Existing riparian classification and its impact on RVD values

Within RCAT, in order to calculate riparian vegetation departure (RVD), LANDFIRE layers for existing (EVT) and historic (BpS) vegetation are reclassified into a binary classification of Riparian or Non-riparian. In the BpS layer, there is generally a riparian specific classification along all but the smallest drainage networks. In contrast, in the EVT layer, the 30 m pixels are too coarse to pick up riparian vegetation that may exist along very small channels, and as a result the pixel gets classified as whatever vegetation type is dominant in the rest of the pixel. This causes RVD in small channels to consistently show up as 0 (i.e., there is no remaining riparian vegetation). This is simply a limitation of the vegetation resolution. In order to deal with this, and attempt to make the model useful even in small channels, we broaden our classification of riparian in the EVT raster to classes that are not explicitly riparian, but often occur along with (or alongside) riparian vegetation in small channels (e.g., deciduous forest species). In some cases, this causes the model to show more riparian vegetation now compared with the historic riparian layer (negative RVD values). 

Negative RVD values, therefore, are generally interpreted as areas where existing vegetation communities along channels are relatively in tact and we infer that riparian vegetation departure is minimal. When averaging values across an entire basin, large negative values can skew the average, "washing out" meaningful values along larger, higher-order channels.