『Abstract
　Previous research in agricultural catchments showed that past inputs of nitrate continue to influence present observations and future characteristics of nitrate concentrations in stream water for a long period of time. This persistence manifests itself as a “memory effect” with a prolonged response of stream water nitrate levels to reductions of nitrate inputs on the catchment scale. The question we attempt to resolve is whether such a memory effect also exists in mountainous catchments with a snowmelt-dominated runoff regime. We analyzed long-term records (〜20 years) of nitrate-nitrogen concentrations measured in stream at three stations on the upper Vah（aの頭に´） River (Slovakia). Applying spectral analysis and detrended fluctuation analysis, we found a varying degree of persistence between the three analyzed sites. With increasing catchment area, the fluctuation scaling exponents generally increased from 0.77 to 0.93 (fluctuation exponents above 0.5 are usually considered as a proof of persistence while values close to 0.5 indicate “white” uncorrelated noise). The nitrate-nitrogen signals temporally scaled as a power-low function of frequently (1/f noise) with a strong annual seasonality. This increase in persistence might be attributable to the catchment areas upstream the sampling sites. These results have important implications for water quality management. In areas where reduction of nitrate in surface waters is imposed by legislation and regulatory measures, two catchments with different persistence properties may not respond to the same reduction of sources of nitrogen at the same rate.

Keywords: Persistence; Memory effect; Nitrate concentrations; Detrended fluctuation analysis; Spectral analysis』

1. Introduction
2. Materials and methods
　2.1. Study area and databases
　2.2. Analyses
3. Results
4. Discussion
5. Conclusions
Acknowledgments
References