Project 3.2.5 - NESP Tropical Water Quality Hub

0 downloads 156 Views 1MB Size Report
Dec 2, 2017 - Indicators: Chlorophyll-a and Secchi Depth. • Scoring: Fixed capex (0.5x and 2x) modified amplitude meth
NESP PROJECT 3.2.5 - Improvements to the marine water quality metric for Reef Report Card Cedric Robillot*, Murray Logan, Mark Baird, Britta Schaffelke, Katherine Martin, Jane Waterhouse

20 November 2017

Project objectives and timelines • NESP Project 3.2.5 aimed to develop a new metric to report on GBR marine water quality condition as part of Reef Report Card • Phase 1 – June 2017 - Addressing limitations experienced with previous metric based on remote sensing optical properties • Phase 2 – December 2017 – Identify improvements required to develop a long-term metric for Report Card (from 2018)

20 November 2017

Perceived shortcomings of previous metric In 2016, the ISP expressed a lack of confidence in the WQ metric used in Report Card and agreed unanimously that a new approach was needed for the next Report Card (published in 2017). Perceived shortcomings included: • Only based on remote sensing-derived data and concerns about the ability to obtain accurate estimates from optically complex waters • Limited valid satellite observations in the wet season due to cloud cover • Limited to reporting on two indicators and not incorporating other water quality data collected through the MMP, IMOS and others • Outputs inconsistent with expected large scale properties (impact of rainfall and flooding history, gradient north-south) • Limitations of binary assessment of compliance relative to the water quality guidelines and aggregation and averaging over large spatial and temporal scales

20 November 2017

Approach Remote sensing in situ sampling Remote loggers eReefs BGC

TSS Chl-a

Data preparation

Secchi depth

Calculation of indices Aggregation options Success criteria?

Fit-forpurpose? Preferred scoring methodology

20 November 2017

Approach Remote sensing in situ sampling Remote loggers eReefs BGC

Assimilation of remote sensing reflectance observation data

TSS Chl-a

Data preparation

Secchi depth

TSS, Chl-a and secchi depth outputs of eReefs BGC

Calculation of indices Aggregation options

Calculation of indices and aggregation

20 November 2017

Activities • Collation and processing of remote sensing, eReefs BGC, in situ sampling and logger data for a range of parameters including Chl-a, TSS, Secchi depth and NOx • Selection of data centered around specific in situ monitoring locations and comparison of time series to assess suitability of source data • Analysis of eReefs model outputs, both as a control run (no data assimilation, from 2011) and as a reanalysis (assimilating remote sensing data, from 2012) • Comparison of index calculation methods including binary compliance, frequency of exceedance, maximum duration of exceedance and modified amplitude • Generation of reports and engagement with stakeholders and science panels

20 November 2017

Activities • Collation and processing of remote sensing, eReefs BGC, in situ sampling and logger data for a range of parameters including Chl-a, TSS, Secchi depth and NOx • Selection of data centered around specific in situ monitoring locations and comparison of time series to assess suitability of source data • Analysis of eReefs model outputs, both as a control run (no data assimilation, from 2011) and as a reanalysis (assimilating remote sensing data, from 2012) • Comparison of index calculation methods including binary compliance, frequency of exceedance, maximum duration of exceedance and modified amplitude • Generation of reports and engagement with stakeholders and science panels

20 November 2017

Comparison of data sources

20 November 2017

Activities • Collation and processing of remote sensing, eReefs BGC, in situ sampling and logger data for a range of parameters including Chl-a, TSS, Secchi depth and NOx • Selection of data centered around specific in situ monitoring locations and comparison of time series to assess suitability of source data • Analysis of eReefs model outputs, both as a control run (no data assimilation, from 2011) and as a reanalysis (assimilating remote sensing data, from 2012) • Comparison of index calculation methods including binary compliance, frequency of exceedance, maximum duration of exceedance and modified amplitude • Generation of reports and engagement with stakeholders and science panels

20 November 2017

eReefs data assimilated BGC

+ Simulated true colour eReefs 4km model 12:00 12 Sep 2013

Observed true colour MODIS - Aqua 13:20 12 Sep 2013

=

Assimilated true colour Assimilate RSR 12:00 12 Sep 2013

Assimilating model has all BGC quantities – nutrients, chl, TSS, pH, oxygen, production etc.

20 November 2017

eReefs data assimilated BGC Two estimates of state: • Control: this simulation only uses weather and river forcing, but does not include observations in the run. • Re-analysis: Mean of 108 member ensemble that uses the mis-match between the observed and simulated ratios of remote-sensing reflectance to move the model towards the observations in a manner consistent with the model itself.

