Appendix B — Version changes

The PostgreSQL version of the database on the Waterway Ecosystem Research Group ‘water’ server always holds the most up-to-date version of the mwstr database. Most recent changes are documented at the end of this appendix. If the final listed version is pending, the changes have not yet been committed to the downloadable gpkg version of the database. Past versions are available here.

B.1 Version 1.2

The changes made converting version 1.1 to 1.2 are described throughout the main part of this document. The following describes changes from version 1.0 to version 1.1 in January 2020.

Many reaches of the stream layer in version 1.0 had many small non-contiguous segments allocated to incorrect reaches. These have all been corrected now so that all confluences meet at the terminal nodes of the relevant segments.

The subcatchment layer was recalculated using the corrected stream layer, and now every subcatchment contains a matching stream reach (in v1.0 7,546 subcatchments had no matching stream reach). This improvement removed the need for a linking table (mwsubc_streams in v1.0) between the streams and subcs tables. All information that was in the mwsubc_streams table in v1.0 is in the subcs table of v1.1).

Selecting and calculating related reaches upstream and downstream in the network is handled more efficiently in v1.1 using array fields in the postgreSQL database for the allus (all upstream) and allds (all downstream) fields. This increased efficiency, and the derivation of the cats table removed the need for the ‘florets’ table, which was a feature of v1.0, no longer included in v1.1.

~1,100 km of additional stream length have been added with additional stream extensions

The reach code conventions have been revised from version 1.0. Instead of each reach being identified by a set four-digit code equal to 1000 times log10(catchment area in 100-m2), a variable-length integer of the catchment area in ha is now used. All subcatchments differ from their neighbouring subcatchments on the same stream line by at least 1 ha. 16 subcatchments downstream of small tributaries differed by <1 ha. For those cases, catchment areas were rounded up or down to ensure no duplicate reach codes.

Numerous minor changes made to reachcode field of small segments of mwstr to correct mismatches with mwsubc, in most cases where the subcatchment boundaries (angular polygons based on 5-m gridcell DEM) crossed the finer-resolution stream lines. This class of errors affects a small percentage of stream lines, and have been corrected opportunistically as we use the network. Certainly other errors remain, but should not cause significant errors in stream length estimation. Please report any such remaining errors to Chris Walsh for correction in future versions.

The strcode for WLB (trib of QC7) was recoded to Q71, and WLB was used for consistency with past practice for Wallaby Creek in the Goulburn catchment in the north of Kinglake State Park. Stream lines and new subcatchments were added for Wallaby Creek (as this stream is managed by Melbourne Water). Therefore two new stream names (WLB, Wallaby Creek, and PO1 Poley Creek) were added to the stream_names table (and Q71 revised); and five new stream lines added to mwstr and five new subcatchments added to subcs.

Twenty streamcodes such as 3E3 were easily misread by computers as scientific notation (e.g. 3E3 could be 3000), causing potential confusion. All 20 were converted to non-ambiguous codes in all tables. The altered codes and stream names are listed in Table B.1.

Finally we checked the stream layer and subcatchment spatial data for topological errors in ArcGIS. The “Repair Geometry” tool was used to remove null geometries. Topology checks found numerous overlaps in stream lines at catchment boundaries, self-intersections and overlaps in the subcatchment polygons, all of which we fixed manually.

Table B.1: Ambiguous stream codes using in v1.0 of the stream database and the new codes that replace them in v1.1.

Old code

New code

3E3

GL9

3E4

GHR

3E5

GFR

3E6

FSC

3E7

NE8

3E8

N6U

3E9

LFC

4E3

MQW

4E4

L9H

4E5

KRO

4E6

KNI

4E7

JQB

4E8

H89

5E3

FP3

5E4

FL4

5E5

EST

5E6

ESJ

5E7

DGO

5E8

YM6

5E9

NUI

The nextds_strcode field in the stream_names table was updated for all streams that drain to the sea. In v1.1 nextds_strcode equalled “sea”. This field now specifies the marine segment to which the stream drains or, in the case of several inland streams, the Goulburn catchment. The marine segments used in the table are listed in Table A2-2.

B.2 Version 1.3

I transformed the projection of all spatial tables from GDA 1994 MGA Zone 55 (EPSG 28355) to GDA 2020 MGA Zone 55 (EPSG 7855).

A small number of line and subcatchment corrections were made from 1.2 to 1.3, mostly at confluences that are in fact waterbodies, to make the stream network database more compatible with the waterbodies network (for DoR calculation). A new headwater stream (RJ3) was added as identified by Walsh et al. (2022).

The allus_ds table listing all upstream and downstream sites for each site in the subcatchment table included in earlier versions is omitted from version 1.3, as its utility has been replaced by implementation of the rapid igraph (C and Nepusz 2006) algorithm for returning upstream and downstream sites, and the method of rapid iterative compilation of full network data using the agg_order field in the subcs table.

