**Draft** The effect of wet season river flows on flood plume distribution across northern Australia; contribution to primary productivity, and future extent under climate change. (NESP MaC 3.4, JCU)

Full description

This dataset consists of satellite imagery across northern Australia from 2003 to 2023, along with spatial files of the peak flood plume events from each wet season over that time. It covers the regions of the Flinders and Gilbert River outflows in the Gulf of Carpentaria, and the Daly River outflow in the Timor Sea (Anson Bay). It contains climate precipitation projections for both regions. Northern Australia is notable for its expansive river catchments, which represent vital components of the region's hydrological and ecological systems. The wet season, extending from November to April, constitutes a pivotal period during which copious rainfall inundates the landscape, profoundly impacting floodplains, and inland ecosystems, before emptying into the ocean and extending outward, sometimes for 100’s of kilometres, as large plumes of sediment and nutrient laden water. Many coastal processes, plants, and animal species, rely on this freshwater influx that reduces the hyper-salinity of the late dry season, and brings nutrients that drive the coastal and offshore productivity vital for fisheries species and migrating birds. In recent years, there has been increased interest in northern Australian river catchments as a resource for further water development to support activities such as irrigated agriculture and critical minerals mining. With climate change placing uncertainty around future rainfall levels in the region, understanding the relationship between wet season river flows and coastal flood plume extent is imperative for comprehending the ecological resilience of northern Australia to increased water allocation within catchments. This research aimed to identify how wet season rainfall events in the Flinders, Gilbert and Daly River catchments drives flood plume extent and primary productivity in the adjacent coastal seas. Further, it investigates how climate change could impact further the removal of environmental water across these regions. Methods: Hydrological data Hydrological flow data from the Flinders and Gilbert River catchments between 2003-2023 was retrieved from the Queensland Government Water Monitoring Information Portal (https://water-monitoring.information.qld.gov.au/). For the Daly River, hydrological data was extracted from the Bureau of Meteorology Water Data Online portal (http://www.bom.gov.au/waterdata/). Flow hydrographs were created for each river, and peak flow events for each year 2003- 2023 were identified. For the Gilbert River, flow levels for the years 2003-2015 (when downstream station 9170014A was not operational) were obtained from three upstream tributaries (Einasleigh, Etheridge and Gilbert Rivers, station numbers 917106A, 917104A and 917001D respectively), and the flow levels were summed. Calendar years (as opposed to fiscal years that are often used to encapsulate wet season rainfall) were used, as very little flow occurred before January each wet season, and the analysis was related to peak flow events of 7-day durations, not annual rainfall averages. Flood plume analysis A search was conducted for satellite imagery from the Moderate Resolution Imaging Spectroradiometer (MODIS) true colour, corrected radiance products, covering the period immediately following peak flow events. The MODIS satellite images from the south-eastern Gulf of Carpentaria region that encompassed the Flinders and Gilbert River outflows (Coordinate limits -18.580, 137.250, -14.361, 142.031) and from Anson Bay region that encompassed the Daly River outflow (Coordinate limits -12.673, 129.446, -14.009, 130.534) were downloaded in GeoTIFF format from the NASA Worldview Snapshots Portal (https://wvs.earthdata.nasa.gov/). The products were loaded onto ArcMap (ArcGIS Desktop 10.8) for spatial plume size analysis. Flood plume categories were defined as in Devlin and Schaffelke (2009), that were based on the plume concentration of water quality parameters that can be distinguished through ocean colour remote sensing, as follows: (i) Primary water types were defined as having a high total suspended mineral (TSM) load, minimal chlorophyll (Chl-a) and high coloured dissolved and organic matter (CDOM). (ii) Secondary water types were defined as a region where CDOM is still high however, the TSM has been reduced and increased light and nutrient availability has prompted phytoplankton growth. Thus, the secondary plume exhibits high Chl-a, high CDOM and low TSM. (iii) Tertiary water types are the region of the plume that exhibits no elevated TSM and reduced amounts of Chl-a and CDM when compared with that of the secondary plume. Tertiary plumes can be described as being the transition between a secondary