Po River discharges: a preliminary analysis of a 200-year time series

Abstract

Daily stage data measured at the closing section of the Po River (Northern Italy), collected from historical archives of the “Hydrological Office of the Po River—Parma”, allowed to estimate daily discharges for the period 1807–1916, therefore to extend the time series of Po River discharges for 110 years before the actually published time series. This paper provides a detailed description of the stage–discharge conversion and of the tests performed for checking homogeneity of reconstructed data. In particular, monthly discharge data were compared with approximated catchment-average data of precipitation and evapotranspiration for the period 1831–2003, which were estimated, respectively, from monthly data of local precipitation (at Milan, Turin and Parma) and local air temperature (at Milan and Turin). It emerges that estimated values of precipitation, evapotranspiration and discharge provide a coherent picture of the hydrological dynamics in the basin throughout the study period. Specifically, an apparent progressive depletion of basin reservoirs is observed since 1920, i.e., when concomitant sudden changes (regime shifts) occurred in precipitation (downward shift) and evapotranspiration (upward shift). The 1920-shift is amongst the likely causes of the fact that prolonged drought periods as those observed in the 1940s and since 2003 are not observed in the pre-shift period, when accumulation of reservoirs occurred. The increase in peak-flow discharges observed in recent decades, with values well above the maximum discharge estimated for the nineteenth century as well as for the early twentieth century, is apparently the result of the massive levee works along the river network that were completed in the 1960s. On decadal time scales, discharge variability is found to essentially reflect the changes in precipitation patterns. In particular, peaks of comparable magnitude are found in the 128-month (∼11 years) wavelet spectra of precipitation and discharge. Furthermore, concurrent changes are observed in the persistence (i.e., autocorrelation) of precipitation and discharge data. Since the red-noise background spectrum of discharge is much lower than that of precipitation, river discharge is the likely hydrological variable to be preferred for assessing the basin’s response to the background climatic variability occurring at the decadal and multi-decadal time scales, notwithstanding the fact that changes in water management and other anthropogenic impacts can be important on long time scales. Concerning the seasonal to interannual response to climatic forcing, a robust dependence of wintertime precipitation and discharges on the state of the NAO was observed. This dependence results in stronger (weaker) precipitation and higher (lower) discharges during negative (positive) anomalies of the NAO index.