Lecture 3: Predictive Models and Trends

Question 1

Forecast

Answer 1

The public announcement that a hazardous event is likely to occur during a specified period, commonly with a statement of probability
ie. There is a 10% chance the forest will be burned unintentionally this summer, with 95% confidence
Rain is the most difficult to measure

Question 2

Hindcast

Answer 2

Means of testing models, evaluate uncertainty
Uses data from past events, even if the short-term situation is monitored precisely
Usually the longer the record, the better. For weather hazards, usually try for 40+ years, for seismic hazards, may be hundreds of thousands of years
Helps define magnitude limits, boundary conditions
Need to capture events with high rates of change, or anomalous events, as these rare events may be the important ones

Question 3

Prediction

Answer 3

A statement regarding the occurrence of a process or outcome . May specify time, location, intensity, or magnitude.
Can be hindcast or forecast
May be considered more specific than a forecast

Question 4

Warning

Answer 4

An announcement that a hazard event is happening or expected to happen in the near future in a defined regio

Question 5

Watch

Answer 5

An alert issued by a government agency that weather or other environmental conditions are favourable for a hazardous event in a general region

Question 6

Hazard trends: why are there increased losses in modern times?

Answer 6

We are more vulnerable

Higher population density and more infrastructure

Question 7

What hazards are we becoming less vulnerable to?

Answer 7

Geological
Better warnings, building codes
As well, geological events occur over much longer time periods

Question 8

Data

Answer 8

An archive of data from years of observation of past events
Current parameters related to the intensity and location of the actual hazard
Current parameters that are indirectly related
Precursor indications

Question 9

Models

Answer 9

Numerical, physical (analogue)
Trends or relationships derived from past events
Complex, non-linear, chaotic relationships
Simulations, tests, hindcasts

Question 10

Analogue (physical) models

Answer 10

Use powder and gel that have same property as rocks (on very large scale)
Make layers, then deform it
Add magnetic grains to sand, to detect magnetic fields – put into computer to get 3D analog model
Examine the processes along a fault type
Simulate and study a specific fault
Predict where earthquakes may occur
Predict the magnitude of those earthquakes
Predict patterns in space and time

Question 11

Numerical model

Answer 11

· “Algorithms”
Uses forward and inverse modeling
Typically involves significant computation
Convert (empirical) observations to (theory) equations
Make models more complex to include more factors

Question 12

Examples of numerical models (2)

Answer 12

1. Dept Oceanography Storm Surge Model (Keith Thompson)

2. Dept Oceanography/Earth Sciences Geodynamics Model (Chris Beaumont)

Question 13

Path to prediction

Answer 13

1. Observation of past events (verify data, hypothesis, test hypotheses, parameterize variables, develop rules)
2. Model (test hypothesis, hindcast, evaluate uncertainty, forecast)
3. Prediction (alert regional and local office, continue monitoring, issue warning, update model)

Question 14

What are forecasts based upon?

Answer 14

Our forecasts are based on the premise that those global oceanic and atmospheric conditions which preceded comparatively active or inactive hurricane seasons in the past provide meaningful information about similar trends in future seasons
Data, models