Airborne and satellite observations show that when large rivers join
they can take hundreds of kilometers to mix completely but, on
occasion, may mix very rapidly. Application of established
semitheoretical analyses shows that long mixing lengths should be
expected. The first measurements of mixing processes at a large river
junction (Rio Parana and Rio Paraguay, Argentina, combined width
similar to 2.8 km) are presented at two occasions: first when they mix
in > 400 km, and second when mixing is complete in only 8 km downstream
of the junction. For the case of slower mixing, at-a-point surveys
showed that mixing driven by turbulent shear associated with a
near-vertical shear layer was restricted to close to the junction (to
0.272 multiples of the postconfluence width downstream). Transect
surveys showed penetration of more turbid water from the Rio Paraguay
underneath the Rio Parana, but this was insufficient to promote more
rapid mixing. There was no clear channel-scale circulation present and
slow mixing was compounded by reverse topographic forcing on the
mainstream Rio Parana side of the river. This kept more turbid water on
the Rio Paraguay side of the river, close to the bed. In the case of
rapid mixing, we found clear channel-scale circulation. The momentum
ratio between the combining flows reinforced the effects of the
discordance in bed height between the tributaries at the confluence and
allowed penetration of more turbid Rio Paraguay water further across
the channel width deeper within the flow. The importance of the
interaction between momentum ratio and bed morphology at channel
junctions makes mixing rates at the confluence dependent upon
basin-scale hydrological response, which is more likely to differ
between large confluent rivers than small rivers, as a result of the
different climatic/topographic zones that they may capture.