This paper develops a sequential entropy filter for disaggregating nonpoint sources from ambient data. A numerical simulation based on sediment loading is provided to illustrate the ability of the sequential entropy filter to recover the underlying parameters and optimally disaggregate ambient sediment load among nonpoint sources. In the process we show the equivalence of this sequential entropy filter with Bayes' theorem and, given this equivalence, argue that the sequential entropy filter is more applicable than traditional Bayesian estimators are when the parameter distributions are unknown or when the sample is undersized, which is typically the case when dealing with natural resource data.