Author:Antonin Kusbach (Tony)

Supervisors:A. Persson, J. Connolly

Dept. of Physical Geography and Ecosystem Science, LU

Analysis of Arctic peak-season carbon flux estimations based on four MODIS vegetation indices

June 8, 2016

Project background

• Pan-Arctic Net Ecosystem Exchange (PANEEx 2015)

• Aim: improve understanding of CO2 dynamics in the Arctic

• Upscale satellite-derived data to produce NEE (Net Ecosystem Exchange)

Why study C in the Arctic ?

• high-latitude boreal and tundra biomes: sequester a atmospheric CO2

– 119 Pg of soil organic carbon• Importance of the Arctic:

– Stable stocks of C (if frozen)– Heat budget

• Delicate region: Arctic amplification(Screen & Simmonds, 2010; Serreze et al., 2009)

Introduction

• NEE – exchange of C between land and atmosphere– Uptake, release (-,+) – Measured in g/m2/yr or µmol m2/s

• Arctic regarded as a net C sink– future? (McGuire et al., 2010; Kimball et al., 2009)

• Arctic tundra C sequestration:– Abisko: -1.74 µmol m-2 s-1 (Stoy et al., 2013) Source: AMAP 2012

Measurements of C exchange• Eddy covariance towers

(Baldochi, 2003)

• Gas chambers (Williams et al., 2006)

• Environmental modelling

(Mbufong,In Prep; Kimball et al., 2009)

Site: Barrow, Alaska, source: SpecNet (http://specnet.info/)

• PIRT - Photosyntesis & temp. sensitive (Williams et al., 2006)

• TCF - GPP, soil temperature (Kimball et al., 2009)

• PANEEx - Air temp, LAI, PAR (Mbufong, In

Prep)

Examples of C models

• PANEEx - Air temp, LAI, PAR

LAI: ~0-2[m2 / m2]

Air temperature: ~ 7-10 [ ͦC]

PAR: light photon density [W / m2]

Problem statement

• Vegetation properties are important in understanding of C exchanged in the Arctic

(Mbufong et al., 2015)

• Need for an accurate vegetation proxy (LAI, NDVI) - Which one, based on what ?

Main objectives

• Objective: (1) Reduce knowledge gap between RS and environmental modelling

(2) Assess suitability of satellite-derived products

• Goal: Model and evaluate Arctic NEE/LAI at 12 sites

• Assumptions:(i) Vegetation proxies are expected to yield discrepancies in the Arctic

NEE estimations

(ii) The product with the finest spatial resolution is likely to yield most realistic NEE estimations compared to in situ

Product Spatial Resolution Temporal resolution Source

MCD15A3 - LAI 1 km2 4 - day LP DAAC, USGS*

MOD13A1 - NDVI 500 m2 16-day GEE, MODIS Terra

MOD13Q1 - NDVI 250 m2 16-day GEE, MODIS Terra

MYD09GA - NDVI 1 km2 Daily GEE, MODIS Aqua

Methodology, used materials

* LP DAAC – Land Processes Distributed Active Archive Center

• Scope: 2008-2010

• Peak of the growing season: July

• NDVI to LAI conversion for consistency (Van Wijk & Williams, 2005)

NEE model Study domain

• Sensitivity analysis: selective (non-probability) sampling

• Quantitative analysis of “big data”

• Average daily NEE [µmol m2/s]

(Mbufong et al., 2014)

Air temp: Fcsat

LAI: Fcsat

PAR:

NEE

Methodology, used software

• ESRI ArcGIS (vs. 10.2.2 Desktop)• (Google Inc.)• Matlab (vs. R2014a, The MathWorks, Inc.)

Results, LAI

Results, NEE

Slope = 0.29 Intercept = -0.94r2 = 0.15; rmse = 0.30

NEE

Modelled vs. Observed NEE

0 - 0.25

0.25 - 0.5

0.50 - 0.75

0.75 - 1

1.1 - 1.5

1.5 - 2

2.1 - 2.5

2.5 - 3.5

Land fill

Water bodies

0.16

 

0.87

 

1.7

 

1.35

 

0.18

 

0.960.11

 

1.62

 

In situ LAI : ~0.24 & ~ 0.63

Results summary

• LAI – under & overestimated; 1/4 correlations is consistently overestimated (slope = 2.45, R2 = 0.8) in MCD15A3

• LAI – MOD13Q1 shows least overestimation (slope =1.22, R2 = 0.59, RMSE = 0.23)

• NEE- MCD15A3, statistically insignificant (P>> 0.05)• NEE – MOD13Q1 & MYD09GA show best estimations