plume and ambient conditions. Primary, secondary, and tertiary plumes were drawn using the ArcMap polygon tool, and the spatial extent in square kilometres for each plume type was calculated. This hand digitising method has been shown to provide better accuracy when determining plume boundaries over shallow benthic features compared to semi-automated processes that can struggle to distinguish seafloor from turbid water (Evans et al., 2012). To determine the relationship between flow and plume size, linear regression analyses were conducted between plume extent (sum of primary, secondary, and tertiary plumes) and 7-day hydrological flows. The analysis utilised complete hydrological data from 2003 to 2023 for the Flinders River and Daly River. For the Gilbert River, hydrological data from 2016 to 2023 only, was utilised. Primary productivity MODIS-aqua Level 2 ocean colour satellite products from the same satellite pass as the GeoTIFF products used to identify plumes, were downloaded from NASA (https://oceancolor.gsfc.nasa.gov) and the chlorophyll-a (chlor_a) bands opened in software package SeaDAS version 8.3.0 (NASA). The ‘chlor_a’ band, as described in Hu et al. (2019), returns the near-surface concentration of chlor_a in mg m-3 (µg L-1). Each product was subsetted to the same spatial extent which was a) the south-east section of the Gulf of Carpentaria defined by the oceanic region south of latitude 15.5°S and bounded by the mainland, and b) the region of Anson Bay in the Timor Sea defined by the latitudes 13.0°S to -13.55°S, and longitudes 129.7°E to 130.3°E. The statistical analysis tool was run to quantify mean chlor_a following each flow event. Because primary productivity commonly increases as secondary/ tertiary plumes expand seaward, multiple products encompassing lags of 3-5 days (where cloud-free satellite imagery was available) were used to assess quantifiable changes in the chlor_a spatial extent in the week following peak flow events. Regression analyses were conducted between 7-day flow during peak flood events and total plume size, and 7-day flow and tertiary plume size. Climate change and rainfall Future climate predictions were retrieved from the Climate Change Web Portal (Earth Systems Research Laboratory, 2014), developed by the National Oceanic and Atmospheric Administration’s (NOAA) Earth System Research Laboratory to collate and regionally downscale (to approximately 1° spatial resolution) the climate model outputs from the Coupled Model Intercomparison Project phase 6 (CMIP6; Eyring et al., 2016). The portal calculates the anomaly as the difference in the mean precipitation between the future climate (we used 2070–2099) and the model baseline reference period of 1985–2014, under the different Shared Socio-economic Pathways (SSP’s). Six climate models were chosen for use in the analysis as they have shown the best performance against historical climate for northern Australia (Ridder et al., 2021). The models used were CESM2-WACCM (NSF-DOE-NCAR, USA), FGOALS-G3 (Chinese Academy of Sciences), INM-CM5-0 (INM, Russia), ACCESS-CM2 (CSIRO-BOM, Australia), MRI-ESM2-0 (MPI-M, Germany) and NorESM2-MM (NCC, Norway). The scenario SSP5-8.5 was used as it projects the most global warming of all Shared Socio-economic Pathways, representing the continuation of a fossil fuel intensive world. In research applications, SSP5-8.5 is often used as the climate signal is strongest under this emissions scenario, making the signal most easily identifiable from the background noise of natural climate variability. Projected future rainfall anomalies, both seasonal and annual, were extracted for each climate model and the percentage increase or decrease in rainfall against baseline values (1995-2014) was calculated for each region (Gilbert, Flinders, and Daly catchments). The projected percentage reduction in rainfall was applied to the 7-day flow – plume size regression models, with the assumption that the percentage reduction in rainfall would be the same as the percentage reduction in flow, and the projected change in flood plume extent in square km calculated. Limitations of the data: This study has quantified the spatial extent of flood plumes in three catchments across northern Australia however, it does not quantify Total Suspended Sediments (TSS) i.e., the total discharge (in weight) of sediment. The analysis utilised complete hydrological data from 2003 to 2023 for the Flinders River and Daly River. For the Gilbert River, hydrological data from 2016 to 2023 only, was utilised (unavailable prior to this at the downstream site), which limits the strength of the resulting statistical analysis for the Gilbert. Climate change models for predicting future rainfall events across northern Australia do not offer consistent predictions for future rainfall, with model projections ranging from decreased rainfall to no