Development of observation model for predicting the phenomena of Rill and Gully erosion using Machine learning

I am a doctoral candidate in University of Trento (first year) in Earth Observation at the Department of Civil, Environmental and Mechanical Engineering. My administrative University is Sapienza University of Rome and I am writing to express my interest in presenting my research in developing the strategies to predict Rill and Gully erosion in EGU general assembly 2024.

Soil moisture plays a major role in assisting crop productivity and weather forecasting to satisfy the ever-growing demands for land resources.

Soil erosion mechanics involve fluid (water/wind) detachment or entrainment followed by transport of soil particles and subsequent deposition as soil sediment. Soil movement by water often starts when a raindrop impacts the soil surface and initiates splash erosion, i.e., raindrops break aggregates into finer soil particles, displacing those particles and aggregates to create depressions in the soil surface. It depends on rainfall intensity, soil erodibility, and field slope among other factors. 

Various factors affecting soil erosion, including soil slope and length or supporting control practices like contour rows, strip cropping, and terrace systems, were recognized as independent factors influencing soil erosion by their inclusion in regional soil-loss equations. Water Erosion Prediction Project has been used to predict soil loss in a range of environments such as rangeland and forest for simulating runoff and sediment yield from the untreated watershed with good accuracy using continuity equation :

(dG/dx)  =  D_r + D_i 

G = sediment load (kg·s^-1· m^-1)

x = distance down slope (m)

D_r = rill erosion rate (+for detachment, - for deposition)

D_i = interrill sediment delivery (kg·s⁻¹·m⁻²).

Water Erosion Prediction Project relates sediment load in the runoff to the distance downslope as a function of the interrill and rill erosion rates calculated on a daily time step. Interrill erosion is the process of sediment delivery to more concentrated flow in rills, but rill erosion depends on the potential detachment capacity as limited by the sediment transport capacity of runoff in the rill. Soil loss through interrill and rill erosion is associated with the factor known as Revised Universal Soil Loss Equation (RUSLE), formulated as :

A  = R∗K∗L_s∗C∗P

A is the annual soil loss due to erosion [t/ha year];

R the rainfall erosivity factor;

K the soil erodibility factor;

L_S the topographic factor derived from slope length and slope gradient;

C the cover and management factor; and

P the erosion control practice factor.

The limitations of RUSLE are that it only accounts for soil loss through sheet and rill erosion and ignores the effects of gully erosion. 

The objective is to generate gully erosion susceptibility maps (GESMs) by applying three machine learning algorithms to identify the area of the basin with respect to total area, which prones to have higher or lower susceptiblity to gully erosion.

These erosions require Persistent Monitoring with the combination of three main elements. High resolution, Revisit rate and global coverage. The models can be developed using GIS or R software and from SAR technologies.

Considering my academic performance so far, I hope to have this opportunity to present in EGU assembly, as I am confident that I will be able to meet your expectations.

Thanking you.