The role of vegetation optimality in the Budyko-framework

Post on 03-Oct-2021

1 views 0 download

Transcript of The role of vegetation optimality in the Budyko-framework

MethodsIntroduction

The role of vegetation optimality in the Budyko-frameworkR.C. Nijzink1, S. Schymanski1

Hypotheses

Conc lusi ons

Budyko-framework

Vegetation Optimality

Study sites

Vegetation OptimalityModel

Results

Convergence to the curveby optimality

Modifying precipitation

Sensitivity n-values

Supported by the Luxembourg National Research Fund (FNR) ATTRACT programme (A16/SR/11254288)

Next

1 Luxembourg Institute of Science and Technology, Belvaux, Luxembourg,

Hydrological Models

Experiments

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

2

• Catchments around the world plot close to water and energy limits:

• Ea/P < 1

• Ea/P < Ep/P• Empirical curve by Budyko (1974)

THE BUDYKO FRAMEWORK

Water limit

Ene

rgy

limit

• Why do catchments converge to the curve?

• What happens under changing climate?

NextPrevious

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

3

• Different formulations of the curve• Parametric formulation:

THE BUDYKO FRAMEWORK

Water limit

Ene

rgy

limit

Budyko-parameter n• Widely assumed as a catchment

property• Changes with changes in

catchment properties?• Changes with climate?

NextPrevious

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

4

VEGETATION OPTIMALITY

Net Carbon Profit :Total difference of carbon uptake by photosynthesis and carbon costs of the system

AssimilationEvaporation

Carb

on cos ts

Root uptake

Vegetation Optimality Model: Optimizes vegetation properties to maximize NCP More info

NextPrevious

Available resources:• Water• Light

• CO2

Natural selection:• Optimally adapted vegetation• Uses resources in the best

possible way

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

5

RESEARCH QUESTIONS

Assimilation

Evaporation

Carb

on cos ts

Root uptake

NextPrevious

• Does optimality explain convergence on the Budyko-curve?

• Does climate change move a catchment along its individual curve?

• Does a change in vegetation properties result in shifting between curves?

Climate change?

Vegetation change?

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

6

HYPOTHESES

• Model simulations based on vegetation optimality lead to a better reproduction of the empirical Budyko-curve than model simulations without self-optimized vegetation.

• The empirical parameter n stays constant as precipitation changes, as long as vegetation and other meteorological forcing variables stay constant.

• Changes in n-values are a result of slowly varying, long-term vegetation properties.

Go to results ➔

Go to results ➔

Go to results ➔

NextPrevious

Go to results ➔

Go to results ➔

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

7

North Australian Tropical Transect • Mean annual rainfall: 500-1800 mm• Pronounced wet season: Nov-Feb• Evergreen trees + seasonal grass• Sites:

• Five flux tower sites

• Six catchments

• 36 additional locations

CAMELS-data• Catchments around the contiguous

United States• Selection of 357 catchments

STUDY SITES

NextPrevious

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

8

Tre

e r

oo

tin

g d

epth

Tree coverGrass cov.

Gra

ss r

oo

tin

gd

ep

th

Root distributions

VEGETATION OPTIMALITY MODEL

Optimized constants• Tree cover fraction• Tree rooting depth• Grass rooting depth• Water use strategies

Dynamically optimized variables:• Grass cover fraction• Photosynthetic capacity• Stomatal conductances• Fine root surface area

NextPrevious

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

9

FLEX

CONCEPTUAL HYDROLOGICAL MODELS

GR4J

Perrin, Michel, and Andréassian. “Improvement of a Parsimonious Model for Streamflow Simulation.” JoH 279, no. 1–4 (2003): 275–89. https://doi.org/10.1016/S0022-1694(03)00225-7.

NextPrevious

TUW (HBV)

• Simple bucket-models• Calibrated• Applied to:

– Australian catchments

– CAMELS-catchments

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

10

EXPERIMENTS

Unmodified situation• Optimize VOM to maximize the Net Carbon Profit• Calibrate hydrological models to observed streamflow

Increase/decrease precipitation• Run VOM:

➔ Vegetation from unmodified situation• Run hydrological models:

➔ Parameters from unmodified situation

Let vegetation adjust…• Re-optimize VOM for new precipitation

NextPrevious

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

11

CONVERGENGE BY VEGETATION OPTIMALITY

Flux tower sites

Australian catchments

Extra locations NATT

NextPrevious

• VOM with full optimization of vegetation properties• VOM without vegetation → bare soil

Optimizing vegetation leads to a higher curve!

Higher and more realistic n-values for optimized vegetation!

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

12

MODIFYING PRECIPITATION• Rainfall multiplied: 0.2 - 2.0, steps of 0.2• VOM with constant long-term vegetation• VOM with re-optimized vegetation for new precipitation

Howard Springs

Optimizing vegetation leads to a lower standard deviation!

Non-optimal vegetation deviates from curve!

NextPrevious

See also:

Adelaide River

Daly Uncleared

Dry River

Sturt Plains

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

13

MODIFYING PRECIPITATION

Curve moves down for increased precipitation...

…but moves back if vegetation re-optimizes!

• 36 additional locations, precipitation +20%• VOM with constant vegetation• VOM with re-optimized vegetation

NextPrevious

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

14

MODIFYING PRECIPITATION• 36 additional locations, precipitation +20%• VOM with constant vegetation• VOM with re-optimized vegetation

Water use strategy parameters

Perennial vegetation cover

Perennial vegetation rooting depth

Seasonal vegetation rooting depth

Biggest changes in perennial vegetation properties

NextPrevious

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

15

MODIFYING PRECIPITATION• Australian catchments • Prec. multiplied: 0.2 - 2.0, steps of 0.2 • VOM with constant vegetation• VOM with re-optimized vegetation• Hydrological models with constant model parameters

Self-optimized vegetation has the best fit!

