Dissaggregation modelling of spring discharges

Disaggregation models are basically divided into three main groups: temporal, spatial and temporal-spatial. The focus of this paper is the application of temporal disaggregation models to disaggregate the seasonal flow in some large time intervals to sub-seasonal flows in some shorter time intervals. Two basic models are applied: the original model of Mejia and Rousselle and the corrected extended Lin model one-stage disaggregation. The flow totals from some karstic springs are used. Data for five springs in different areas of Bulgaria for the aims of the study are executed. The synthetic data generation for the chosen spring stations for a new realisation of thirty years is obtained. The multi-variate lag-one auto regressive model (AR(1) model) is applied for generation of the annual flow sequences. The Lin model single- site is performed for thirty years generation period. The Lin model is an improvement compared to the original extended model. The new Lin approach succeeds in the preservation of the additivity as well as the moments. Applying the Lin model one-stage disaggregation results in consistent model parameter estimates. As a second step in the research multi-site disaggregation schemes are also applied