Cyclone Dylan Up to 250,000 observations per 5 day cycle. When observations are low, errors are poorlyestimated, and reanalysis is similar to control.

20 November 2017

eReefs data assimilated BGC

20 November 2017

eReefs data assimilated BGC When sufficient observations are available to for model assessment, the spatial mean absolute error (MAE) of control and re-analysis is 0.2 ug/L Much of the domain is low, so mean absolute percent error (MAPE) is a better measure MAPE of control is around 50%, with the re-analysis at 42% better during the report card period

20 November 2017

Activities • Collation and processing of remote sensing, eReefs BGC, in situ sampling and logger data for a range of parameters including Chl-a, TSS, Secchi depth and NOx • Selection of data centered around specific in situ monitoring locations and comparison of time series to assess suitability of source data • Analysis of eReefs model outputs, both as a control run (no data assimilation, from 2011) and as a reanalysis (assimilating remote sensing data, from 2012) • Comparison of index calculation methods including binary compliance, frequency of exceedance, maximum duration of exceedance and modified amplitude • Generation of reports and engagement with stakeholders and science panels

20 November 2017

Metric calculation - Scoring

20 November 2017

Metric calculation - Scoring

20 November 2017

Metric calculation - Scoring

Simulated data and associated indices for threshold of 10 and small sample sizes (R=10) 20 November 2017

Metric calculation - Aggregation Temporal hierarchy DAILY

ANNUAL

Measure hierarchy MEASURE

SUB-INDICATOR

INDICATOR

Spatial hierarchy SITE

ZONE

GBR

20 November 2017

Metric calculation - Aggregation SITE (PIXEL)

ZONE

GBR

Measures

Measures

Measures

Sub-indicators

Sub-indicators

Sub-indicators

Indicators

Indicators

Indicators

20 November 2017

Metric calculation – Threshold values Water Body Open_Coastal_Cape York Open_Coastal_Wet Tropics Open_Coastal_Dry Tropics Open_Coastal_Mackay Whitsunday Open_Coastal_Fitzroy Open_Coastal_Burnett Mary Midshelf_Cape York Midshelf_Wet Tropics Midshelf_Dry Tropics Midshelf_Mackay Whitsunday Midshelf_Fitzroy Midshelf_Burnett Mary Offshore_Cape York Offshore_Wet Tropics Offshore_Dry Tropics Offshore_Mackay Whitsunday Offshore_Fitzroy Offshore_Burnett Mary

Indicator chl chl chl chl chl chl chl chl chl chl chl chl chl chl chl chl chl chl

Unit ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1 ugL-1

Annual 0.45 0.45 0.45 0.45 0.45 0.45 0.45 0.45 0.45 0.45 0.45 0.45 0.4 0.4 0.4 0.4 0.4 0.4

Justification/Source GBRMPA WQ guidelines used as threshold values.

GBRMPA WQ guidelines used as threshold values.

GBRMPA WQ guidelines used as threshold values.



20 November 2017

Metric calculation – Temporal aggregation “Water year” October 1st September 30th

20 November 2017

Report Card 2017 • Water year: October 1st to September 30th • Data sources: • eReefs BGC 4km resolution with assimilation of satellite reflectance using the MODIS sensor • In-situ river flow monitoring • Source Catchments concentration data using dedicated plug-in • Indicators: Chlorophyll-a and Secchi Depth • Scoring: Fixed capex (0.5x and 2x) modified amplitude method • Threshold values: Based on GBRMPA marine WQ guidelines • Aggregation zones: Open coastal water body in each of 6 NRM regions • Uniform grading scale from A to E

20 November 2017

Report Card 2017

20 November 2017

Report Card 2017

20 November 2017

Next steps – Future improvements • Suspended solids modelling (more and finer classes with different settling and resuspension properties) • Additional rivers to better represent the inputs on the catchment side, both in terms of flow and contaminants • Integrate 1km resolution BGC model and potentially RECOM (200m) to better represent each region and water body • Include a pesticide modelling framework using multiple conservative tracer and decay curve functionalities • Cover more years to get a better representation of the base levels • Contribute to the review of the water quality guidelines and establish appropriate threshold values and fixed caps for the modified amplitude scoring

20 November 2017

Next steps – Future improvements • Suspended solids modelling (more and finer classes with different settling and resuspension properties) • Additional rivers to better represent the inputs on the catchment side, both in terms of flow and contaminants • Integrate 1km resolution BGC model and potentially RECOM (200m) to better represent each region and water body • Include a pesticide modelling framework using multiple conservative tracer and decay curve functionalities • Cover more years to get a better representation of the base levels • Contribute to the review of the water quality guidelines and establish appropriate threshold values and fixed caps for the modified amplitude scoring