change in rainfall, to increased rainfall, by 2079-2099. Whilst an ensemble of the models was used here, use of individual or different combinations of ensembles will result in different projections. Shape files (plume extent) – the Flinders and Gilbert plumes have separate shape files for primary, secondary, and tertiary plumes. The Daly shape files have one shape file for each entire plume with the primary, secondary and tertiary plumes available as selectable layers within the shape file. Overlap with a neighbouring river plume occurred for the Gilbert 2023 tertiary layer, with the Mitchell River to the north. An arbitrary cut off mark was chosen to distinguish between the two plume sources. An initial Area calculation of 3095km2 was reduced to 1644km2 (in the area2 attribute). Format of the data: 1.The satellite files consist of filenames in .txt only, with the actual files (~40mb each) available from (https://oceancolor.gsfc.nasa.gov). The file names contain the date of the satellite swath. 2.The flood plume shape files have spatial reference WGS_1984 Equal Earth Greenwich. 3. The Worldview snapshots are in Geo Tiff format. The name of the file contains the date of the satellite image. 4. The hydrological data and plume area size are in excel spreadsheets. 5. The climate anomaly files names are netCDF files in .nc format with r script for opening files provided. Filenames are generic and require opening to view climate model used. Climate master spreadsheet is in excel. 6. Code for statistical data analysis are in R files. Data dictionary: Flow analysis spreadsheets: FlowML(last7days)917014A : sum of flow in ML over the previous 7 days at hydro station # 917014A Primary: size in square km of primary flood plume Secondary: size in square km of secondary flood plume Tertiary: size in square km of tertiary flood plume 7daysum: sum of all stations flow over past 7 days (Gilbert River only) Plume_sum: sum in sq km of all three plume types (primary, secondary, tertiary) Chl: mean chlorophyll-a within spatial limits Chl_lag: mean chlorophyll-a 3-7 days post date of plume analysis Climate model anomalies spreadsheet: Future clim av: precipitation by 2079-99 Histclim/ hist_av: Historical rainfall at baseline 1985-2014. The number of entries are indicative of the size of the catchment with each entry covering 1° spatial area. Annual anom: Change by 2079-2099 from baseline Av sd: average and standard deviation of anomaly across the catchment %change: change between baseline and future rainfall eAtlas Processing: The original data were provided as follows: Climate Data: 95 nc files, 1 xlsx file, 1 R file Flow Analysis 3 x csv files, 1 R file Satellite data: 2 x txt files, 91 tif files Shapefiles: 162 shapefiles No modifications were made to the to the underlying data for the first three data folders (Climate data, Flow Analysis, Satellite Data) and the data packages are provided as submitted. Minor modifications were made to the shapefiles for eAtlas visualisation on the maps. Each of the three layers for year year (primary, secondary, tertiary) were compiled into a single layer with the naming convention 'River_year', i.e. 21 layers for each river. Attributes were added to each layer for mapping integration (River, Year, PlumeType, and Area_km2). Care was taken prior to compiling the year-composite layer where a single plume type layer had multiple vectors with individual areas. These areas were merged and the area values summed prior to joining the three plume type layers. All 21 layers were then joined for a composite River layer ('River_2003-2023_All-Plume-Types), and subsequent whole of project layer with all three rivers (including the 60 layers), 'NT-QLD_JCU_NESP-MaC-3.4_Flood-Plume_Gilbert-Daly-Flinders_2003-2023'. - As noted above in limitations for Gilbert 2023, area2 attribute was retained for the Area_km2 value for the tertiary vector in the Gilbert_2023 composite layer. - Erroneous vector areas were removed from the Gilbert 2007 layer (which had an area value of '0') and were covering the Flinders area. The version presented in the composite layer contains these changes for mapping consistency. - The original files for Gilbert 2009 contained two Tertiary files, and we confirmed the tert2 file was the correct version to use. The composite layers only contain the Gilb_2009_tert2.shp. Location of the data: This dataset is filed in the eAtlas enduring data repository at: data\\custodian\xxxx

Lineage

Maintenance and Update Frequency: asNeeded

Notes

Credit
The data collections described in this record are funded by the Australian Government Department of Climate Change, the Environment, Energy & Water (DCCEEW) through the NESP Marine and Coastal Hub. In addition to NESP (DCCEEW) funding, this project is matched by an equivalent amount of in-kind support and co-investment from project partners and collaborators.