Temporal dynamics of monthly evaporation in Lake Urmia

Abstract

As a UNESCO biosphere, Lake Urmia is a shallow hypersaline lake which is facing a rapid water surface degradation. Evaporation from the surface of the Lake, as a physical process which accelerates the Lake's degradation, was evaluated using chaos theory. Seven hydrometeorological stations scattered around the Lake were selected, and a 40-year time span between October 1974 and September 2014 was used at each station. Missing data in time series was removed and the whole time series was tested for consistency, randomness, and presence of trend. Since evaporation at each station was measured by means of class A evaporation pan, time series at each station was multiplied by a pan coefficient to incorporate the effect of saline water and free water surface environment simultaneously. Measurement errors arising from assumption of zero evaporation in winter were removed from the time series using locally weighted scatterplot smoothing method after which unification of time series into a single time series is achieved. Results of the data transformation and information loss were monitored by means of auto-correlation, partial-auto-correlation, mutual information, power spectrum, false nearest neighbor, and correlation dimension. A local prediction method is then used to capture the temporal dynamics of the evaporation with consideration of an appropriate time delay and embedding dimension. Finally, the representative model was projected on a 3-dimensional phase space to evaluate the temporal dynamics of the evaporation. Results indicate that the chaotic approach shows accurate predictions in advance.