20 November 2017

Report Card 2017

20 November 2017

Next steps – Future improvements • Suspended solids modelling (more and finer classes with different settling and resuspension properties) • Additional rivers to better represent the inputs on the catchment side, both in terms of flow and contaminants • Integrate 1km resolution BGC model and potentially RECOM (200m) to better represent each region and water body • Include a pesticide modelling framework using multiple conservative tracer and decay curve functionalities • Cover more years to get a better representation of the base levels • Contribute to the review of the water quality guidelines and establish appropriate threshold values and fixed caps for the modified amplitude scoring

20 November 2017

Report Card 2017 Water quality Semi-distributed modelling (SWIFT)

Grid based modelling (G2G) Evaporation

Precipitation

Surface flow-routing

River

Saturation-excess surface runoff

Runoff- producing soil column

River flow Return flow

Drainage Subsurface flow-routing

Streamflow (ungauged)

Moore et al., 2006

Streamflow (gauged) Statistical water quality modelling 20 November 2017

Next steps – Future improvements • Suspended solids modelling (more and finer classes with different settling and resuspension properties) • Additional rivers to better represent the inputs on the catchment side, both in terms of flow and contaminants • Integrate 1km resolution BGC model and potentially RECOM (200m) to better represent each region and water body • Include a pesticide modelling framework using multiple conservative tracer and decay curve functionalities • Cover more years to get a better representation of the base levels • Contribute to the review of the water quality guidelines and establish appropriate threshold values and fixed caps for the modified amplitude scoring

20 November 2017

Report Card 2017

20 November 2017

Next steps – Future improvements • Suspended solids modelling (more and finer classes with different settling and resuspension properties) • Additional rivers to better represent the inputs on the catchment side, both in terms of flow and contaminants • Integrate 1km resolution BGC model and potentially RECOM (200m) to better represent each region and water body • Include a pesticide modelling framework using multiple conservative tracer and decay curve functionalities • Cover more years to get a better representation of the base levels • Contribute to the review of the water quality guidelines and establish appropriate threshold values and fixed caps for the modified amplitude scoring

20 November 2017

Next steps – Future improvements • Suspended solids modelling (more and finer classes with different settling and resuspension properties) • Additional rivers to better represent the inputs on the catchment side, both in terms of flow and contaminants • Integrate 1km resolution BGC model and potentially RECOM (200m) to better represent each region and water body • Include a pesticide modelling framework using multiple conservative tracer and decay curve functionalities • Cover additional years to get a better representation of the “climatology” • Contribute to the review of the water quality guidelines and establish appropriate threshold values and fixed caps for the modified amplitude scoring

20 November 2017

Next steps – Future improvements • Suspended solids modelling (more and finer classes with different settling and resuspension properties) • Additional rivers to better represent the inputs on the catchment side, both in terms of flow and contaminants • Integrate 1km resolution BGC model and potentially RECOM (200m) to better represent each region and water body • Include a pesticide modelling framework using multiple conservative tracer and decay curve functionalities • Cover additional years to get a better representation of the “climatology” • Contribute to the review of the water quality guidelines and establish appropriate threshold values and fixed caps for the modified amplitude scoring

20 November 2017

Next steps – Future improvements • Reference areas where in situ samples can be given more weight, where in situ sampling campaigns can take place and time series can be developed. It also allows to specifically identify sites that: • Are responsive or not to changes in various drivers/pressures • Encompass quality model/RS/in situ data (not too close to shore, appropriate depth etc…) • Can provide useful information spatially (good covariance, transects, flood discharges, supports model verification…)

• Sentinel 3 – Think about the overlap and finding ways to cover the previous climatology and events (cyclones etc…) • Continually updated index and indicator outputs (weekly, monthly?) can be an extension to the reference area concept

20 November 2017

Comparison of data sources

20 November 2017

Next steps – Future improvements • Reference areas where in situ samples can be given more weight, where in situ sampling campaigns can take place and time series can be developed. It also allows to specifically identify sites that: • Are responsive or not to changes in various drivers/pressures • Encompass quality model/RS/in situ data (not too close to shore, appropriate depth etc…) • Can provide useful information spatially (good covariance, transects, flood discharges, supports model verification…)

• Transition to Sentinel 3 series of satellites and sensors • Continually updated index and indicator outputs (weekly, monthly?) as a potential extension to the reference area concept

20 November 2017