Erosion risk index: Correlation of ROM-scale and Mackintosh Probe at Sungai Langat tributary

The assessment of soil erosion risk has been widely studied, and various methods have been established. However, most of the methods require extensive field and laboratory works that are time-consuming. The previous literature mainly focused on utilizing the existed empirical models like USLE and its derivatives. The establishment of empirical correlation can increase the efficiency to estimate the potential of soil erosion in a shorter time. This study was performed to develop an empirical correlation between the number of blows per unit penetration (M-value) obtained from Mackintosh Probe test and ROM-scale number from soil particle size distribution analysis. Both tests were conducted at three different points at one of Sg. Langat tributary riverbank nearby Universiti Kebangsaan Malaysia, Bangi. The soil at each location was analyzed at three different layers, from 0 m to 1.5 m with intervals of 0.3 m and 0.5 m for Mackintosh Probe test and ROM-scale, respectively. The result indicates that ROM-scale number is inversely proportional to M-value with a coefficient of determination of 0.5885.


Introduction
Soils in general, are earthy materials that made up of minerals, organic matter, living organisms, gas, and water. Soil is very important as it is a natural medium for planting crop and act as the foundation for construction. Soils are also known to have a heterogeneous composition, and residual soils in Malaysia are made up of different compositions of soil particles, namely sand, silt and clay [1]. More than three-quarters of the land in Peninsular Malaysia is covered by residual soils [2]. This physical property is directly related to its texture and particle size distribution which means the soil heterogeneity influences the process of soil erosion.
The occurrence of soil erosion is a serious worldwide geological problem. Determination of soil erosion risk is crucial to make sure that the preventive measurements can be taken appropriately before any catastrophic events happen that may lead to the loss of life and change the surrounding ecosystem. A lot of studies around the world have implemented the empirical model such as Universal Soil Loss Equation (USLE) and its derivative, namely Modified Universal Soil Loss Equation (MUSLE) and Revised Universal Soil Loss Equation (RUSLE), to estimate soil loss. However, these methods require a large-scale laboratory and field works as external factors like rain erosivity, soil erodibility, slope and steepness, cropping management and conservation practice need to be considered [3][4][5]. Thus, developing a soil erosion risk index can be an alternative to estimate the future occurrence of soil erosion through the development of a correlation between the existing and conventional methods.
To simplify the assessment of soil erosion, ROM-scale was developed in 2001 by Malaysian researchers (Roslan and Mazidah) to categorize the degree of soil erodibility using the percentage of sand, silt, and clay of the soil sample. This index focused on categorizing one of the governing erosion factors proposed by Hudson in 1979: the soil erodibility (Kfactor) using the particle size distribution [6]. ROM-scale can be evaluated using the mathematical expression as in Equation 1, and its corresponding degree of soil erodibility can be classified into different levels, as shown in Table 1 [7].

EI
% sand % silt 2 % clay (1)  The local researchers have used ROM-scale as this scale was developed specifically for soil series in Malaysia that can cover different conditions of SI location. For instance, some studies had been conducted at the riverbank [7] and slopes [8][9][10]. Furthermore, although this method is specifically for Malaysian soil series, it was recently adapted to one of the bank erosion studies conducted in India [11].
Soil stiffness also contributes to the possibilities of soil erosion events aside from the soil erodibility factor. Mackintosh Probe (MP) test is a conventional site investigation (SI) method that is widely used to find soil strength. This method is relatively simple in terms of the execution process. The soil strength is directly reflected by counting the number of blows for every 0.3 m penetration into the soil. However, there are limitations to this method as it cannot penetrate hard layers and is not suitable for hard clay or soils containing gravel or cobbles [12]. Besides, the test needs to stop when the depth reaches 15 m, or the number of blows reaches 400. In 1994, Sabtan and Shehata have established the correlation between the consistency of cohesive soil based on the M-value from MP test. They reported that the Mvalue could be regarded as the soil stiffness as its consistency can be related to the unconfined compressive strength and undrained shear strength as shown in Table 2 [13].   [14], Standard Penetration Test (SPT) together with resistivity test for soft soil assessment [15] and Vane Shear Test to find undrained shear strength [16]. However, the description of the correlation between Mackintosh Probe test and ROM-scale for soil erosion risk index is limited in the literature. Thus, the developed correlation from this case study will allow future work to assess soil erosion potential using a simpler method. In addition to that, the tedious and timeconsuming field work and laboratory testing needed to evaluate the ROM-scale can be skipped.

