Improved process descriptions in the PESERA model

Table of Contents

1. Authors
2. Abstract
3. Introduction
4. General extensions to PESERA
   1. Grazing
   2. Fire
   3. Wind Erosion
5. Particular extensions to PESERA
   1. Mulching and/or maintaining ground cover vegetation within tree crops
   2. Retention of crop residues as litter layer at harvesting of arable and other crops
   3. Irrigation and water harvesting for croplands
   4. Invasion and clearance of unpalatable species
   5. Terracing and strip cropping
   6. Nitrogen budgeting and rotations
6. Other extensions to PESERA
   1. Mass movements
   2. Data collection calibration of PESERA against reservoir data
7. Conclusions
8. References

Mass movements

A detailed model for mass movements is being developed within the PESERA model by CNR-IRPI in WB 5.2. Because of the finer scale of this application, it has been necessary to simplify the conceptual principles for partial inclusion within the coarse scale PESERA model. Mass movements are driven by rupture within the soil or rock along a defined surface within the soil. Here we represent only shallow slides with a shallow slide surface sub-parallel to the ground surface. Slides occur when the downslope component of the weight of overlying soil overcomes the resistance to movement, or the 'shear strength' of the material. This resistance is made up of two components; friction and cohesion. The cohesive strength is a constant force per unit area, highest in intact consolidated clays but generally very small for materials weathered near the surface, and here ignored. The frictional strength is proportional to the normal stress (pressure or force per unit area) across the slide surface, with a constant of proportionality that is the 'coefficient of friction', commonly expressed as the tangent of the 'angle of friction'. In a pile of sand or gravel, this angle is equal to the maximum stable angle for the pile. Archimedes' principle states that the upthrust is equal to the weight of water displaced: applying this to a sloping soil mass, the normal stress due to the weight of overlying soil is reduced as the soil becomes saturated. Since soil has a density of approximately twice that of water, a fully saturated soil applies only about half of the normal stress compared to a dry soil, so that the frictional strength is proportionally reduced, and failures in a wettable soil generally occur when it is saturated, and at a gradient that is about half of the angle of friction.

Applying this simplified model as an extension of PESERA, we need to forecast the spatial frequency of susceptible slopes (at or close to half the angle of friction) and the temporal frequency of saturated conditions. Some approximation is needed for both at a coarse scale, so that we recognise the need for calibration and evaluation of the results, preferably against both the CNR-IRPI fine scale model and against observed events.

Gradient can be estimated from DEM data, but needs to be analysed at the finest scale available. Previous work has suggested that a resolution of 10m or better is desirable, whereas PESERA currently relies mainly on the 90m SRTM data set. Gradient measured at these coarser scales generally underestimates the steepest slopes, so that the critical slope value should be made to respond to the DEM resolution, calibrating against progressive degradation of a fine scale data set. For each grid cell, the DEM will provide a frequency distribution of gradients (120 values per km² for the SRTM; more with a finer resolution DEM). This distribution, f(g) is then compared with a stable angle, , (half the angle of friction) estimated from the soil texture, interpreted from soil and/or geological maps.

Average deficit can also be simulated from the modelled monthly deficit, multiplied by the frequency of rainfall events in the month, each offsetting the average deficit. We obtain a gradient-dependent expression for the probability p(D,g) of a given deficit, D. Combining these two distributions, we try to fit an expression of the form:

This expression convolutes the two distributions to provide a probability of landslide occurrence in any grid cell for each month. As with wind erosion this is used primarily to estimate the frequency of crop destruction rather than a sediment transport volume. However, by combining this frequency with an average slide volume, this can be used to estimate volumes removed, which are required to reconcile PESERA model estimates with reservoir data.

The input and output of the fine scale model has been prepared in a way that is compatible with PESERA, and can be treated as an addendum to the basic PESERA model, although final coupling between the models has not yet been completed. The prototype fine scale model has been tested against data for the Rendina catchment and the final version of the model should be ready in May 2010.

Priority has been given to the modelling of mass movements. Other aspects of fine scale modelling will benefit, particularly from the frequency distributions of gradients developed for the landslide model, as a basis for representing details of topography at sub-grid scales. This methodology is being particularly applied to modelling of tillage erosion and land levelling.