Submitted to: Soil Science Society of America Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: February 2, 2000
Publication Date: N/A
Interpretive Summary: Conservationists and scientists use soil erosion models for many purposes. There is a soil conservationist in nearly every county of the United States who is trained to use a soil erosion model, and to help farmers decide what practices to use on their land. Engineers use soil erosion models when they design reservoirs from the size of farm ponds to the size of electrical generating reservoirs such as the Three Gorges project in China. Municipal, state, and national governments use soil erosion models to set regulations for erosion and sediment control. The EPA uses them to predict pesticide movement to lakes, rivers, and streams. Soil erosion models play a key role in protecting our environment. One important question for scientists is: How do we know when an erosion model is working adequately? Nature is highly variable. Scientists find that erosion rates on two areas with soil types, slopes, rainfall, land management, and other factors which are not measurably different can have very different erosion rates. Thus when we ask the question about how well the model works, the answer is not so simple. One cannot just compare the model output to an erosion rate. One must simultaneously ask the question: How variable is nature? This study asks that question, and provides us the insight we need to evaluate just how well a model is working when we compare model results to measured soil erosion information.
Technical Abstract: One of the important methods used to evaluate the effectiveness of soil erosion models is to compare the predictions given by the model to measured data from soil loss collected on plots taken under natural rainfall conditions. While it is recognized that plot data contains natural variability, this factor is not quantitatively considered during such evaluations because our knowledge of natural variability between plots which have the same treatments is very limited. The goal of this study was to analyze sufficient replicated plot data and present methodology to allow the model evaluator to take natural, within- treatment variability of erosion plots into account when models are tested. A large number of data from pairs of replicated erosion plots were evaluated and quantified. The basis for the evaluation method presented is that if the difference between the model prediction and a measured plot data value lies within the population of differences between pairs of measured values, then the prediction is considered acceptable. A model effectiveness coefficient was defined for studies undertaken on large numbers of prediction vs. measured data comparisons. This method provides a quantitative criteria for taking into account natural variability and uncertainty in measured erosion plot data when that data is used to evaluate erosion models.