dynatopmodel (version 1.2.1)

discretise: Discrete a catchment into hydrological response units (HRUs)

Description

Discrete a catchment into a set hydrological response units (HRUs) according to any number of landscape layers and cuts

Usage

discretise(layers, chans, cuts = list(a = 10), area.thresh = 2/100,
  order.by = names(cuts)[[1]], riv.cells.na = FALSE, renumber = FALSE,
  order = FALSE, burn.hrus = NULL, chan.width = 5, remove.areas = TRUE,
  hrus = NULL)

Arguments

layers

A multi-band raster (stack) comprising the catchment data. This should be in a projected coordinate system (or none) and have regular cells. The first layer should be the elevation raster, and subsequent (named) layers should supply the landscape data drawn in to create the discretisation

chans

Raster containing channel reach locations, of the same dimensions and resolution of the DEM and other catchment layers. The reaches should be numbered sequentially and any areas not containing part of the channel should be NA. If a second band is supplied with values 0-1 then this is taken to be the proportion of the corresponding non-zero cell occupied by the channel. If this layer is not present then the proportion is inferred from the channel width as p = min(1, chan.width/xres(dem))

cuts

A list of cuts of the form layer_name=number. Each name should correspond to a layer name in the layers parameter.

area.thresh

Minimum area for response units, expressed as a percentage of the catchment plan area, excluding channel cells. Areas smaller than this are aggregated with adjacent areas until exceeding the threshold area

order.by

Name of layer whose values will be use to sort the response units, in decreasing order. Defaults to the name of the first cut

riv.cells.na

Boolean Remove river cells from discretisation (default FALSE)

renumber

Boolean Renumber HRUs after discretisation so that their IDS are in numerical order (default TRUE)

order

Boolean Reorder HRUs after discretisation so that those with highest TWI come first (approximate to distances from channel). Default FALSE

burn.hrus

list Named list of geometries (supplied as rasters) to burn into discretisation of HRUs that will be stamped onto the classification. Overrides any classification already defined

chan.width

Channel width, in same units as DEM. Only used if chans doesn't contain a layer to specify the proportion of each river cell comprised of the channel.

remove.areas

Boolean Whether to remove areas that fall into classes with smaller areal contribution than the supplied threshold (default FALSE)

hrus

list Unused, maintained for backward compatibility

Value

A list comprising the following:

groups A data frame whose rows comprising the names, plan area and model parameters of each response unit. See Beven and Freer (2001) and Metcalfe et al. (2015) for a description of these parameters

weights Flux distribution (weighting) matrix for routing of subsurface flow downslope through the response units. If n is the number of response units (including channel "unit(s)") this is an n x n matrix. Row sums should thus always add to 1. The elements of the i-th row give the proportion of flow directed from response unit i to the other units

cuts list Cuts applied to produce these HRUs

area.thresh Area threshold specified

layers Multi-band raster comprising the the original rasters that the specified cuts were applied to produce the discretisation; the channel network

chans The channel raster

hru The resultant response unit locations

Details

This applies the given cuts to the supplied landscape layers to produce areal groupings of the catchment.

Examples

# NOT RUN {
# Landcover and soils are fairly homogenous throughout the Brompton catchment.
# Due to the extensive artifical sybsurface drainage discharging directly into 
# the channel it is hypothesied that the storm response is largely mostly controlled 
# by proximity to the network. A simple discretisation according to flow distance 
# from the nearest channel thus appears to capture the dynamics without introducing
# unnecessary complexity.
# }
# NOT RUN {
require(dynatopmodel)

data(brompton)

chans <- build_chans(brompton$dem, drn=brompton$drn, chan.width=2)
sort by distance from the channel network, but want areas closest the channel to come first
layers <- addLayer(brompton$dem, 2000-brompton$flowdists)
disc <- discretise(layers, cuts=c(flowdists=10), chans=chans, area.thresh=0.5/100)
rm(chans)
rm(layers)
write.table(disc$groups, sep="\t", row.names=FALSE)
# }