Site investigation (SI) -Mackintosh Probe (MP) test and collection of soil sample for ROM-scale evaluation
The location to conduct this case study is at one of Sg. Langat tributary riverbank nearby . and its simplified schematic illustration shows the current condition of SI

Laboratory testing -Sieve analysis and hydrometer test
A portion of soil samples was collected from the SI locations (separated based on the location and soil depth) had been subjected to a 24-hour drying process before conducting laboratory testing for particle size distribution analysis, as shown in Fig. 3. The dried soil samples then were submitted for sieve analysis. The fines particles of the soil samples that pass through the 75μm sieve opening were collected and used for the second laboratory test, the hydrometer test. Sodium hexametaphosphate (Na 6 (PO 3 ) 6 ) and sodium carbonate (NaCO 3 ) were used as a dispersing agent during the hydrometer test. All laboratory testing was conducted according to the British Standard (BS 1377) system. Based on British Standard, the particle size (

Mackintosh Probe test
The Mackintosh Probe test results were tabulated as in Table 4, and the graph of M-value versus soil depth was illustrated in Fig. 4. According to Sabtan and Shehata (1994), the M-value for every 0.3 m of penetration obtained at different location can be directly used to categorize the consistency of the soil at each layer based on  Based on Fig. 4, the trends show that for the first layer (topsoil to 0.3 m) of MP test, Location 1 has stiffer soil compared to Point 2 and 3. Still, its strength drops slightly but remains within the medium soil consistency as the cone penetrometer penetrates deeper into the soil until 1.2 m. The soil consistency for Location 1 after that drastically dropped from medium to a soft region when the cone penetrometer travelled from 1.2 m to 1.5 m, which gives the smallest M-value (15 blows) for this case study. In contrast, the trend for Location 2 and 3 show that they have soft soil from the topsoil until 1.2 m deep. After the depth of 1.2 m, Location 2 and 3 became stiffer as M-value exceeded the limit for soft soil, which is 33 blows for every 0.3 m penetration. In general, the soil at this SI location falls between soft to medium consistency.

Soil sample -Particle size distribution analysis and ROM-scale
In total, there are nine soil samples collected from three different locations at the interval of 0.5 m deep, as shown in Fig. 5. The physical appearance of the samples does reflect its composition, especially for sandy soil. It can be seen from the picture, Location 3 (1.0 m -1.5 m) seems like it contains the most sand particles compared to the rest of the soil samples. However, it is a bit tricky and difficult to gauge silt and clay content in a soil sample with naked eyes.  Through the sieve analysis and hydrometer test, the percentage of sand, silt and clay can be determined from the particle size distribution curve ( Fig. 6-8) and later was compared to the given particle size according to British Standard (BS) as tabulated in Table 3. The experimental results and the computed ROM-scale values are tabulated in Table 5. The soil compositions from this case study were grouped and categorized into three different levels, while ROM-scale were classified based on Table 1.    The soil samples from Location 1 have a relatively high percentage of clay and a low percentage of sand, resulting in a low ROM-scale at this location. Low ROM-scale means that the soil at this location is less susceptible to erosion since it has a low erodibility degree. According to the results from MP test in Table 4, the M-values for this location are relatively higher compared to Location 2 and 3, starting from topsoil until 1.2 m deep. Higher M-values indicates that the soil in Location 1 within the range of 0.0 m-1.2 m is stiffer and hard to erode, as suggested by ROM-scale.
The soil at Location 2 contained a low clay percentage for each layer but with a combination of a moderate and high percentage of sand. ROM-scale shows that the soil at this location has a moderate risk of soil erosion and has relatively low M-values. Low Mvalues show that the soil at Location 2 is softer than soil at Location 1 and easier to get eroded.
Although the trend at Location 3 shows a slight inconsistency between sand/clay content, ROM-scale, and corresponding M-values, the results still show the same pattern as the soil samples from Locations 1 and 2. The soil at Location 3 still has high clay when the percentage of sand is low, resulting in a low ROM-scale and vice versa. Lower ROM-scale is because clay and sand hold the most significant percentage in every soil composition, making the effect of silt content less meaningful in evaluating ROM-scale. As per MP test results, Location 3 fell into the soft region except when the depth went beyond 1.2 m, just like soil at Location 2.
Therefore, from the analysis of experimental results, the percentage of clay and sand is inversely related. Clay is classified as cohesive soil, while sand is cohesionless soil. Sand particles are prone to detachment compared to clay as the cohesion force between the clay soil particles impacts the soil detachability [17]. Thus, the soil at Location 1 is less vulnerable to erosion than Location 2 and 3 because the soil at Location 1 has lesser loose soil grains.

