1.1.1

Test release for update to tide modelling file directory name

1.1.0

1.1.0 (2020 annual shoreline)

In December 2021, the DEA Coastlines product was updated to version 1.1.0. This includes the following key changes to the data and web services:

Improvements and additions:

• Added annual shoreline data for 2020. The 2020 shoreline is an interim dataset that is subject to change, and will be updated to a final version in the following 2021 DEA Coastlines update.
• Implemented a new temporal filter that improves mapping performance across small islands and dynamic sandbars while reducing noise in areas with poor satellite coverage (#35).
• Time series graphs are now available for non-significant rates of change points on DEA Maps (previously this was only available for retreating and growing points) (#40).
• Australian Coastal Geomorphology Smartline and Australian Coastal Sediment Compartments datasets are now available in the "Supplementary data" folder on DEA Maps to assist in interpreting observed patterns of coastal change.
• Added new valid_obs and valid_span fields to rates of change points that summarise the total number and temporal coverage of valid non-outlier shoreline observations. Results from valid_obs are now used to filter out unreliable rates of change points on DEA Maps by removing points with fewer than 25 valid observations (#38).
• Climate indices (e.g. SOI) are now detrended prior to regression to remove false positives associated with trends in index values over time, and to allow climate index relationships to be modelled for coastlines with temporal erosional or progradation trends through time (#42).
• Numerous minor improvements to inland water and estuary masking to fix areas of missing shoreline data and remove non-coastal features (#32, #33, #36).

Backwards incompatible changes:

• Removed experimental breaks field from rates of change points (#41).
• Removed retreat, growth fields from rates of change points (this data is already captured in rate_time, with negative values indicating retreat and positive values indicating growth).
• Updated WMS layer group names from "dea:DEACoastLines" to "dea:DEACoastlines", and "dea:AnnualCoastlines" to "dea:AnnualShorelines"

1.0.0

Official 1.0.0 release (2019 annual shoreline; 15 March, 2020)

Official version 1.0.0 of DEA Coastlines code base, as submitted to Remote Sensing of Environment.

Updates

• Implemented morphological operators to reduce false positive cloud detection along narrow, bright sandy beaches (see #37)

1.0.0-beta

1.0.0-beta (1 Oct, 2020)

Updates

• Added the following new columns to "Statistics" points:
  • rate_soi, sig_soi, se_soi, outl_soi: Columns giving the results of regression analysis of annual distances from the 2019 baseline coastline and the Southern Oscillation Index. This can be used to plot the spatial footprint of ENSO on coastal change (e.g. indicating if and by how much coastlines retreat during La Nina years). (#24)
  • breaks: A column listing any years that were identified as breakpoints. This can be useful for verifying that a significant trend is indeed linear, or identifying areas of rapid non-linear change e.g. associated with coastal development or management (experimental, interpret with care) (#21)
• Continental summary includes SOI as well as temporal rates of change (#27)
• Code cleaned to use virtual products and remove dependency on dea-notebooks code (#26)
• Add support for multiple polygons and file export for DEA CoastLines utility notebook (#16)
• Minor updates to waterbody mask to exclude additional poorly mapped tropical rivers
• Fully documented all script functions (#23)
• Re-written readme doc to follow official product description and provide examples of how to load data from WFS in both R and Python

Bug fixes

• Greatly reduced data gaps caused by an issue with the way nodata pixels were identified for gapfill years (#19)

0.3.0

0.3.0 (July 15, 2020)

Run on almost complete GA Landsat Collection 3 archive (v.s. c3-samples database that was missing many scenes).

Major updates

• Added the following new columns to "Statistics" points:
  • se_time: gives standard error of regression, will be used to add error bars to rates of change (#7)
  • growth/retreat: booleans indicating whether a point eroded or prograded, will be used in Terria config for plain english descriptions of change (#6 )
  • sce: Shoreline Change Envelope; gives the entire distance across which shorelines changed over the time series (#13)
  • nsm: Net Shoreline Movement; gives the difference between the first and the last valid shoreline position (#15)
  • max_year/min_year: gives the year where shorelines were at their most land-ward and most sea-ward (#15)
• Set all 1991/1992 annual coastlines to "uncertain" northward of -23 degrees latitude on account of Mt Pinatubo aerosol issue (#12)
• Standardise all continental outputs to use Australian Albers EPSG:3577 projection (#9)
• Renamed year columns with dist_ prefix to clarify their purpose (i.e. distances from baseline) (#5)
• Add notebook utility tool based on DEA CoastLines WFS for interactively drawing profiles and returning transects (#3, #8)

Bug fixes

• Restored missing contours on western end of Rottnest Island, WA
• Removed Papua New Guinea territory from data
• Noise caused by low data availability is greatly reduced, particularly in southern Tasmania and Cambridge Gulf

Outstanding issues

• Two tiles fail due to KeyError in skimage.measure.find_contours (#11)

0.2.0

0.2.0 (May 5, 2020)

Features

• Add 'uncertainty' column to contour data for pixels with either less than 5 observations or > 0.25 MNDWI standard deviation in more than 50% of years
• Add new DEACoastLines_utilities.ipynb notebook containing additional functions for using DEA CoastLines data
• Add rates_hist utility function for plotting DEACoastLine statistics data as a histogram based on an interactive selection
• Add ability to modify waterbody mask by providing a vector files of features to add or remove from the mask
• Add new deacoastlines_summary.py script for generating continental summary of point statistics based on average values within a buffer of X km of a subset of points. This allows hotspots to be visualised at full zoom out

Bug fixes

• In deacoastlines_generation.py: Fixed missing grid cell areas by buffering both the input grid cell extent and tide modelling extent by 0.05 degrees (i.e. a total of 0.10 degrees). This ensures that enough tidal modelling points are available for interpolation, and improves tidal modelling consistency between neighbouring grid cells
• In deacoastlines_generation.py: Break script early if grid cell correspondes to 1 or less tidal modelling points, as this makes tidal interpolation impossible
• In deacoastlines_generation.py: Increase dask_chunks to {'time': 1, 'x': 2000, 'y': 2000} for improved data load performance
• In deacoastlines_statistics.py: Fixed missing shorelines caused by the waterbody mask by limiting mask to specific waterbody features ('Aquaculture Area', 'Estuary', 'Watercourse Area', 'Salt Evaporator', 'Settling Pond' and perennial 'Lakes')
• In deacoastlines_statistics.py: Fix CRS of exported GeoJSON contours and statistics files by converting to EPSG:4326
• In deacoastlines_statistics.py: Remove previous zeroing of yearly distance values to 1988. Yearly distances are now relative to the 2018 baseline (e.g. 0 for the 2018 contour), which should be simpler easier to interpret
• In deacoastlines_statistics.py: Increase min_vertices for contour extraction to 30 to reduce noise
• In deacoastlines_statistics.py: Move vector directory creation until after data load, so no directory is created if rasters do not exist

v0.1.0 (April 1, 2020)

Features

• First continental run of DEACoastLines
• Outlier detection now uses a more robust Median Absolute Deviation method

Bug fixes

• In deacoastlines_statistics.py: Remove mask applied to contour generation due to excessive missing data