Daniel Cluis

Analysis of the El Niño effect on the discharge of selected rivers in the Asia-Pacific Region

Abstract Discharge records originating from 78 rivers within the Asia-Pacific region are used to assess a possible relationship between a temporal El Niño-Southern Oscillation (ENSO) Index and regional discharges of rivers. The rivers selected have long monthly discharge series and are geographically distributed throughout the whole region. Analyses of variance are used to identify the sub-regions most teleconnected to the ENSO signal. Composite analyses show the temporal patterns of the relationships between river discharge and the ENSO signal and the possible presence of a time lagged relationship. The analyses show that the rivers of the Oceania-Pacific region are the most affected and that a majority of rivers show a similar temporal pattern, thus allowing the construction of a regional composite streamflow index. The temporal patterns of the identified relationships demonstrate the possibility of forecasting the occurrence of abnormally high or low flows (floods/droughts) months ahead of time.

Potential eutrophication assessment in rivers; relationship between produced and exported loads

In contrast with biological processes in lacustrine environments, the physical characteristics of river waters greatly complicate the relationship between nutrient input and primary productivity.In this paper a new approach to quantify this relationship is developed, linking land use within the watershed to the bioavailability of nutrients in the receiving waters. It estimates by two different ways an enrichment index derived from N/P ratios.A load model links land-use within the watershed area to nutrient export, and allows the estimation of the potential enrichment index from the ratio of calculated annual exports of nitrogen and phosphorus. This index value correlates well with the real enrichment index representing the annual mean value of the corresponding measured ratios in running waters. The latter index can then be related to nutrient availability expressed by chl-a and the algal growth potential.This model, established from 7 sampling stations on the Yamaska-North river, allowed assessment of nutrient bioavailability in the river water as a function of hydrological (low or high water flow) as well as of seasonal (summer or winter) events.

Global Variance and Root Mean Square Error Associated with Linear Interpolation of a Markovian Time-Series

Abstract Most general-purpose data acquisition networks provide equispaced instantaneous information; the frequency of measurements necessary to obtain this information efficiently is related to the intrinsic temporal variability of the given phenomenon. Thus, meteorological and hydrometric phenomena are sampled more intensively than are more stable groundwater variates. Once the data have been acquired, the estimation of values which should have been taken by a time-series at higher frequency or of the combination of two or more time-series measured at different frequencies is a frequent but unsolved problem. in the field of water quality monitoring, for example, the estimation of mass-discharges is a prerequisite for the interpretation of transport phenomena, source-effects relationships and trend detection. To evaluate this important secondary variate, one must combine high frequency / high variability flow data with low frequency / low variability concentration data; this can be done by using some combination of aggregation and interpolation of data. the aggregation of high frequency data has relatively minor effects, e. g. a reduction of the variance and a modification of the persistence structure in the transformed data. However, the spreading of the information resulting from linear interpolation creates a certain level of heteroscedasticity and also produces an error of estimation, the variance of which increases with the number of partitions. For phenomena exhibiting short-term positive Markovian persistence, the analytical expression for the global variance of the estimation error was first established for skipped series derived from actual measurements: using the self-similar persistence structure of the Markovian processes, the Root Mean Square Error (RMSE) of the interpolated time-series was then deduced. Thus, a criterion relating the short-term persistence to the number of partitions allows to control and limit the level of error in the transformed time-series.