A hydrochemically guided landscape-based classification for water
quality: a case study application of process-attribute mapping (PoAM) at
a national scale
Abstract
Spatial variation in landscape attributes can account for much of the
variability in water quality compared to land use factors. Spatial
variability arises from gradients in topographic, edaphic, and geologic
landscape attributes that govern the four dominant processes
(atmospheric, hydrological, microbially mediated redox, physical and
chemical weathering) that generate, store, attenuate, and transport
contaminants. This manuscript extends the application of Process
Attribute Mapping (PoAM), a hydrochemically guided landscape
classification system for modelling spatial variation in multiple water
quality indices, using New Zealand (268,021 km²) as an example. Twelve
geospatial datasets and >10,000 ground and surface water
samples from 2,921 monitoring sites guided the development of 16
process-attribute gradients (PAG) within a geographic information
system. Hydrochemical tracers were used to test the ability of PAG to
replicate each dominant process (cross validated R2 of
0.96 to 0.54). For water quality, land use intensity was incorporated
and the performance of PAG was evaluated using an independent dataset of
811 long-term surface water quality monitoring sites
(R2 values for total nitrogen of 0.90 - 0.71 (median =
0.78), nitrate-nitrite nitrogen 0.83 - 0.71 (0.79), total phosphorus
0.85 - 0.63 (0.73), dissolved reactive phosphorus 0.76 - 0.57 (0.73),
turbidity 0.92 - 0.48 (0.69), clarity 0.89 - 0.50 (0.62) and E.
coli 0.75 - 0.59 (0.74)). The PAGs retain significant regional
variation, with relative sensitivities related to variable geological
and climatic histories. Numerical models or policies that do not
consider landscape variation likely produce outputs or rule frameworks
that may not support improved water quality.