Erosion Analysis with USLE and RUSLE2

• Activity Number: 471
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): Yang Li*+ and Zhengyuan Zhu
• Companies: Iowa State University and Iowa State University
• Keywords: USLE ; RUSLE2 ; erosion ; random forest ; MARS

Abstract

Universal Soil Loss Equation is a model that predicts the long term average annual rate of soil erosion caused by rainfall. It will be replaced by the more sophisticated RUSLE2 (Revised USLE 2) model in 2006. Data for computing both USLE soil loss estimates and RUSLE2 estimates are collected between 2002 and 2006 for validation. To estimate long term soil erosion trend and maintain consistency as an input to other models, it is necessary to impute USLE estimates after 2006 and RUSLE2 estimates before 2002. We first present statistical models to predict soil loss estimates from one soil loss model using the variables from the other one. With the aid of cross validation, we assess three different fitting methods: simple linear regression, multiple linear regression (MLR), and multivariate adaptive regression splines (MARS). Next we investigate imputation of USLE estimates shortly after 2006 and RUSLE2 estimates shortly before 2002 using nearest neighbor, classification and regression tree, and random forest. Reliable imputation is achieved by exploiting the relationship among the changes between neighboring years.