Tuesday, 15 January 2002
Spatial-temporal structure of the interannual variability of precipitation over "Rio de la Plata" basin
A combination of two statistical analyses, spatial principal component analysis (PCA) and multichannel singular spectrum analysis (MSSA), were applied to characterize the spatial-temporal structure of trend and quasi-oscillatory modes of interannual variability of precipitation over Rio de la Plata basin. The data used in the study are from a work-grid (1 deg. lat-long resolution), which contains monthly precipitation records along the period 1901-1993. The series of monthly precipitation were converted in monthly anomalies (removing the annual cycle) and a low pass digital was applied to remove higher frequencies associated with faster than annual variations. First of all, PCA was performed on the data set of filtered monthly precipitation, and the first 10 leading S-PCs, which together account for about 79.23% of the total variance, were retained. MSSA analysis, with a window length of 96 months, was applied to the retained S-PCs. Trend components of the leading S-PCs were found in the monthly precipitation series for the last half of the century, but the space-time contributions show large areas with significant increases and decreases of precipitation after 1950. Extreme increases of precipitation were found in the upper basin of the Paraguay river, a region with big swamps denominated Pantanal. On the other hand, the sources of Pilcomayo and Bermejo rivers show negative trends. In the interannual timescales, signals with 6.2, 3.6 and 2.4 yr. were found in the space-time principal components of precipitation. The oscillatory pair with 6.2 years. being stronger over the Iguazú river basin whereas those corresponding to 3.6 and quasi-biennial oscillations present regional differences in amplitude and timing.