(slope=0.78, R2=0.73; slope= 0.86, R2 = 0.56), p<0.05

Discussion – other studies• Implementation of 250 m NDVI better than 1 km GPP

Schubert et al. (2012) • Watts (2014)

– used 13Q1 & 13A1 (250 m) compared to EC

(R2 = 0.8, p < 0.05)

– halved the acquisition time by combining 13Q1 & 13A1 (linear interpolation)

• Arctic heterogeneity 500 m and 250 m vegetation indices may not be sufficient

Discussion - methods• Need for sufficient number of observed

data in global scale(n=12) • Spatial misalignment of external data

(LAI, PAR)– Edge effect, miscalculated pixels

• LAI derivation from NDVI (97% of in situ variation, Abisko) (Van Wijk & Williams, 2005)

• Spectral errors imbedded in LAI images

• Investigation of finer-resolution satellites (Sentinel 2 project) www.sentinel.esa

Conclusions

• Three MODIS products generated NEE (p<0.05)

• Best estimation: 250 m 16-day LAI (MOD13Q1)- Slope = 0.78, R2=0.73

• Spatial resolution is important for result accuracy- Need for finer scale (30 to 60m)

• Heterogeneity of Arctic landscape can be modelled via RS; accuracy of NEE is correlated to the configuration of sat.-derived products

THANK YOU FOR LISTENING

Methodology cont.

• Air Temperature (Ta)

Processing PAR

• 00:00• 03:00• 06:00• 09:00• 12:00• 15:00• 18:00• 21:00

Fine 3-hr resolution creates illumination shadows

References

• Baldocchi, D.D. 2003. Assessing the eddy covariance technique for evaluatiing carbon dioxide exchange rates of ecosystems: past, present and future. Glob. Change Biol 9:479-92.

• McGuire, A.D., Hayes, D.J., Kicklighter, D.W., Manizza, M., Zhuang, Q., Chen, M., Follows, M.J., Gurney, K.R., McClelland, J.W., Melillo, J.M., Peterson, B.J., Prinn, R.G., 2010. An analysis of the carbon balance of the Arctic Basin from 1997 to 2006. Tellus B 62. doi:10.3402/tellusb.v62i5.16587

• Kimball, J., Jones, L., Zhang, K., Heinsch, F., McDonald, K., Oechel, W., 2009. A Satellite Approach to Estimate Land-Atmosphere CO2 Exchange for Boreal and Arctic Biomes Using MODIS and AMSR-E. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING 47, 569–587.

• Mbufong, H.N., Lund, M., Aurela, M., Christensen, T.R., Eugster, W., Friborg, T., Hansen, B.U., Humphreys, E.R., Jackowicz-Korczynski, M., Kutzbach, L., Lafleur, P.M., Oechel, W.C., Parmentier, F.J.W., Rasse, D.P., Rocha, A.V., Sachs, T., van der Molen, M.K., Tamstorf, M.P. 2014. Assessing the spatial variability in peak season CO2 exchange characteristics across the Arctic tundra using a light response curve parameterization. Biogeosciences 11, 4897-4912.

• Mbufong, H. N., 2015. Drivers of sesasonality in Arctic carbon dioxide fluxes. PhD thesis. Arhus University, Department of Bioscience, Denmark. 144 pp.

• Watts, J.D., Kimball, J.S., Parmentier, F.J.W., Sachs, T., Rinne, J., Zona, D., Oechel, W., Tagesson, T., Jackowicz-Korczyński, M., Aurela, M., 2014. A satellite data driven biophysical modeling approach for estimating northern peatland and tundra CO2 and CH4 fluxes. Biogeosciences 11, 1961–1980. doi:10.5194/bg-11-1961-2014

• Stoy, P.C., Williams, M., Evans, J.G., Prieto-Blanco, A., Disney, M., Hill, T.C., Ward, H.C., Wade, T.J., Street, L.E., 2013. Upscaling Tundra CO2 Exchange from Chamber to Eddy Covariance Tower. Arctic, Antarctic, and Alpine Research 45, 275–284. doi:10.1657/1938-4246-45.2.275

• Land Processes Distributed Active Archive Center (LP DAAC), 2000. Land Cover Type Yearly L3 Global 500 m SIN Grid. Version 051. NASA EOSDIS Land Processes DAAC, USGS Earth Resources Observation and Science (EROS) Center, Sioux Falls, South Dakota (https://lpdaac.usgs.gov), accessed 04/11, 2015, at https://lpdaac.usgs.gov/dataset_discovery/modis/modis_products_table/mcd12q1.

• Van Wijk, M. T. and Williams, M., 2005. Optical Instruments for Measuring Leaf Area Index in Low Vegetation: Application in Arctic Ecosystems. Ecological Applications, 15: 1462–1470. doi:10.1890/03-5354y.