Correlation between M-value and ROM-scale
The soil series in Malaysia can be divided into three different texture layers at the interval of 0.5 m deep, namely layer A (surface soil), B (subsoil), and C (substratum) [18]. Thus, the Mvalue at 0.3 m was chosen to indicate the M-value from the MP test at layer A, 0.9 m for layer B, and 1.2 m for layer C. These depths for MP are reasonable as they are in between the represented layers by ROM-scale. This adjustment was made to develop the one-to-one relationship between M-values with ROM-scale since the MP test and soil samples collection were not done on the same scale. Hence, Table 6 summarizes the selected M-values and the corresponding ROM-scale for each layer, while Fig. 8 shows the linear regression of a scattered plot between the two parameters.  Based on the scattered plot between M-value and ROM-scale, these two parameters can be correlated as inversely proportional. The linear regression for this correlation has a coefficient of determination (R 2 ) of 0.5885. Aminaton (1996) stated that if the range of R2 is between 0.25 to 0.55, the coefficient of determination is considered moderately good [19]. From Fig. 9, there are two outliers in the dataset which belong to Location 3-A and 3-C. At Location 3-A, the M-value is 30 while the ROM-scale is equal to 0.72, and at Location 3-C, the M-value is 20 while ROM-scale is equivalent to 2.84. These outlier data might be existed due to human or instrument error during the MP test and laboratory testing. This is because MP tests and laboratory testing are highly dependent on human accuracy. This might be the case for outlier at Location 3-C where supposedly the M-value should be lesser when the percentage of sand is higher. The cone penetrometer can easily probe into the soil that contains a lot of loose-particle soil.
The moisture content of the soil might also contribute to the inconsistency of these results for soils in Location 3 as MP test does get affected by the saturation of soil. For Location 3-A, the percentage of clay is 40.97% with the corresponding 17.07% of moisture content, while soil at Location 1-B has 40.69% of clay and 20.36% moisture content. This implies that Location 3-A has drier soil compared to Location 1-B, with a relatively same percentage of clay. The Location 3-A ideally should have a greater M-value To match with the regression line, as it suggests that it is harder for Mackintosh Probe to penetrate the dry clayey soil.

Conclusion
The Mackintosh Probe (MP) test and the gradation of soil analysis using sieve and hydrometer test has been carried out for soil samples collected from Sungai Langat riverbank. The soil erosion index using ROM-scale has been determined from the result obtained from a series of laboratory testing. Based on the result, the M-value from MP test is correlated with the evaluated value of ROM-scale of the soil sample. The soil with high sand amount has a lower M-value, whereas the soil with a clay amount has a higher M-value. By correlating the M-value with ROM-scale at different soil locations, it was found that these two data are inversely proportional with the coefficient of determination (R 2 of 0.5885). More data will be needed in the future to improve the accuracy of this correlation with the precaution measures taken during SI and laboratory testing to avoid experimental and human error. Nevertheless, this study indicates that the assessment of soil erosion index using ROM-scale