System for bathing Water quality Modelling (SWIM)

31
System for bathing Water quality Modelling (SWIM) Professor Gregory O’Hare, School of Computer Science University College Dublin (UCD) SWIM Launch event, Titanic Centre, 7 th December 2017

Transcript of System for bathing Water quality Modelling (SWIM)

System for bathing Water quality Modelling (SWIM)

Professor Gregory O’Hare,

School of Computer Science

University College Dublin (UCD)

SWIM Launch event, Titanic Centre, 7th December 2017

SWIM Partnership

SWIM Partnership

SWIM Funding Instrument

Interreg VA United Kingdom-Ireland (Northern Ireland-Ireland-

Scotland)

2014-2020

Environment, Objective 2.2

Manage Marine Protected Areas and Species

The Directive recognises that short-term pollution of bathing

waters may arise (e.g., caused by high-rainfall events);

It allows up to 15% of the results for an assessment period to be

discounted (i.e., to be disregarded and not included in the

compliance calculations);

However discounting may be applied only to short-term

pollution that is PREDICTABLE.

Short-term pollution and‘discounting’

Predict short-term pollution events.

Warn the public not to bathe.

Objective: protect the public health.

The Directive’s Predict and Protect strategy

“Predict and Protect”

The Directive’s Predict and Discard Strategy

Predict short-term pollution events.

Warn the public not to bathe for health protection.

Then, the microbial results may be discarded, potentially

with the effect of protecting the bathing water

classification from downgrading.

A check sample and a replacement sample must be

taken.

Management measures must be taken.

Predictive models are required to underpin discounting

Their development has been undertaken within an INTERREG

IVA project SmartCoasts

The Partnership

Multiple Regression Modelling

Model 3 - Tolerance 0.2, r2 (adj.): 75.9%

Dependent (Y): Mean log10 Confirmed intestinal enterococci (cfu/100 ml)

Predictor Slope Value Partial r Sig. Tolerance

Intercept 5.459 .000

UVA Rad. in previous 12 Hrs. (MJ/sq. m) X1 -1.803 -.274 .000 .623

Max. Tide Height on sampling day (m) X2 .149 .244 .000 .945

Log10 Afon Afan Q in previous 48 Hrs (cub. m) X3 -1.233 -.440 .000 .303

Log10 Clyne R. Gauge. Max. Q in previous 48 Hrs (cub. m) X4 .679 .463 .000 .385

ETR in previous 24 Hrs. (MJ/sq. m) X5 -.015 -.145 .030 .656

Log10 Afon Dulais Max. Q on sampling day (cub m.) X6 .493 .257 .000 .311

Mean Wind Sp. in previous 24 Hrs (m/s) X7 -.105 -.155 .021 .742

Y = 5.459 – 1.803X1 + 0.149X2 – 1.233X3 + 0.679X4 – 0.015X5 + 0.493X6 - 0.105X7 ± 0.255

Intestinal enterococci

Intestinal enterococci

Confirmed Enterococci by Hour of Day

Decline in hourly GM towards the afternoon, increase in early evening

Model operationNRW Hydrometry

inputs

SCSC Installation Inputs

Calculated Variable Inputs

Black Pill

Clyne

Marcroft

Cilfrew

Workbook

Output

Sign:GoodPoor

Met Eireann station

Rainfall stationWeather station

Bray

WQ sampling points

The Dargle catchment has been instrumented with dynamically linked sensors, measuring

rainfall, river-flow and other parameters.

Met Eireann station

Rainfall stationWeather station

Bray

WQ sampling points

Met Eireann station

Rainfall stationWeather station

Bray

WQ sampling points

Hydrodynamic model: the ‘test-bed’ catchment

Sampling points – sampled

in 2011

STW outfalls – sampled in 2011

Sampling points – sampled

in 2012

STW outfalls – sampled in 2012

0

Parameters measured:

current speed and direction,

temperature, salinity, E. coli,

Intestinal Enterococci,

MST markers, and turbidity.

The Coastal Zone

Catchment models: NAM & MIKE11

Coastal model: MIKE3 FM

Legaslative Drivers

To ensure effective and efficient implementation of these directives, water resource managers need to know the water quality in order to take appropriate mitigating actions for social and ecological benefits in the event of pollution. This is particularly so for the Bathing Water Directive, where water quality is defined in terms of Escherichia coli and intestinal enterococci (IE) concentrations as percentile limit values.