The subcatchment raster, r_site, was revised using the corrected subcatchments, and the r_d2ol (distance to outlet) raster was revised to correct several compilation errors in the 1.2 version.

The major change for version 1.3 was the finalization and collation of the environmental variable tables following the delivery of the Nearmap land cover data in early 2023.

B.3 Version 1.3.1

New tables and new fields to existing tables were added for version 1.3.1. In all cases, the metadata table was updated to define these additions.

B.3.1 New tables

conn_subcs and conn_pps, used to calculate effective impervious variables (ei_*). (see Chapter 6, and new cat_env and subc_env fields below).

Seven linked tables quantifying stream barriers (stream_barriers, barriers_waterbodies, streamflow_gauges, barriers_pipe_conc, fishways, barriers_sources, and barrier_types) were added (see Chapter 10, and new subc_env fields below).

B.3.2 New fields in existing tables

Thirteen new fields have been added to the cat_env table:

  • sep_per_km2,c_basalt, and c_granite (see Chapter 5);

  • ei_2005, ei_2014, ei_2018 and ei_2022, eia_2005_m2, eia_2014_m2, eia_2018_m2 and eia_2022_m2 (see Chapter 6);

  • c_vol_dams_megalitre_2010 was added, and dor was renamed dor_2010 to indicate when the waterbodies data from which it was derived was estimated (the data was in fact compiled between 2009 and 2012). dor_res remained undated as all reservoirs used to calculate it are > 50 y old. (see Chapter 9). Note also corrections to dor_2010 below

Ten + new fields have been added to the subc_env table:

  • sc_eia_2005_m2, sc_eia_2014_m2, sc_eia_2018_m2 and sc_eia_2022_m2 (see above and Chapter 6);

  • n_full_barriers_ds_1996, n_part_barriers_ds_1996, n_full_barriers_ds_2023, and n_part_barriers_ds_2023 (see Chapter 10;

  • northness (see Chapter 5);

  • meant_y30_1950, meant_y30_1980 meant_y30_2010, and meant_y30_2022 (see section 5.8).

B.3.3 Data corrections

Table subcs

Errors in stream_length_m and channel_length_m fields were corrected (Most values for these variables were incorrect in v1.3. Other length values in the subcs and streams tables were correct.)

Table streams

A network correction was made to the nextdownid value in the streams table for hydroid 154195 (from 114824 to 114825).

The type field in the streams table for a piped reach of Eummemmering Creek [site IN (357656, 85847, 85778, 85774, 357660, 357658, 357662, 357664, 86639, 86611)] and all reaches of its tributary (strcode 2CF, actually a council stormwater pipe) was changed to pipe. They had been incorrectly classed as “channel” and “stream extensions” respectively.

Table stream_names

str_nm “CREEK 3851” renamed “GUYS CREEK (3851)” for it and its tributaries.

Table cat_env (and subc_env and r_tree_series)

Quality assessment of the Nearmap medium-high vegetation polygons revealed some areas of underestimation for 2022. Appendix 5 was added to the manual to outline the process of error detection and correction in deriving revised tree cover rasters for 2018 and 2022. These corrections resulted in changes to the r_tree_series raster and values of cat_env variables tf_2018, tf_2022, af_2018, af_2022, and subc_env variables tfa_2018_m2 and tfa_2022_m2.

The data underlying the dor_2010 and dor_res fields of cat_env were revised and corrected, resulting in widespread changes to these variables.

B.4 Version 1.3.2 (pending)

r_d2str raster revised by subtracting 5 from all gridcell values, so that gridcells nearest the stream have a value of 0 (and are thus weighted with a value of 1 in exponentially weighted metrics such as AF).

mwstr_dir <- "~/uomShare/wergStaff/ChrisW/mwstr_v13"
# system(paste0("cp ", mwstr_dir, "/r_d2str.tif ",
#        "/servers/shiny-server/archive/mwstr/mwstr_v1.3.1_r_d2str.tif"))
d2str <- terra::rast(paste0(mwstr_dir, "/r_d2str.tif"))
  # Gridcell nearest stream should have d2str = 0. This step was not done for af_2018 in version 1.3.1 (or 1.3)
  # a small section of the upper yarra already has correct 0 values. Leave them
d2str <- terra::ifel(d2str < 1, d2str, d2str - 5)
  # there were a few values just less than 5
d2str <- terra::ifel(d2str < 0, 0, d2str)
terra::writeRaster(d2str, "~/uomShare/wergStaff/ChrisW/mwstr_v13/r_d2str_tif")
# system(paste0("cp ~/uomShare/wergStaff/ChrisW/mwstr_v13/r_d2str.tif ",
#        "/servers/shiny-server/data/mwstr_v13"))