A multi-scale study of wind erosion susceptibility along the South African Wild Coast
Date
2024-03
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch : Stellenbosch University
Abstract
ENGLISH ABSTRACT: Wind erosion is a significant driver of land degradation, affecting over a third of all land areas in recent centuries. Accelerated wind erosion in South Africa has caused severe localised land degradation, similar to that observed in parts of the ecologically important Wild Coast region. This erosion-induced degradation has led to localised desertification and poses risks to vulnerable wetland and river ecosystems. In data-sparse regions such as the Wild Coast, identifying highly susceptible areas becomes crucial to mitigate the detrimental effects of accelerated wind erosion. This study aims to determine the spatial distribution of wind erosion features along the Wild Coast, investigate factors influencing their occurrence and growth, and model the area's future susceptibility to wind erosion.
Historical aerial photography, Google EarthTM imagery, and multi-temporal mapping spanning an 85-year period were utilised to create a wind erosion inventory map. It revealed an uneven spatial distribution of wind erosion sites, primarily clustered within a 2 km stretch along the coastal study area. These sites were concentrated in specific locations such as Xolobeni, Mkambati, Mngazi River Mouth - Noxova - Mbolompo Point, Wavecrest, and Kei River Mouth. Human activities in wind-exposed areas, such as disturbed agricultural fields, bare patches in grasslands, informal sand mines, and tracks, were identified as the key locations where these features initiated. Over the 85-year period, some erosion features expanded significantly, while others remained relatively stable due to the establishment of peripheral vegetation that acts as wind barriers.
Long-term remote sensing analyses focused on the Xolobeni area, a representative subset of the broader Eastern Cape Wild Coast study region, aimed to comprehend the influence of long-term changes in land cover, vegetation status, soil texture, and soil moisture conditions on the occurrence and evolution of wind erosion features. This analysis utilised multi-temporal Landsat 5 Thematic Mapper (L5 TM) and Landsat 8 Operational Land Imager (OLI) imagery covering the period from 1987 to 2020, in conjunction with available topographical data from 1982, 1993, and 2004. The application of the Random Forest classifier successfully mapped land cover for the years 1987, 1991, 1999, 2004, 2010, 2015, and 2020, achieving overall accuracies exceeding 80.00% and Kappa indices surpassing 0.77 for each of these seven years. A primary finding of the land cover change assessment reveals the susceptibility of degraded grasslands to wind erosion and noted a rapid expansion of wind erosion features between 1987 and 1999, followed by a subsequent period of stability. The analysis of computed time-series Normalised Difference Vegetation Index (NDVI), Topsoil Grain Size Index (TGSI), and Normalised Difference Moisture Index (NDMI) data revealed that regions impacted by wind erosion consistently exhibited lower NDVI values, indicating reduced vegetation cover conditions, reaffirming the influence of vegetation on wind erosion development. Higher TGSI values denoted areas associated with higher wind erosion susceptibility and emphasised the significance of sandy soils with reduced clay content in erosion vulnerability. Lower NDMI values associated with affected regions highlighted that drier soil conditions promote wind erosion processes. The multi-temporal analyses of topographical data revealed that abandoned cultivated lands and zones with high track density were prone to erosion, highlighting a connection between human activities and wind erosion susceptibility in the study region.
General concepts of the Wind Erosion Equation were adopted in this study to map the regional wind erosion susceptibility conditions. Two regional susceptibility methods were implemented and compared. Model 1 employed a geostatistical approach, based on erosion factor class frequency ratio data and Analytical Hierarchy Process importance weights. Model 2 utilised the data-driven Weights of Evidence modeling technique. Model 1 classified large areas of the study area as having low susceptibility (46%), while Model 2 classified more than 90% of the areas as very low susceptible zones. Both models show that less than 4% of the study region has a high to very high susceptibility to wind erosion. In general, areas associated with higher wind erosion susceptibility are poorly vegetated, wind-exposed coastal zones characterised by unconsolidated, erodible sandy soils. Model 1 and Model 2 are associated with area under the receiver operating characteristic curve values of 0.987 and 0.946, respectively, displaying satisfactory average performances.
Recommendations for combating wind erosion along the ecologically sensitive Wild Coast include avoiding wind-aligned trackways, protecting existing vegetation, minimising bare soil patches in vulnerable areas, and establishing indigenous vegetation barriers. Utilisation of the developed wind erosion inventory and susceptibility maps will aid stakeholders in developing targeted conservation strategies required to shield vulnerable regions from further degradation.
