Assessing Soil Erosion Hazards Using Land-Use Change and Landslide Frequency Ratio Method: A Case Study of Sabaragamuwa Province, Sri Lanka

Abstract

This study aims to identify the vulnerable landscape areas using landslide frequency ratio and land-use change associated soil erosion hazard by employing geo-informatics techniques and the revised universal soil loss equation (RUSLE) model. Required datasets were collected from multiple sources, such as multi-temporal Landsat images, soil data, rainfall data, land-use land-cover (LULC) maps, topographic maps, and details of the past landslide incidents. Landsat satellite images from 2000, 2010, and 2019 were used to assess the land-use change. Geospatial input data on rainfall, soil type, terrain characteristics, and land cover were employed for soil erosion hazard classification and mapping. Landscape vulnerability was examined on the basis of land-use change, erosion hazard class, and landslide frequency ratio. Then the erodible hazard areas were identified and prioritized at the scale of river distribution zones. The image analysis of Sabaragamuwa Province in Sri Lanka from 2000 to 2019 indicates a significant increase in cropping areas (17.96%) and urban areas (3.07%), whereas less dense forest and dense forest coverage are significantly reduced (14.18% and 6.46%, respectively). The average annual soil erosion rate increased from 14.56 to 15.53 t/ha/year from year 2000 to 2019. The highest landslide frequency ratios are found in the less dense forest area and cropping area, and were identified as more prone to future landslides. The river distribution zones Athtanagalu Oya (A-2), Kalani River-south (A-3), and Kalani River- north (A-9), were identified as immediate priority areas for soil conservation.

GIS and RS Based Soil Erosion Modelling in Sri Lanka: A Review

“…Thus, K values of above ~85% of studies are more accurate while K values of Dissanayake et al (2019) and Wijesundara et al (2018) are the most accurate according to the explanation by Renard et al (1991). (Wijesekara and Samarakoon, 2001) From available literature of previous studies (Joshua, 1977) 02 (Senanayake et al, 2013) From available literature of previous studies (Bandara and Somasiri, 1991) (Zijister, 1989) 03 (Jayarathne et al, 2010) From available literature of previous studies 04 (Senanayake et al, 2020) From available literature of previous studies (Fayas et al, 2019) (Senanayake et al, 2013) (Wijesekara and Samarakoon, 2001) 05 (Dissanayake et al, 2019) Nomography is used to compute the values of K based on soil properties (Wischmeier and Smith, 1978) (Ganasri and Ramesh, 2016) (Mapa et al, 2010) 06 (Panditharathne et al, 2019) From available literature of previous studies (Jayarathne et al, 2010) (Wijesundara et al, 2018) (Fayas et al, 2019)…”

“…(Senanayake et al, 2013) The LS layer has derived directly from slope and flow accumulation layers, both of which can be computed using the digital elevation model (DEM). 02 (Senanayake et al, 2020) 03 (Jayarathne et al, 2010) L = (μ/72.6)m…”

“…As shown in Table 05, out of the above mentioned two methodologies, ~93% of the Sri Lankan studies (Piyathilake et al, 2020;Senanayake et al, 2020;Dias et al, 2019;Fayas et al, 2019;Panditharathne et al, 2019;Jayasekara and Kadupitiya, 2018;Lekamge et al, 2018;Jayarathne et al, 2010;Thuraisingham and Weerasinghe, 2009;Udayakumara and Gunawardena, 2016;Wijesundara et al, 2018;Wijesekara and Samarakoon, 2001) have used previous literature to determine C factor which is the simplest and time-saving method since the NDVI method requires advance RS applications viz. satellite imagery, aerial photography, and image processing.…”

“…Although ~93% of studies have incorporate C factor values to model soil erosion using GIS/ RS technologies, Senanayake et al (2013) explains that, since the C factors are highly localized and vary at individual farm level, its effects are (Munasinghe et al, 2001) (Biswas and Pani, 2015) 02 (Jayarathne et al, 2010) Not mentioned 03 (Wijesundara et al, 2018) (Munasinghe et al, 2001) (Senanayake et al, 2013) 04 (Fayas et al, 2019) ( Munasinghe et al, 2001) (Senanayake et al, 2013) (Wijesundara et al, 2018 05 (Thuraisingham and Weerasinghe, 2009) 06 (Jayasekara and Kadupitiya, 2018) (Kuok et al, 2013) 07 (Dias et al, 2019) (Udayakumara and Gunawardena, 2016) (Weerasinghe et al, 2016) 08 (Udayakumara and Gunawardena, 2016) (Kuok et al, 2013) 09 (Lekamge et al, 2018) Not mentioned 10 (Piyathilake et al, 2020) (Biswas and Pani, 2015) (Kuok et al, 2013) (Munasinghe et al, 2001) (Senanayake et al, 2013) (Udayakumara and Gunawardena, 2016) (Weerasinghe et al, 2016) ) (Wijesundara et al, 2018) 11 (Wijesekara and Samarakoon, 2001) Not mentioned 12 (Senanayake et al, 2020) (Senanayake et al, 2013) (Munasinghe et al, 2001 scattered and it will not vital to apply C factor for regional-scale studies. Thus, the only land cover factor has been taken into consideration.…”

“…For instance, Senanayake et al (2013) has given C factor values concerning land management practices as well and in this case, the P factor is omitted. Moreover, ~86% of Sri Lanka soil erosion modeling studies have used previously published data for obtaining P factor values for each land use and land cover classes (Wijesekara and Samarakoon, 2001;Thuraisingham and Weerasinghe, 2009;Udayakumara and Gunawardena, 2016;Lekamge et al, 2018;Jayasekara and Kadupitiya, 2018;Wijesundara et al, 2018;Dias et al, 2019;Fayas et al, 2019;Panditharathne et al, 2019;Piyathilake et al, 2020;Senanayake et al, 2020).…”