupta et al. 2023; Jadhao et al. 2023).

In the context of Italy, one of the most effective ways to determine a rainstorm's potential for erosive impact is to use the comprehensive rainfall–runoff erosivity factor R, which is included in the USLE and Revised Universal Soil Loss Equation (RUSLE) soil erosion prediction models. USLE-based models are commonly used to estimate soil loss rates at various scales across Europe. Accurate calculation of the R-factor requires continuous, high-temporal-resolution rainfall data, but due to limitations in available data, it is often computed at 15- to 30-min intervals with the use of conversion factors to compensate for underestimations. However, the lack of pluviographic records and the time-consuming nature of R-factor calculation have limited its application in some studies, especially at regional and large-scale levels. Some studies have employed geostatistical methods that incorporate secondary information (elevation and environmental variables) to predict RE spatially, achieving good results in regions like the Algarve in Portugal, the Ebro catchment in Spain, Switzerland, and Greece (Borrelli et al. 2016). A study has focused on assessing RE in São Paulo State, Brazil which employs hierarchical clustering analysis to define homogeneous regions based on RE. This categorization highlights distinct patterns of erosivity across the state. Regionalized regression models are created to estimate RE within each of the defined homogeneous regions, providing accurate tools for predicting RE (Teixeira et al. 2022). There are two main approaches to estimate RE, this method takes into account the kinetic energy of rainfall. It recognizes that higher rainfall kinetic energy corresponds to a greater potential for causing erosion. It considers the energy associated with the falling raindrops. The second approach is simpler and directly uses daily, monthly, or annual rainfall data to estimate RE. It does not factor in the kinetic energy of rainfall but focuses on quantifying erosivity based on the volume or amount of rainfall received over a specific period. This approach is more straightforward and does not involve the specific energy component of the rainfall (Wang et al. 2024). The USLE prediction equation involves RE, or the R-factor, as one of its input parameters. These mean monthly R-factor values are used to predict global monthly erosivity datasets at a 1-km resolution using ensemble machine learning. The resulting monthly raster data in GeoTIFF format has applications in soil erosion prediction modeling, sediment distribution analysis, climate change assessments, and flood and disaster evaluations, and can serve as valuable inputs for Land and Earth Systems modeling (Panagos et al. 2023). Globally, soil erosion is predicted to occur at mean rates of 12–15 Mg ha⁻¹ yr⁻¹ (Marondedze & Schütt 2020). According to the baseline scenario, at a continental level, South America shows the highest prediction of average soil erosion rate (3.53 Mg ha⁻¹ yr⁻¹) in 2001, followed by Africa (3.51 Mg ha⁻¹ yr⁻¹) and Asia (3.47 Mg ha⁻¹ yr⁻¹) (Borrelli et al. 2017). In order to identify the regional variability of soil erosion in farming systems, the study used a time-series analysis of several factors, including rainfall, land use land cover (LULC), and crop diversification. Regression analysis, in conjunction with rain-use efficiency (RUE) and residual trend analysis (RESTREND), was used to separate soil erosion resulting from both human and climate-induced land degradation (Senanayake et al. 2022).

Determining and mapping the Samanalawewa watershed's rate of soil erosion is, thus, the main goal of this project. Using remote sensing and the geographic information system, the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) Sediment Delivery Ratio model was used. The study's findings showed that the watershed's current rate of human-induced soil erosion ranges from 0 to 320 t/ha/year, with an average of 12.51 t/ha/year. Due to irresponsible forest removal, the bulk of the area has experienced land degradation at a rate that exceeds the natural rate of soil erosion (Kumarasiri et al. 2023). The close connection between the Sustainable Development Goals (SDGs) and soil-related activities, emphasizing the crucial role of healthy soil ecosystems in sustaining life on Earth has been studied in another study which has focused on the Central Highlands of Sri Lanka. This is s region vulnerable to climate threats and susceptible to soil erosion. The research employs a RUSLE (Piyathilake et al. 2021; Somasiri et al. 2022) to track and predict soil erosion in response to rainfall variations. It aims to reduce soil erosion in agricultural systems and contribute to the achievement of the SDGs. The study analyzed rainfall trends and erosivity to understand soil erosion dynamics. Using various machine learning techniques like artificial neural networks (ANNs), support vector machines (SVMs), and adaptive network-based fuzzy inference system (ANFIS), (Avand et al. 2022) the research predicts that soil erosion rates will increase (Sahour et al. 2021). Specifically, it highlights a positive trend in RE and extreme rainfall indices. The highest monthly erosion rates are expected in October, and the erosion rates could range from 4 to 22% by 2040 based on 2020 rates. Under different climate scenarios (RCP 2.6 and 8.5), the study predicts an increase in soil erosion rates. The findings underscore the growing vulnerability of the Central Highlands and the threat this poses to achieving the SDGs. In response, the research offers strategies for monitoring and anticipating soil erosion to safeguard the soil ecosystem and support the SDGs. The methodology developed in this study can potentially be applied to reduce soil erosion in agricultural systems in other regions as well.

