RPhosFate v2.0.1

This version introduces a new flow method, but also comes with several breaking changes.

Overview

• Use of the D-infinity (Tarboton, 1997) instead of the D8 flow method, however, the D8 flow method can still be mimicked by rounding D-infinity flow directions to the nearest multiple of 45 degrees.
• Weighted flow accumulations are no longer supported for the time being.
• The basis for the calculation of the D-infinity slopes is the original digital elevation model and not the breached one with reversed stream burning as it used to be for the calculation of the D8 slopes.
• The default value for the parameter holding the minimum bounding slope (ns_slp_min) is now 1 instead of 0.001%.
• The L factor is now calculated using equation (9) as published by Desmet and Govers (1996) and not as modified by Kovacs (2013).
• The channel retentions of the substance outlet loads of subsurface drainages are now calculated analogous to the overland retentions of the local emissions, i.e. by using half of the respective flow path length.
• RPhosFate now utilises the terra instead of the raster package.
• Ceased the support for ERDAS IMAGINE (*.img) raster files and added img2tif() function can be used to convert all ERDAS IMAGINE raster files in a directory and its subdirectories into GeoTIFF raster files.
• Dropped backward compatibility to major version zero.

Technical details

• Input data changes:
  • Removed the layers holding the (weighted) D8 flow accumulations (acc and acc_wtd) and added acc_inf holding the D-infinity flow accumulations.
  • Removed the layer holding the D8 flow directions (dir) and added dir_inf holding the D-infinity flow directions.
  • Removed the layer holding the D8 slopes (slp) and added slp_inf holding the D-infinity slopes.
• DEMrelatedInput() function changes:
  • Considerable adjustments to reflect the aforementioned input data changes.
  • Addition of ns_cha argument: allows for specifying the minimum D8 flow accumulation determining a channel.
  • Addition of ls_mD8 argument: allows for mimicking D8 flow directions by rounding the D-infinity flow directions to the nearest multiple of 45 degrees. Please note that this treatment is always applied to channel cells independently of this argument.
  • Removal of the cs_wgs and cs_dir arguments: These input data are no longer supported for the time being.
• Switched to utilising the SpatRaster and SpatVector classes from the terra package instead of the RasterLayer class from the raster and the Spatial*DataFrame classes from the sp packages.
• Added is_ths argument to the RPhosFate() and catchment() constructors: allows for specifying the number of threads to use for processing, where applicable.
• Removed the layer holding the hydraulic radii (rhy), as the calculation of the hydraulic radii is now integrated into the transport() method. This implies that the transportPrerequisites() method does not save it to disk any longer.
• Removed the parameter holding the D8 outflow direction vector (iv_fDo). Existing parameter files containing it can still be used, but the parameter will definitely be lost upon saving the project’s state.
• Removed the transportCalcOrder() method, as the determination of the cell transport calculation order is now integrated into the transport() method. This implies that the firstRun() as well as subsequentRun() methods do not call it and the saveState() method does not save it to disk any longer.
• The calibrationQuality() method now returns its return value invisibly.
• Considerably revised the internal RPhosFateHelpers class.
• Removed spatstat.geom from the imported packages list (utilised functionality is now also provided by terra).
• Bumped the minimum tested R version from 4.2.3 to 4.3.2 using the corresponding Posit public package manager snapshot.
• Slightly improved the documentation.
• Major internal code improvements.

RPhosFate v1.0.4

• Removed hydroGOF from the imported packages list (maptools is retiring and hydroGOF depends on it via hydroTSM). Thanks to Roger Bivand for raising this issue (#17).
• Bumped the minimum tested R version from 4.2.2 to 4.2.3 using the corresponding Posit public package manager snapshot.
• Slightly improved the documentation.

RPhosFate v1.0.3

• Removed rgdal from the suggested packages list and set the minimum required version of the raster package to ≥ 3.6.3 (rgdal is retiring and raster ≥ 3.6.3 does not depend on it any longer). Thanks to Roger Bivand for raising this issue (#17).
• Bumped the minimum tested R version from 4.1.2 to 4.2.2 using the corresponding MRAN repository snapshot.
• Slightly improved the documentation.
• Minor internal code improvements.

RPhosFate v1.0.2

• Removed NRMSE from calibration quality metrics and added KGE as well as RCV.

RPhosFate v0.12.0

• Added autoCalibrate2() method: allows for calibrating the overland and channel deposition rate in one go.
• The DEMrelatedInput() function can handle multiple catchment outlets now and so does the calibrationQuality() method.
• The DEMrelatedInput() function now returns the breached DEM with reversed stream burning if applicable instead of the original one.
• The DEMrelatedInput() function now calculates “correct” slopes even if the channels used for stream burning contain gaps.
• RPhosFate now makes sure that the x- and y-coordinates of gauges used for calibration lie within the extent of the river catchment object.
• Renamed the "inChannelRetention" output of the calibrationQuality() method to "inChannelRetentionRatio".
• RPhosFate requires R ≥ 3.5.0 now.
• Improved the documentation.

RPhosFate v0.11.0

• Monte Carlo simulation mode can now make use of repeated random samples, i.e. raster data, of distributions of about all input data.
• Added the following arguments to the subsequentRun() method, which allow for calling the respective methods:
  • erosionPrerequisites
  • erosion
  • emission
  • transportPrerequisites
  • transportCalcOrder
• Added cv_MCl argument to the RPhosFate() and catchment() constructors: allows for specifying the names of the layers, which shall be written to disk with the associated Monte Carlo iteration in their filenames upon calling the appropriate methods.
• Initialising a project in Monte Carlo simulation mode now also reads model results produced by a possible earlier run associated with the specified iteration. This implies that Monte Carlo input data additionally can reside in the project root subdirectories and not only in a separate directory.
• The plot produced by the calibrationQuality() method is now prettier.
• Added a vignette describing the basic modelling workflow.
• RPhosFate now depends on the spatstat.geom instead of the spatstat package.
• Fixed the minimum required version of the whitebox package (≥ 2.0.0).
• Added test for the standard use case of the DEMrelatedInput() function.
• Slightly improved the documentation.

RPhosFate v0.10.0

• The DEMrelatedInput() function now calculates the slopes from the breached DEM (stream burning is undone beforehand).
• Switched to utilising GeoTIFF (*.tif) instead of ERDAS IMAGINE (*.img) raster files.
• Added cs_fex argument to the DEMrelatedInput() function: allows for using ERDAS IMAGINE raster files for backward compatibility.
• Added cs_dir argument to the DEMrelatedInput() function: allows for utilising an existing D8 flow directions raster using ArcGIS codes.
• Added demoProject() function providing training data.
• Added examples to the documentation.
• Fixed a backward incompatibility.
• Added tests utilising the unit testing framework of the tinytest package.
• Added means to measure code coverage with the help of covr.