A non-parametric analysis of qualitative and quantitative data erosion modelling: a case study for Ethiopia

*Centre for World Food Studies of the Vrije Universiteit (SOW-VU). De Boelelaan 1105 1081 HV, Amsterdam, The Netherlands. *Corresponding author: b.g.j.s.sonneveld@sow.econ.vu.nl ABSTRACT The objectives of this paper are twofold. First, it compares the discriminatory power of qualitative expert judgements with actual soil losses to express class boundaries in physically measured, quantitative terms. Secondly, it investigates the properties of a postulated functional relationship between soil loss and readily available explanatory variables on both, their reliability of fit and behaviour. The study uses quantitative soil erosion data of runoff plots of the Soil Conservation Research Project in Ethiopia. Qualitative expert judgements on the state of erosion for the same runoff plots were obtained through a questionnaire. The study applies a non-parametric technique that uses a flexible method of curve fitting. The first exercise applies this technique to determine the quantitative boundaries (soil losses) of qualitative classes. The results reveal a positive relationship between erosion hazard assessment by the expert and actual soil losses, however, experts tend to overestimate. In the second exercise, the mollifier program is used to visualize non-parametric estimates in 3-D graphs that show non-linear relationships and reliability of the estimates. The results indicate that soil loss should be modelled separately for annual crops and land use types with a permanent coverage. Further findings show that annual runoff has an almost linear relation with annual soil loss. An index derived from monthly rainfall data and the adjusted Cooks’ method seems promising to represent the hydrological factor in the model. Most relations show a poor ‘goodness of fit’, which anticipates low correlation coefficients in future parametric, models and indicates that additional variables should be included.

[1]  B. Silverman.  Density estimation for statistics and data analysis , 1986 .

[2]  Rattan Lal,et al.  Soil Erosion in the Tropics: Principles and Management , 1990 .

[3]  Michiel A. Keyzer,et al.  Using the mollifier method to characterise data sets and models. The case of the universal soil loss equation , 1997 .

[4]  G. Jacks,et al.  Soil Conservation , 1940, Nature.

[5]  Rattan Lal,et al.  Sustainable Management of Soil Resources in the Humid Tropics , 1995 .

[6]  P. J. Albersen,et al.  Water erosion assessment based on expert knowledge and limited information using an ordered logit model , 1999 .

[7]  Donald Gabriëls,et al.  Assessment of Erosion. , 1980 .

[8]  Bofu Yu,et al.  Plot-scale Runoff Modelling for Soil Loss Predictions , 1997 .

[9]  G. R. Foster,et al.  Predicting soil erosion by water : a guide to conservation planning with the Revised Universal Soil Loss Equation (RUSLE) , 1997 .

[10]  M. A. Keyzer.  Estimation of real-valued models from discrete and limited dependent observation: a programming approach , 1996 .

[11]  Rome.  Guidelines for soil profile description. , 1977 .

[12]  Mark A. Nearing,et al.  A Single, Continuous Function for Slope Steepness Influence on Soil Loss , 1997 .

[13]  G. R. Foster,et al.  A Process-Based Soil Erosion Model for USDA-Water Erosion Prediction Project Technology , 1989 .

[14]  C. D. Kemp,et al.  Density Estimation for Statistics and Data Analysis , 1987 .

[15]  D F Yule,et al.  Soil management and rroduction of alfisols in the semi-arid tropics. I. Modelling the effects of soil management on runoff and erosion , 1996 .

[16]  H. Arnoldus.  An approximation of the rainfall factor in the Universal Soil Loss Equation. , 1980 .

[17]  Ckk Gachene,et al.  Evaluation and mapping of soil erosion susceptibility: an example from Kenya , 1995 .

[18]  Soil management and production of alfisols in the semi-arid tropics. IV. Simulation of decline in productivity caused by soil erosion , 1996 .

[19]  John N. Quinton.  Reducing predictive uncertainty in model simulations: a comparison of two methods using the European Soil Erosion Model (EUROSEM) , 1997 .