The country's forests and vegetation have reduced, whereas the amount of land utilized for agriculture in Sri Lanka has expanded over the past three decades (Palliyaguru et al. 2022). It has been shown verification that the Kelani River basin experiences greater rates of soil erosion. Approximately 70% of the river basin area exhibited low to moderate erosion severity (less than 12 tons per hectare per year), highlighting the urgent need for erosion control measures (Fayas et al. 2019). Therefore, it is now essential that soil be conserved in this basin (Kottagoda & Abeysingha 2017). Thus, to obtain an in-depth comprehension of the soil parameters and the circumstances of the riverbed, this study completes the knowledge of the soil erosion pattern of the Kalani River Basin over a period of 31 years.

Therefore, this research aimed to provide evidence of how important sediment deposition in downstream reservoirs is for the long-term management of water resources and hydraulic systems in the Kelani River basin, Sri Lanka. In order to achieve this aim, the specific objective is to estimate the quantitative variations in RE and erosivity density (ED) on a monthly, annual, and seasonal basis. Soil erosivity is unlikely to be significant on a daily basis, which is why the assessment was conducted across these three scenarios.

To fulfill this goal, authors employed two main equations, one internationally recognized, and one locally adapted. After a careful comparison, it was evident that the international Wischmeier and Smith equation provided the most accurate results. Overall, this combination provides a comprehensive framework to find RE and DE, versatility, grounded in an empirical basis, accessible to wide range, and especially longstanding use are made both of these equations special in this study. Not only that the Premalal's equation plays a crucial role in soil erosivity studies in Sri Lanka by providing a locally calibrated model for estimating soil erosion rates (Piyathilake et al. 2021). Its importance lies in its ability to offer more accurate predictions, inform policy and planning decisions, support research and monitoring efforts, and contribute to capacity building in soil erosion management. Therefore, by utilizing this equation, the study successfully achieved the particular aim.

The data used are monthly rainfall data. The data were collected from 10 stations for a period of 31 years (1990–2020) (Renard et al. 1991). The rainfall stations are Angoda, Awissawella, Chesterford, Colombo, Labugama, Maliboda, Norton, Champion, Castlereigh, and Deraniyagala. The locations and elevation of the rainfall stations are given in Table 1. The monthly rainfall data collected were summarized as monthly, annual, and seasonal rainfall data for the assessment of the RE (Cecílio et al. 2013; Parisouj et al. 2020). For the weighted average rainfall, RE, and ED of basin, the Thiessen polygon method was used (Zekai 1998; Милентијевић et al. 2021).

RE results were estimated using two equations as above mentioned(1) Premalal Equation (2). Wisehmeier and Smith equation. The results obtained using both equations separately were compared to obtain the most accurate results (Tesfahunegn et al. 2014; Karunaratne et al. 2022).

The equation developed by Arnoldus (Singh & Singh 2020) requires monthly rainfall data and annual rainfall data in order to calculate the RE. Furthermore, the monthly RE and annual RE can be obtained using this equation. The annual RE varies between 592.84 and 2,831.41 MJ mm ha⁻¹ h⁻¹ yr⁻¹. The highest RE was in 2016 and the lowest RE was in 2001. The annual RE results obtained for the basin are shown in Figure 2(b).

Because the input data are different, there is a significant difference between the two equations. Premalal's equation uses annual average rainfall data, while the Wischmeier and Smith equation uses monthly and annual rainfall data. The Wischmeier and Smith equation yields more accurate answers since it initially determines the monthly RE before calculating the annual RE, whereas Premalal's equation directly calculates the annual RE simultaneously (Cherif et al. 2020; Chen et al. 2023). Table 2 illustrates the corresponding results. When referring to the RE listed in Table 2, the value has dropped in 2016 in Premalal's equation but in Wischmeier and Smith equation, the value has increased by a large amount. Meanwhile considering the reason for this, the annual rainfall of 2016 was 3,012.54 mm and the rainfall in the month of May 2016 was 1,145 mm. As Premalal's equation calculation is based on annual rainfall data, monthly rainfall does not affect the calculation but in Wischmeier and Smith's algorithm the annual RE is calculated by summation of monthly RE, as we can see it depends on monthly rainfall, the annual RE differs concerning monthly rainfall.

As mentioned above in 2016, the RE spiked according to the Wischmeier and Smith algorithm. If we consider the rainfall data of May 2016 it has increased in large value due to Tropical Storm Roanua occurring on 15 May 2016 and causing severe damage to the lives of people as well as properties. Due to the severe landslides and floods, 104 deaths, and 99 missing were recorded and more than 300,000 people were affected. Therefore, we can come to the conclusion the results obtained using Wischmeier and Smith's algorithm are more accurate than Premalal's equation.

As it is discussed, the Kelani River basin in Sri Lanka is one of the most important river basins to the economy of the country. This includes various water resources management processes and direct and indirect relationships to the water resources. Therefore, understanding soil erosion in this tricky catchment is highly essential which was pointed out in the introduction. The sustainability of food production is one of the most important and challenging aspects and is given a higher priority even in the SDGs. Zero hunger is considered the second sustainable development goal and soil erosion can adversely influence agricultural production. Therefore, continuous research has been carried out to understand soil erosion and its correlations to the water sector, agriculture and food sector, energy sector, etc. throughout the world. Being one of the most important catchments in Sri Lanka, the presented research in this paper has a greater potential to look at the required responsible strategies to mitigate any adverse impacts.

