Upload
ami-houston
View
221
Download
3
Tags:
Embed Size (px)
Citation preview
*http://www.nlh.no/?avd=54
Problems using downscaled scenarios in studies of climate effects.
Lars Bakken
- The EACC project- Our first trials (and errors)*
More scenarios?Better scenarios?Simple wheather manipulations?Wheather generators?
COUP
Soil moisture
Soil temp.
Water transport
Snow cover
Waterstress
Frost/thawDriving data
Soil erosionPlant growthSoil biology- C- and N model-Trace gas model
The scenarios:
Dynamic downscaling (HIRHAM) based on: - AOGCM ECHAM4/OPYC3 with the GSDIO integration - IPCC IS92a scenario
The control period: 1980-99:Obs80-99 = Observed wheather 1980-99Scen80-99= Simulated weather 1980-99
The future wheather: Scen30-49= simulated wheather 2030-49
Snow depth Frost depth
Much less snowin a warmer climate
And soil frost is somewhat more shallow
Colder soils in a warmer climate?
Number of ”frost spells” in the soil:
More frequent freeze/thaw events in a warmer climate?
The summer is more problematic:The Scenarios are wet and cold!
And the Scenarios have a peculiar distribution of rain:
The cool and wet summers delayed the phenological developmentThe high and frequent precipitation resulted in moist grains=> Much delayed grain harvest!
The high Scenario-precipitation in July-August resulted in much too high drainage and surface runoff
High erosion!
High erosion in summer
High early-springerosion due to snowmelt on frozen soil
Quo vadis?
Don Quijote & Sancho Panza (Daumier)
Several scenarios => More uncertainty Longer scenarios => Better statistics
But the climate models will gradually become better…
- Simulations driven by climate scenarios tend to be anecdotic- the simulations are not causally transparent- general respons patterns cannot be ”extracted”
The difference between two scenarios are multidimensionalThe simulated ecological processes are nonlinear To understand the reasons for contrasting results is difficult
Alternative 1: Simple manipulations of observed wheather- Additive for temperature (seasonal og whole year)- Multiplicative for prcipiation- Or stochastic
Factorial model experiments => interactions detected
Winter temperature: ∆T winter= -1 0 +1Summer temperature: ∆Tsummer = -1 0 +1Winter precipitation : KPwinter = 0.9 1.0 1.1Summer precipitation: KPsummer = 0.9 1.0 1.1
All combinations: 81 simulations
Wheather generators
- Based on existing weather => reliable weather ”quality” ?
- Parameterized by contrasting climate scenarios plausible combinations ?
-Many or long time periods simulated=> Statistics will be OK