AFRIKAANSE OPSOMMING: Winderosie is 'n beduidende drywer van gronddegradasie, wat meer as 'n derde van alle landoppervlaktes in die afgelope eeue beïnvloed het. Versnelde winderosie in Suid-Afrika het ernstige gelokaliseerde gronddegradasie teweeg gebring, soortgelyk aan dié wat in dele van die ekologies belangrike Wildekusstreek waargeneem word. Hierdie erosie-geïnduseerde degradasie het gelei tot gelokaliseerde woestynvorming en hou risikos in vir kwesbare vleiland- en rivierekosisteme. In data-skaars streke, soos die Wildekus, is die identifisering van hoogs kwesbare gebiede noodsaaklik om die nadelige gevolge van versnelde winderosie te bekamp. Die doel van die studie is om die landelike verspreiding van winderosie-kenmerke langs die Wildekus te bepaal, om faktore wat hul voorkoms en groei beïnvloed te ondersoek en die gebied se toekomstige blootstelling aan winderosie te modelleer. Historiese lugfotografie, Google EarthTM-beelde en multi-temporele kartering oor 'n tydperk van 85 jaar, is gebruik om 'n winderosie-inventariskaart te skep. Dit het 'n onewe landelike verspreiding van winderosieterreine aan die lig gebring, hoofsaaklik gegroepeer binne 'n 2 km-strook langs die kusstudiegebied. Hierdie gebiede is gekonsentreer op spesifieke plekke soos Xolobeni, Mkambati, Mngaziriviermond - Noxova - Mbolompopunt, Wavecrest en Keiriviermond. Menslike aktiwiteite in wind-blootgestelde gebiede, soos versteurde landbouvelde, ontblote kolle in grasvelde, informele sandmyne en spore, is geïdentifiseer as die sleutelplekke waar hierdie kenmerke hul oorsprong gehad het. Oor die verloop van 85 jaar het sommige erosie kenmerke aansienlik uitgebrei, terwyl ander relatief stabiel gebly het as gevolg van die vestiging van perifere plantegroei wat as windversperrings dien. Die doel van langtermyn afstandwaarnemingsontledings wat gefokus was op die Xolobeni-gebied, 'n verteenwoordigende onderafdeling van die breër Oos-Kaapse Wildekus-studiestreek, was om die invloed van langtermynveranderinge in grondbedekking, plantegroeistatus, grondtekstuur en grondvogtoestande op die voorkoms en evolusie van winderosie-kenmerke te ondersoek. Hierdie ontleding het gebruik gemaak van multi-temporele Landsat 5 Thematic Mapper (L5 TM) en Landsat 8 Land Imager (OLI) beelde wat die tydperk van 1987 tot 2020 dek, in samewerking met beskikbare topografiese data van 1982, 1993 en 2004. Die toepassing van die Random Forest-klassifiseerder het grondbedekking vir die jare 1987, 1991, 1999, 2004, 2010, 2015 en 2020 suksesvol gekarteer, met ’n algehele akkuraatheid van meer as 80.00% en Kappa-indekse wat 0.77 vir elk van hierdie sewe jaar oorskry.
AFRIKAANSE OPSOMMING: Winderosie is 'n beduidende drywer van gronddegradasie, wat meer as 'n derde van alle landoppervlaktes in die afgelope eeue beïnvloed het. Versnelde winderosie in Suid-Afrika het ernstige gelokaliseerde gronddegradasie teweeg gebring, soortgelyk aan dié wat in dele van die ekologies belangrike Wildekusstreek waargeneem word. Hierdie erosie-geïnduseerde degradasie het gelei tot gelokaliseerde woestynvorming en hou risikos in vir kwesbare vleiland- en rivierekosisteme. In data-skaars streke, soos die Wildekus, is die identifisering van hoogs kwesbare gebiede noodsaaklik om die nadelige gevolge van versnelde winderosie te bekamp. Die doel van die studie is om die landelike verspreiding van winderosie-kenmerke langs die Wildekus te bepaal, om faktore wat hul voorkoms en groei beïnvloed te ondersoek en die gebied se toekomstige blootstelling aan winderosie te modelleer. Historiese lugfotografie, Google EarthTM-beelde en multi-temporele kartering oor 'n tydperk van 85 jaar, is gebruik om 'n winderosie-inventariskaart te skep. Dit het 'n onewe landelike verspreiding van winderosieterreine aan die lig gebring, hoofsaaklik gegroepeer binne 'n 2 km-strook langs die kusstudiegebied. Hierdie gebiede is gekonsentreer op spesifieke plekke soos Xolobeni, Mkambati, Mngaziriviermond - Noxova - Mbolompopunt, Wavecrest en Keiriviermond. Menslike aktiwiteite in wind-blootgestelde gebiede, soos versteurde landbouvelde, ontblote kolle in grasvelde, informele sandmyne en spore, is geïdentifiseer as die sleutelplekke waar hierdie kenmerke hul oorsprong gehad het. Oor die verloop van 85 jaar het sommige erosie kenmerke aansienlik uitgebrei, terwyl ander relatief stabiel gebly het as gevolg van die vestiging van perifere plantegroei wat as windversperrings dien. Die doel van langtermyn afstandwaarnemingsontledings wat gefokus was op die Xolobeni-gebied, 'n verteenwoordigende onderafdeling van die breër Oos-Kaapse Wildekus-studiestreek, was om die invloed van langtermynveranderinge in grondbedekking, plantegroeistatus, grondtekstuur en grondvogtoestande op die voorkoms en evolusie van winderosie-kenmerke te ondersoek. Hierdie ontleding het gebruik gemaak van multi-temporele Landsat 5 Thematic Mapper (L5 TM) en Landsat 8 Land Imager (OLI) beelde wat die tydperk van 1987 tot 2020 dek, in samewerking met beskikbare topografiese data van 1982, 1993 en 2004. Die toepassing van die Random Forest-klassifiseerder het grondbedekking vir die jare 1987, 1991, 1999, 2004, 2010, 2015 en 2020 suksesvol gekarteer, met ’n algehele akkuraatheid van meer as 80.00% en Kappa-indekse wat 0.77 vir elk van hierdie sewe jaar oorskry.
Description
Thesis (DPhil)--Stellenbosch University, 2024.