SWIM Objective

The SWIM project will enable short-term pollution to be predicted, through the development of a bathing water quality prediction model and deriving from this, the capacity to inform the public through a series of media channels including text alerts and automatic web updates, and real-time communication via alert services delivered through electronic signage installed strategically at beach entrances.

This will help to protect public health, significantly improve communication to members of the public, and in doing so contribute to promoting tourism.

Legaslative Drivers

Achieving and maintaining high marine water quality standards is required under stringent EU environmental legislations (e.g., Bathing Water Directive (2006/7/EC), Shellfish Waters Directive (2006/113/EC), and Water Framework Directive (2000/60/EC).

Legaslative Drivers

To ensure effective and efficient implementation of these directives, water resource managers need to know the water quality in order to take appropriate mitigating actions for social and ecological benefits in the event of pollution. This is particularly so for the Bathing Water Directive, where water quality is defined in terms of Escherichia coli and intestinal enterococci (IE) concentrations as percentile limit values.

SWIM Approach

Acquire all pre-existing available bathing water microbial water-quality.

Determine sources of, and acquire all available retrospective relevant environmental data.

Determine which bathing waters had less than ‘Excellent’ classifications(category 1).

Determine which had one or more sample results that exceeded ‘Sufficient’ standard values (category 2).

Operate the Discard Model for categories 1 and 2.

Validate successful model performance.

Develop multivariate and other models where the Discard Model has not been successfully validated.

Beach Selection Rationale

Cranfield Rationale: Shared cross-border tidal water

BallyholmeRationale: High footfall. Public health risk. High amenity value (local boat clubs and organised swimming events)

Castlerock Rationale: Shared cross-border tidal water. Public health risk

Newcastle Rationale: High footfall. Public health risk. High amenity value (key NI tourist destination)

Crawfordsburn Rationale: High footfall. Public health risk.

Portrush (East Strand) Rationale: Shared cross-border tidal water. High footfall. Public health risk. High amenity value (local surf schools)

Enniscrone, Sligo Rationale: Public Health Risk

Lady’s Bay, Donegal Rationale: Public Health Risk and high footfall.

Anticipated Beaches

A number of beaches will be selected. The anticipated candidate beaches are:

Northern Ireland

Cranfield

Ballyholme

Castlerock

Newcastle

Crawfordsburn

Portrush (East Strand)

Republic of Ireland

Enniscrone, Sligo

Lady’s Bay, Donegal

Aviation Issues

Informing the Public

Intelligent

Orchestrated

Sensing

Predictive Modelling

Citizen

Engagement

Sampling

Multi-Sensor Deployment

Informing the Public

Manual signing at the DSP – 7 days

Weekdays - 09:00, 12:00, 15:00 GMT

Weekends – 09:00, 12:00 GMT

Sign wording is as per SEPA

Informing the Public

• IpV6 Addressable signage

• Personalised contextualized messaging using geofencing

• Alternate revenue streams

Towards Autonomic Middleware

NEXUS

Agents

HOTAIR

Agents

ACCESS

Agents

Agent Factory Framework

AF Deployment Process

System

Specification

Platform

Configuration

Community

Initialization

BBA

Development Kit

ALPHA

Development Kit

AFAPL

Development Kit…

Agent Factory Run-Time Environment

Agent System Architecture

FIPA-compliant Agent Platform

Collaborative Sensor Network

Sensor Network Visualisation

Shrink Wrapped Middleware Agents

Intelligent Infrastructure

Fro

m V

isualisatio

nto

Dep

loym

ent

Fro

m U

nco

nst

rain

ed t

o C

on

stra

ined

From Miniaturisation to Simulation…

Sensing Intelligently is Very Difficult

Very often we think of sensing as a relatively

straightforward process of data capture given

an appropriate sensing infrastructure.

However, this blind gathering of data is an

overly simplistic view, which naively fails to

consider the use to which the data will be put,

and the power envelope within which it must be

assembled.

The manner in which data is gathered should

be influenced by the ongoing use of that data in

a variety of application and user contexts.

In practice, this means that not only should

research provide mechanisms for harmonising,

synchronising, representing and filtering data

but it should also be moderated based upon

feedback resulting from its very usage.

Conclusion: To Dream by Day

“All people dream, but not equally. Those who dream by night in the dusty recesses of their mind, wake in the morning to find that it was vanity.

But the dreamers of the day are dangerous people, for they dream their dreams with open eyes, and make them come true.”

D.H. Lawrence