Lecture 8 - Real World Examples & Insights

Long Bay Modelling

Long Bay has become a place for urban development, when it was previously rural land, especially for early Māori setters.

The urban development has led to increased sediment input in the shote, estuaries, and other water bodies.

• This then has ecological impacts on the species and ecosystem (interrupts breeding, worsens qater quality, and so on)

Various models were developed to help them understand different urban development scenarios and its possible effect to under what course of actions would be the best (lowest imapct on ecosystem, but maximises residential yield)

It allowed them to decide what parts of Long BAy should and shouldn’t have further development

• eg. places that were less risky aka. less erosion-prone, further away from water bodies, and so on were places where development should continue

• However, there were some uncertainties between models, but still served useful in forming decisions & planning

Overseer & Its Controversies

Was a model that regulatory entities tried to use to better understand and make decisions regarding fertiliser usage and its environmental impacts in New Zealand farms.

However, it wasn’t designed to specifically do this, instead, it was designed to quantify waterflow, water intakes, chemical intakes, and so on.

The way regulatory minsitries wanted to use it was outside of the scope of it’s design and function. In the end, it didn’t work out because it didn’t produce reliable data and there was mistrust between those who were pushing for it and people who saw the truth

Safeswim

A model that was used to understand the water quality of rivers, streams, beaches, and other urban water budies to communicate the safety of recreational use of said water bodies.

Prior to the development of safeswim, there were samples that were taken to try and record & understand water quality. However, it was systematically biased because of how they sampled the data: it was done in normal conditions & during times/occasions that wouldnt pose a risk to water quality.

• In other words, the way they sampled the data made it seem like the water quality was always good.

So, to develop the model, more sampling was done, specifically in specific places of interest where sampling would occur throughout the day (to account for tide cycles) and during (before & after too) different rain events (different types of rain)

This sampling data was then used to inform a predictive model that also acocunted for winds & sun that established a relationship between rain type and contamination (safety)

• It wasn’t perfect of course, at times, it would inaccurately predict when contamination events would occur.

• But, they were able to refine the model by constantly checking, comparing ,and validating what the model predicted w/ real world observations & data

• However, this model was only able to predict based on environmental conditions, it can take into account more random anthropological causes of contamiantion (that can occur too).

Stormwater Quality Management Tool

A model used to understand the likely stormwater contaminant loads & volumes to better inform maangement practises & allow regional councils to be more prepared with dealing with them.

Rural and urban spaces have different conditions that cause different patterns in stormwater load & contaminants. Different contamination parameters were used to identify how contamined stormwater could be , which then allowed us to see which areas are better off and which places needed required our attention

Then was this used to inform the costs of how much it’d be to improve the places of attention to argue for budget for the clean ups & implementing better management practises.

Things to Remember

• Purpose: having the right conceptual model for the job

• Good Sampling: having good data to produce good models

• Validation: know how reliable your predictions are

• Uncertainty: explain what this means for the predictions and how it can be used (is it limited by poor data?)

• Transparency: let people ‘look under the hood’ to build trust in your model

• Plain Language: explain clearly how the model works and what it’s predictions mean

• Teamwork: you can’t do all this alone!