Fractal Features Of Loess Plateau Landforms In Northern Shaanxi Using Digital Elevation Models

The North Shaanxi Province's Loess Plateau in the past years received considerable attention in the academia because of its historical strategic importance.Several methods have been implemented for the regional classification for the landforms in The Loess Plateau.Few researchers applied Fractal Dimension(FD)to characterise some aspects of geomorphology in this region.However,not sufficient work was conducted to provide regional classified landforms based on FD only.In addition,the consideration of the role of the FD in the Erosion Potential Mapping(EPM)was not considered.Therefore,the research focussed on two main objectives:(1)to explore the ability of fractal dimension to describe the topographical features of loess plateau,which could not be presented by other methods and(2)to propose methodology for regional classifications of the landforms and erosion potential for the loess plateau in Northern Shaanxi based on fractal dimension.Three methods for the estimation of Fractal Dimension(FD)were adopted for use after a careful review of several methods.These included – Box Counting Method(BCM),Higuchi Method(HIG)and Hurst Method(HUR).In application of these methods the BCM was found be most reliable in terms of consistency and accuracy of results and hence adopted as the key method for the results interpretation.The HIG method was also consistent but in some cases failed as a result of its sensitivity to noise in the data input.HUR method overestimated the FD and hence its results were not considered to be reliable.For the evaluation of the Erosion Potential Mapping(EPM),the Analytical Hierarchical Method(AHP).Using the AHP,eight(8)Scales were analysed.They included Logarithmic,Linear,Root Square,Inverse Linear,Balanced-n,Power,Geometric and Adaptive.These different Scales were assessed and analysed for their Consistency Ratio(CR)and Mean Relative Error(MER).The Inverse Linear was adopted as it provided the lowest CR and MER.For the first objective,Catchment Boundary Profiles(CBPs)were generated for each of the watersheds from Advanced Spaceborne Thermal Emission and Reflection Radiometer(ASTER)30m DEM.The FD of each CBPs were determined to characterize the geomorphological features of small watersheds in the research area.The CBPs FDs show a close link with previous studies conducted on the classification of the various types of landform in the research area such as Loess Low Hill,Loess Hill,Loess Rocky Low Hills and Loess Tableland and their phases of anthropogenic and geomorphic activity.Therefore,the fractal dimension of the CBPs were applicable for the characterizing the geomorphic eatures of the Loess plateau.The findings from the CBPs also demonstrated that peaking and harmonic effect of the time series had a direct correlation to the fractal values obtained.This implies the higher the peaking and the harmonic effect within a time series of a CBPs,the higher the FD value obtained.As there was no similar study to confirm this fact,the series of CBPs graph confirmed this result.The next objective was to evaluate the spatial FD of the positive terrain.The results showed that the spatial values of the FD varies from low to higher distribution as you move from Grid Size(GS)16 to GS 64.Therefore,it implied that GS 64 had higher values compared to GS 16.The GS 32 provided much realistic estimation of the FD compared to the others and was used for further interpretation of the results.In so doing,the contour FD was generated for the GS 32.The FD variation for the negative-terrain(stream networks)was also determined.It was found that:(1)the higher the Strahler Numbers(SN),the higher the Fractal Dimension values obtained.(2)Also,the stream characteristic patterns in terms of the level of branching(dendritic,trellis,parallel)have an influence on the FD values obtained.Though,there was not in-depth studies on the river network patterns and their influence on the FD,it was revealed that,the denser the river pattern,the higher the FD obtained.(3)Larger catchments ten to have a much higher FD.These findings for the stream networks were found to be comparable with other studies.EPMs were generated using six indicators – Land Use Land Cover,Slope,FD,Flow Accumulation,Geomorphological Map and Elevation maps using two scenarios:(1)EPM with FD(Epm Fd)and(2)EPM without FD(Epm Nfd).The purpose of these scenarios was to validate the role of the FD in overall EPM process.The results showed that Epm FD provided more realistic spatial mapping of the erosion for the research area.The Epm Fd classified as Low Erosion Potential,Medium Erosion Potential and High Erosion Potential revealed close proximity to spatial behaviour according the geomorphological complexities and dynamics.This Epm FD was further compared with the contour FD,and show close similarities in terms of the characterising the level of complexity and dynamics.In summary,the five(5)scientific contributions of this study include the following:1.The CBPs are useful for the characterizing of geomorphic features for watersheds and even enriched in their descriptive nature when integrated with the FD for each of the CBP.This is because the fractal nature of each CBPs further describes the statistical complexity of the CBP generated from the watershed.2.GS affect the estimation of FD for DEMs – the larger the grid size window used,the larger the fractal values obtained.However,caution should be taken here,GS are site specific.Therefore,though demonstrated in this study that a grid size of 32 was tenable,it is very possible that in another landforms study,another grid size may be tenable.Therefore care must always to taken to consider at least three grid size windows before a decision made on the grid size to use for the study.3.FD plays a significant role in the determination of EPM.This was confirmed by the comparison between the contour FD and the Epm FD.4.Strahler numbers of stream networks were to a large extent related the fractal dimensions.Higher SN was an indication of higher FD.5.Box Counting and Higuchi methods proved to be appropriate for realistic FD estimates for the loess plateau.For the AHP,the Inverse Linear Scale used in the study prove to provide lowest CR and MER,which could be considered for future studies.The outcomes from this study could be blended other studies such as social dynamics and climate change studies to provide useful guide for national land use planning and further better decision making for the government and enterprises.The EPM results could prove vital for the understanding especially for the initiatives such as Group on Earth Observation(GEO)Land Degradation Neutrality(LDN)which is a global initiative to promote capacity building to avoid,reduce,and reverse land degradation with the aim of achieving LDN for the Sustainable Development Goals(SDGs)15.3 in every country by 2030.