As it was stated, continuous research is being undertaken to assess the RE of river basins throughout the world. (de Sousa Teixeira et al. 2023) have modeled the RE over Brazil from 5,166 rain gauge data. This was a comprehensive study carried out to bring up national-scale policies identifying the importance of soil erosion due to precipitation. Understanding the importance of climate change for future scenarios, (Wang et al. 2023) have showcased the projected RE in mainland China. They have used gridded precipitation data with an acceptable bias correction. They have found that the RE is projected to increase due to intensified precipitation scenarios. However, (Feng et al. 2023) looked at the RE trends in Three Gorges Reservoir, China. This is important in order to understand the temporal behavior of RE and then to plan the other activities. Related studies in mountainous catchments are highly important. (Baniya et al. 2023) analyzed the RE for the Himalaya catchment during the Melamchi flood in Nepal. Analyzing sediment dynamics is essential in river basins due to various issues including understanding impacts on hydropower development and turbine blade damage in countries such as Nepal.

Nevertheless, analyzing RE on highly important catchments in Sri Lanka is limited. Data limitation is one of the major reasons for the lack of studies. In a recent study, (de Silva et al. 2022) investigated the spatial variability of soil erosion in the Nalanda Oya catchment in Sri Lanka. The authors concluded the possible use of their findings in the management of soil erosion due to changing climates. The central highlands of Sri Lanka were chosen by (Senanayake et al. 2024) to investigate the RE and then to understand the possible remedy measures for achieving SDGs. However, a comprehensive study of the Kelani River basin is missing in the literature. Therefore, the presented research herein fulfills the research gap in understanding RE in one of the most rainfall-rich catchments in Sri Lanka. Various policy decisions including sustainable urban drainage systems (Srishantha & Rathnayake 2017) have to be implemented in conserving the Kelani River basin from increased RE. These conserving strategies are highly important under climate change.

In this study, we carried out an assessment of RE in the Kelani River basin spanning a 31-year period, from 1990 to 2020. We utilized two distinct equations for this assessment: The Wischmeier and Smith algorithm and Premalal's equation. Using both equations, we calculated the annual RE and subsequently compared the outcomes. Upon a comprehensive evaluation, it became evident that the Wischmeier and Smith algorithm yielded the most accurate results for determining RE in the Kelani River basin. Furthermore, we employed the Wischmeier and Smith algorithm to determine annual, monthly, and seasonal RE, while we applied Kinnell's algorithm to assess ED. In both the estimations of monthly RE and ED, we noted that the month of May 2016 stood out as a peak period by proving the extreme rainfall event that we had on May 2016. The highest monthly RE recorded was 4,762.51 MJ mm ha⁻¹ h⁻¹ yr⁻¹, while the maximum monthly ED reached 2.96 MJ ha⁻¹ h⁻¹. Likewise, the highest annual RE was observed in 2016, totaling 5,132 MJ mm ha⁻¹ h⁻¹ yr⁻¹, coinciding with the period of May. During the same year, the highest ED was 3.64 MJ ha⁻¹ h⁻¹. When examining the seasonal variations, we found that the highest RE occurred in 2016, reaching 558.34 MJ mm ha⁻¹ h⁻¹ yr⁻¹. This peak was observed during the southwest monsoon period. The ED was also at its highest during the same period, at 0.94 MJ ha⁻¹ h⁻¹ in 2016's southwest monsoon period. Both the annual RE and ED reached their highest levels in the year 2016. Monthly values also peaked in May 2016. These findings indicate a significant soil erosion event occurred in that year. Being evidence of these results, on May 15, 2016, a tropical storm collided with Sri Lanka, leading to severe landslides and floods. This catastrophic event resulted in 104 fatalities, and 99 missing individuals, and affected over 300,000 people. This study proves that assessing the RE of a basin is crucial in calculating soil loss using the USLE method. Based on the observed trends in RE, we can conclude that soil erosion in the basin due to rainfall is likely to remain high in the future. Therefore, it is imperative to take necessary measures to mitigate the damage. In summary, the study's results affirm that the methodology employed was consistently and successfully applied to obtain reliable findings by the end of the study. The Kelani River basin, renowned for its provision of various benefits such as potable water, hydropower, agricultural support, and urban utilities, underscores the need for investments in erosion control. This study serves as a valuable tool for crafting ecosystem-centered benefit distribution and financing strategies. By leveraging potential ecosystem benefits identified through such analyses, these investments can be recouped. Moreover, this approach fosters collaboration among diverse stakeholders and socio-economic sectors. The authors are recommending the following in order to enhance understanding in future studies because there exists a notable research gap concerning these scenarios within this specific study area. It is recommended to model soil erosion in this river basin using GIS as well as in sub catchment basis, also to prepare a soil hazard map for whole catchment and to assess soil retention in the Kelani River basin.

The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

This research received no external funding.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.