Análisis de errores en modelos medioambientales de variables discretas

Abstract

Some of the variables that try to represent the environment are difficult to measure, so they are usually estimated using models based on other, spreading the errors in the source data. This article is a bibliographic compilation about the most important aspects to consider analyzing the error propagation in models with spatial inputs and discrete outputs data on environmental variables. 

Because these models usually have input as discrete variables, their error sour- ces influence is also analized by error propagation. The most important error sour- ces of this type data input are proper completion of the classes, identification of the category —thematic and conceptual errors—, tracing its edges —cartographic preci- sion and accuracy— and scale, allow for variables in the error propagation analysis. 

Once we know the source of the error, then different studies of sensitivity analysis of some models that can serve as reference to the analysis of other discrete environmental variables. Monte Carlo is shown as a suitable method for the analys- is of error propagation for discrete variables, but we have not find literature that compare different methods for the same data set or model. Certain peculiarities of each data model —raster and vector— and its influence on the results are discussed.