Downscaling

Question 1

What is downscaling?

Answer 1

Downscaling is the general name for a procedure to take information known at large scales to make predictions at local scales.

Question 2

Why are upscaling and downscaling both error prone?

Answer 2

Some information transmitted and some lost/altered.

Question 3

Why is it usually unhelpful to impose large grid cell data on small regions unchanged?

Answer 3

This is an extreme example of disaggregation error- values for many locations will be wrong.

Question 4

How can we go from large scale to small scale?

Two options for downscaling:

Answer 4

Dynamical downscaling

Empirical or statisitcal downscaling

Question 5

What is dynamical downscaling?

Answer 5

Regional models (RCMs) that run for smaller areas at higher spatial resolution are provided data from GCM simulations – called dynamical downscaling- much finer grid but can do topography much better.

Question 6

What is empirical or statistical downscaling?

Answer 6

Take underlying heterogeneity (e.g., elevation), calculate the detailed spatial differences and impose on the grid cell (e.g. apply the lapse rate to the surface temperature value) – called empirical or statistical downscaling- if a large climate model just sees something as a single surface you see no variation.

Question 7

Empirical downscaling

Advantages and disadvantages

Answer 7

Advantages: fine scale

Problem is relationships developed based on today’s observations don’t stay constant.

Question 8

Regional model downscaling

Advantagesand disadavtanges

Answer 8

Model has scalable grid- can increase resolution
Includes details in topography
Disadvantages: Input variables come from a GCM run (or modern climate data to test the model)- introduces errors.

Question 9

All models simulate the following:

Answer 9

1. Increase in treeline elevation – considerable loss of alpine tundra in mountains
2. Loss of shrub-steppe and xeric shrub vegetation in the Intermontane West and its replacement by savanna-steppe or open woodland (e.g. more trees)
   ?? Higher CO2 increasing water-use efficiency- allows plants to continue to grow in hot conditions- if drought stomata will close and growth is reduced but CO2 helps.

Question 10

What is the the Uk doing in terms of climate projections?

Answer 10

Uses regional interpolation driven by GCMs and a UK regional model
Basis for estimating likely impacts and for planning, e.g.,
- water supplies
- agriculture
- mortality/ health in the population
- energy demands
- biodiversity management

Question 11

For a given region (e.g., the South East or all the UK)

Answer 11

i) Projection period (e.g., 2020’s, 2050’s)
ii) Scenario (e.g., a high or a low emissions scenario)
iii) Probability level (e.g., 10%, 50%) – this is a bit tricky.

Question 12

At 10%…

Answer 12

only 10% of all the used projections fall at or below the value of the variable presented (say, an increase of 2°C).

Question 13

At 90%…

Answer 13

only 10% of projections fall at or above the value

Question 14

At 50%…

Answer 14

equal numbers of projections are above and below the value. This does not signify the most likely outcome – but the chances of the temperature change being greater or lesser than (our example, 2°C) are equal.

Question 15

As well as the climate projections, there is another website that help organisations plan their adaptation to climate change…

Answer 15

UKCIP

This is an extensive site – examples of impacts and solutions and how to plan adaptation

Question 16

Key points about downscaling

Answer 16

1. GCMs appear to simulate large-scale patterns OK but are not good at spatial detail
2. Using data at one scale to inform another scale is prone to error 3.For useful projections, downscaling needed (dynamic or statistical). But this depends upon the GCM, as well as the capability of the RCM in dynamical downscaling
3. Regional models are only as good as the data used to set them up (e.g. GCM output for experiments) – garbage in = garbage out 5.Empirical-statistical downscaling gives good detail – but assumes modern relationships hold in future
4. UK-CIP attempts to provide useful projections – but has backed off from detailed downscaling as this can be misleading.

Question 17

What is the main problem with ESMs?

Answer 17

They cannot run for a hundred years for the number of grid cells that they have if those grid cells are small.