ABSTRACT
R.J. Chorley and B.A. Kennedy provide a useful classification of systems equilibrium behavior in terms of the perceived response of the system through time by observation of a state variable. In formal terms, neighborhood stability analysis amounts to determining whether the change in the magnitude of the perturbation about equilibrium is positive or negative. This is achieved by obtaining an approximate differential equation for the perturbation using a Taylor expansion, neglecting non-linear terms and then solving to find behavior of perturbation through time. The data series used in this investigation include 15 minute observations obtained by probes from the river Thame and weekly data from the River Stour. In all cases except pH, the series for both rivers can be fitted with models which achieve over 90% ‘explanation’ of the original data series. For pH in both cases the residual variance (white noise under usual tests) constitutes about 60% of the total variance, the rest being accounted for by autoregressive-moving average models.