Adelaide River

NextPrevious

See also:

Dry River

Fergusson River

Magela Creek

Seventeen Mile Creek

South AlligatorCreek

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

16

SENSITIVITY N-VALUES: VOM• Prec. multiplied: 0.2 - 2.0, steps of 0.2 • Budyko-parameter determined for each case, each site• VOM with constant vegetation• VOM with re-optimized vegetation

Re-optimizing vegetation results in constant n

Factors for m

ultip

licati on o

f precipitatio

n

NextPrevious

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

17

SENSITIVITY N-VALUES: ALL MODELS• Prec. multiplied:

• 0.2 - 2.0, steps of 0.2 • n-value:

• each case, each site• VOM:

• constant vegetation

• optimized vegetation• Hydrological models

• constant parameters

Optimized VOM always around one value!

Factors for m

ultiplica

ti on

of

precipitatio n

VOM optimized

VOM not optimized

FLEX

TUW

GR4J

NextPrevious

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

18

SENSITIVITY N-VALUES: MORE LOCATIONS VOM• 36 additional locations • VOM runs:

• Optimized for unmodified precipitation.

• Constant vegetation and increased prec. +20%

• Re-optimized vegetation and increased prec. +20%

• n-values for each case:

• Difference with optimized VOM and unmodified prec.

Re-optimized VOM for increased precipitation returns to the initial n-value!

NextPrevious

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

19

SENSITIVITY N-VALUES: CAMELS-DATA• CAMELS-data• Prec. +20%• Hydrological models with constant parameters• n-value for each catchment

Increasing precipitation results in lower in n-values: happens for all models!

NextPrevious

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

20

• Model simulations based on vegetation optimality lead to a better reproduction of the empirical Budyko-curve than model simulations without self-optimized vegetation.Accepted

• The empirical parameter n stays constant as precipitation changes, as long as vegetation and other meteorological forcing variables stay constant.Rejected

• Changes in n-values are a result of slowly varying, long-term vegetation properties.Rejected

CONCLUSIONS

NextPrevious

Back to results ➔

Back to results ➔Back to results ➔

Back to results ➔

Back to results ➔

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

21

APPENDIX

AssimilationEvaporation

Carbo

n cos ts

Root uptake

HomePrevious Next

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

22

MODIFYING PRECIPITATION• Rainfall multiplied: 0.2 - 2.0, steps of 0.2• VOM with constant vegetation• VOM with re-optimized vegetation

Adelaide RiverNext

Back

Previous

Non-optimal vegetation deviates from curve!

Optimizing vegetation leads to a lower standard deviation!

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

23

MODIFYING PRECIPITATION• Rainfall multiplied: 0.2 - 2.0, steps of 0.2• VOM with constant vegetation• VOM with re-optimized vegetation

Daly RiverNext

Back

Previous

Non-optimal vegetation deviates from curve!

Optimizing vegetation leads to a lower standard deviation!

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

24

MODIFYING PRECIPITATION• Rainfall multiplied: 0.2 - 2.0, steps of 0.2• VOM with constant vegetation• VOM with re-optimized vegetation

Dry River

Back

Previous Next

Non-optimal vegetation deviates from curve!

Optimizing vegetation leads to a lower standard deviation!

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

25

MODIFYING PRECIPITATION• Rainfall multiplied: 0.2 - 2.0, steps of 0.2• VOM with constant vegetation• VOM with re-optimized vegetation

Sturt Plains

Back

Previous Next

Optimizing vegetation leads to a lower standard deviation!

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

26

MODIFYING PRECIPITATION• Australian catchments • Prec. multiplied: 0.2 - 2.0, steps of 0.2 • VOM with constant vegetation• VOM with re-optimized vegetation• Hydrological models with constant model parameters

Optimized VOM has the best fit!

Dry River

Back

Previous Next

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

27

MODIFYING PRECIPITATION• Australian catchments • Prec. multiplied: 0.2 - 2.0, steps of 0.2 • VOM with constant vegetation• VOM with re-optimized vegetation• Hydrological models with constant model parameters

Optimized VOM has the best fit!

Fergusson River

Back

Previous Next

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

28

MODIFYING PRECIPITATION• Australian catchments • Prec. multiplied: 0.2 - 2.0, steps of 0.2 • VOM with constant vegetation• VOM with re-optimized vegetation• Hydrological models with constant model parameters

Optimized VOM has the best fit!

Magela Creek

Back

Previous Next

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

29

MODIFYING PRECIPITATION• Australian catchments • Prec. multiplied: 0.2 - 2.0, steps of 0.2 • VOM with constant vegetation• VOM with re-optimized vegetation• Hydrological models with constant model parameters

Optimized VOM has the best fit!

Seventeen Mile Creek

Back

Previous Next

MethodsMethods

ResultsResults

ConclusionsConclusions

IntroductionIntroduction

Home

AppendixAppendix

30

MODIFYING PRECIPITATION• Australian catchments • Prec. multiplied: 0.2 - 2.0, steps of 0.2 • VOM with constant vegetation• VOM with re-optimized vegetation• Hydrological models with constant model parameters

Optimized VOM has the best fit!

South Alligator River

NextPrevious

Back