Nitrogen retention in rivers: model development and application to watersheds in the northeastern U.S.A.

A regression model (RivR-N) was developed that predicts the proportion of N removed from streams and reservoirs as an inverse function of the water displacement time of the water body (ratio of water body depth to water time of travel). When appliedto 16 drainage networks in the eastern U.S.,the RivR-N model predicted that 37% to 76%of N input to these rivers is removed duringtransport through the river networks.Approximately half of that is removed in1st through 4th order streams whichaccount for 90% of the total stream length. The other half is removed in 5th orderand higher rivers which account for only about10% of the total stream length. Most Nremoved in these higher orders is predicted tooriginate from watershed loading to small andintermediate sized streams. The proportion ofN removed from all streams in the watersheds(37–76%) is considerably higher than theproportion of N input to an individual reachthat is removed in that reach (generally<20%) because of the cumulative effect ofcontinued nitrogen removal along the entireflow path in downstream reaches. Thisgenerally has not been recognized in previousstudies, but is critical to an evaluation ofthe total amount of N removed within a rivernetwork. At the river network scale,reservoirs were predicted to have a minimaleffect on N removal. A fairly modest decrease(<10 percentage points) in the N removed atthe river network scale was predicted when athird of the direct watershed loading was tothe two highest orders compared to a uniformloading.