1 remote sensing and image processing: 9 dr. hassan j. eghbali

29
1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

Upload: elisabeth-wright

Post on 12-Jan-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

1

Remote Sensing and Image Processing: 9

Dr. Hassan J. Eghbali

Page 2: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

2

• Application– Remote sensing of terrestrial vegetation and the global

carbon cycle

Today…..

Dr. Hassan J. Eghbali

Page 3: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

3

Why carbon?

CO2, CH4 etc.greenhouse gasesImportance for understanding (and Kyoto etc...)Lots in oceans of course, but less dynamic AND less prone to anthropogenic disturbance

de/afforestationland use change (HUGE impact on dynamics)Impact on us more direct

Dr. Hassan J. Eghbali

Page 4: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

4

 

The Global Carbon Cycle (Pg C and Pg C/yr)

Atmosphere 730Accumulation + 3.2

Fossil fuels &cement production 6.3

Net terrestrialuptake 1.4

Net oceanuptake 1.7

Fossil organic carbon and minerals

Ocean store 38,000

Vegetation 500Soils & detritus 1,500

Runoff 0.8

Atmosphere ocean exchange 90

Atmosphere land exchange 120

Burial 0.2

(1 Pg = 1015 g)Dr. Hassan J. Eghbali

Page 5: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

5

CO2 – The missing sink

Dr. Hassan J. Eghbali

Page 6: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

6

CO2 – The Mauna Loa record

Dr. Hassan J. Eghbali

Page 7: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

7

Why carbon??

Thousands of Years (x1000)

180 ppm

280 ppm

Dr. Hassan J. Eghbali

Page 8: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

8

Why carbon?

• Cox et al., 2000 – suggests land could become huge source of carbon to atmosphere • see http://www.grida.no/climate/ipcc_tar/wg1/121.htm

Dr. Hassan J. Eghbali

Page 9: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

9

Why vegetation?

• Important part of terrestrial carbon cycle• Small amount BUT dynamic and of major

importance for humans – vegetation type (classification) (various) – vegetation amount (various) – primary production (C-fixation, food) – SW absorption (various) – temperature (growth limitation, water) – structure/height (radiation interception, roughness -

momentum transfer)

Dr. Hassan J. Eghbali

Page 10: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

10

Appropriate scales for monitoring

• spatial: – global land surface: ~143 x 106 km – 1km data sets = ~143 x 106 pixels – GCM can currently deal with 0.25o - 0.1o

grids (25-30km - 10km grid)

• temporal: – depends on dynamics – 1 month sampling required e.g. for crops

Dr. Hassan J. Eghbali

Page 11: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

11

So…… • Terrestrial carbon cycle is global• Temporal dynamics from seconds to millenia• Primary impact on surface is vegetation / soil system• So need monitoring at large scales, regularly, and

some way of monitoring vegetation……• Hence remote sensing….

– in conjunction with in situ measurement and modelling

Dr. Hassan J. Eghbali

Page 12: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

12

Back to carbon cycle Seen importance of vegetation

Can monitor from remote sensing using VIs (vegetation indices) for example

Relate to LAI (amount) and dynamics

BUT not directly measuring carbon at all…. So how do we combine with other measures

Dr. Hassan J. Eghbali

Page 13: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

13

Vegetation and carbon We can use complex models of carbon cycle

Driven by climate, land use, vegetation type and dynamics, soil etc.

Dynamic Global Vegetation Models (DGVMS)

Use EO data to provide…. Land cover Estimates of “phenology” veg. dynamics (e.g. LAI) Gross and net primary productivity (GPP/NPP)

Dr. Hassan J. Eghbali

Page 14: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

14

Basic carbon flux equations

• GPP = Gross Primary Production – Carbon acquired from photosynthesis

• NPP = Net Primary Production– NPP = GPP – plant respiration

• NEP = Net Ecosystem Production– NEP = NPP – soil respiration

Dr. Hassan J. Eghbali

Page 15: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

15

Basic carbon flux equations

• Units: mass/area/time– e.g. g/m2/day or mol/m2/s

• Sign: +ve = uptake – but not always!– GPP can only have one sign

Dr. Hassan J. Eghbali

Page 16: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

16

Dynamic Vegetation Models (DVMs)

• Assess impact of changing climate and land use scenarios on surface vegetation at global scale

• Couple with GCMs to provide predictive tool

• Very broad assumptions about vegetation behaviour (type, dynamics)

Dr. Hassan J. Eghbali

Page 17: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

17

Max Evaporation

Soil Moisture

Litter

Transpiration

Soil Moisture

LAI

Soil C & N NPPSoil

Moisture H2O30

Phenology

Hydrology NPP

Century Growth

e.g. SDGVM (Sheffield Dynamic Global Veg. Model – Woodward et al.)

Dr. Hassan J. Eghbali

Page 18: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

18

Potentials for integrating EO data• Driving model

– Vegetation dynamics i.e. phenology

• Parameter/state initialisation– E.g. land cover and vegetation type

• Comparison with model outputs– Compare NPP, GPP

• Data assimilation– Update model estimates and recalculate

Dr. Hassan J. Eghbali

Page 19: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

19

Parameter initialisation: land cover

EO derived land cover products are used to constrain the relative proportions of plant functional types that the

model predicts

evergreen forest

deciduous forest

shrubsgrasses crops

Land cover

PFTs

Dr. Hassan J. Eghbali

Page 20: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

20

Parameter initialisation: phenology

Day of year of

green-upSpring crops

Green up

Senescence

green-up occurs when the sum of growing degree days above some threshold temperature t is equal to n

Dr. Hassan J. Eghbali

Page 21: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

21

•MODIS Phenology 2001 (Zhang et al., RSE)

•Dynam. global veg. models driven by phenology

•This phenol. Based on NDVI trajectory....

greenup maturity

senescence dormancy

DOY 0

DOY 365

Dr. Hassan J. Eghbali

Page 22: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

22

Model/EO comparisons: GPPSimple models of carbon fluxes from EO data exist and thus provide a point of comparison between more complex models (e.g. SDGVM) and EO data e.g. for

GPP = e.fAPAR.PAR

e = photosynthetic efficiency of the canopy

PAR = photosynthetically active radiation

fAPAR = the fraction of PAR absorbed by the canopy (PAR.fAPAR=APAR)

Dr. Hassan J. Eghbali

Page 23: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

23

Model/EO comparisons: GPP

Dr. Hassan J. Eghbali

Page 24: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

24

Model/EO comparisons: NPP

Dr. Hassan J. Eghbali

Page 25: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

25

Summary: Current EO data Use global capability of MODIS, MISR,

AVHRR, SPOT-VGT...etc. Estimate vegetation cover (LAI) Dynamics (phenology, land use change etc.) Productivity (NPP) Disturbance (fire, deforestation etc.)

Compare with models and measurements AND/OR use to constrain/drive models

Dr. Hassan J. Eghbali

Page 26: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

26

Dr. Hassan J. Eghbali

Page 27: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

27

Future? OCO, NASA 2007

•Orbiting Carbon Observatory – measure global atmospheric columnar CO2 to 1ppm at 1x1 every 16-30 days

•http://oco.jpl.nasa.gov/index.html

Dr. Hassan J. Eghbali

Page 28: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

28

Future? Carbon3D 2009?

http://www.carbon3d.uni-jena.de/index.html

Dr. Hassan J. Eghbali

Page 29: 1 Remote Sensing and Image Processing: 9 Dr. Hassan J. Eghbali

29

Future? Carbon3D? 2009?

Dr. Hassan J. Eghbali