Relationship between rainfall and runoff variability in the La Plata Basin at seasonal to decadal timescale

INÉS CAMILLONI; CARLA GULIZIA

Canmore

Congreso; 8th GEWEX Open Science Conference: Extremes and water on the edge; 2018

International GEWEX Project Office (IGPO)

The main objective of the present study is to comprehend the interaction between precipitation and runoff variability in the La Plata basin (LPB) along a wide range of timescales. The runoff hydrological response amplification is also addressed in order to evaluate to what extent runoff variability can be explained by precipitation fluctuations. For this purpose, six sub-basins corresponding to the main rivers of LPB (Paraguay, Parana, Uruguay and Iguazu) are selected within 1931-2010 period. Monthly runoff data was gathered from hydrological stations at six closing points and areal-averaged precipitation was derived from the Global Precipitation Climatology Centre gridded dataset. Precipitation and runoff seasonal and annual trends are explored. Observed runoff exhibits, in almost all cases (seasons and sub-basins), significant positive trends with 95% confidence, not fully explained by an increase in rainfall, but most probably associated to the land use changes produced in the region. Moreover, comparing sub-periods 1931-1970 and 1971-2010, precipitation exhibit an increase in the latter period (4% to 6% in all sub-basins, except 11% in Uruguay River basin), while runoff positive changes are amplified, ranging between 27% and 37%. Given the non-linear relationship between precipitation and runoff, a non-parametric runoff elasticity index is calculated over each of the six sub-basins considering separately the period 1931-2010 and the two 40-yrs sub-periods, respectively. The elasticity index associates the percent change in runoff given a unit percent change in climate (i.e. represented by precipitation). For example an elasticity of streamflow of about 2.0 indicates that 10% change in rainfall results in a 20% change in mean annual streamflow. The more recent period, after 1970, exhibits smaller elasticity indices than the previous one for almost all river basins. These results could be partially explained by the possible influence of land-use change activities in LPB streamflows. Two additional parameters (coefficient of variation (CV) and amplification) at different timescales (annual, decadal and centennial) are analyzed. The CV is calculated as the ratio between the standard deviation and the long-term mean for each hydrological variable. The ratio between runoff and precipitation CVs is considered as an indicator of the streamflow?s variability amplification relative to that of precipitation. Runoff CVs are considerably higher than precipitation CVs for all basins, which consequently exhibits runoff amplification. Furthermore, traditional Budyko Framework equations are also applied to infer observed runoff variability amplification, with consistent results particularly at the annual and decadal timescales.In addition, spectral analysis based on the maximum entropy method (MEM) was applied to study extensively the complex relationship between observed runoff and the sub-basin precipitation areal averages. The analysis was particularly focused on the amplification of low frequency variability in runoff compared to precipitation. Overall, river discharges show higher spectral density over decadal/interdecadal frequencies compared to precipitation analysis. It is assumed that the watershed acts on precipitation as a spatio-temporal integration operating as a low-pass filter, similar to a moving average. Runoff power spectrum analysis is simulated assuming that the underlying process is an autoregressive moving average (ARMA). Results confirm that the ARMA filter fits effectively the observations. In order to better understand the non-linear rainfall-runoff physical relationship, runoff spectrum could be derived directly and accurately from the precipitation spectrum, transformed by a ?basin? operator, characteristic of the basin itself.