Kamel Didan , Armando Barreto - University of Arizona · Kamel Didan1, Armando Barreto2 ... (Kamel...

Post on 04-Dec-2018

226 views 0 download

Transcript of Kamel Didan , Armando Barreto - University of Arizona · Kamel Didan1, Armando Barreto2 ... (Kamel...

Kamel Didan1, Armando Barreto2

A Reference Global Vegetation Annual Dynamic Profile

1 VIP Lab., ECE Dept & Institute of the Environment, 2 ABE Dept. The University of Arizona, Tucson, AZ 85721, USA

*didan@email.arizona.edu

Acknowledgements: This work was supported by NASA CA # NNX08AT05A (Kamel Didan, PI)

Qu

alita

tiv

e a

nd

qu

an

tita

tiv

e a

naly

sis

.

An

nu

al

sig

nal

co

nstr

ucti

on

Terra record : 2000 20005 2010

10-year VI Records (2 replicates)

DOY 001

DOY 009

*

*

*

*

*

DOY 193

DOY 199

*

*

*

*

*

DOY 353

DOY 346

Aqua record : 2002 20005 2010 Only perfect data retained based on

the reliability index, then averaged.

10-year Average Dynamic profile

(8-day interval)

Introduction

Vegetation indices are by far one of the most

successful and widely used remote sensing

measures of the land surface vegetation

conditions. And although, vegetation indices

are not explicit biophysical parameters, they are

extensively used as proxies for many canopy

state variables (leaf area index, fraction of

absorbed photosynthetically-active radiation,

chlorophyll content, canopy structure) and

canopy biophysical processes (photosynthesis,

transpiration, net primary production).

Remote sensing based Normalized Difference

Vegetation Index (NDVI) and Enhanced

Vegetation Index (EVI) are the most widely

adopted indices. Combined, NDVI and EVI span

more than three decades of consistent

measurements (from NOAA AVHRR and NASA

MODIS, as well as other sensors) and continue

to contribute significantly to understanding the

Earth system functioning and change in

response to disturbances.

To that end vegetation indices depict a

composite property of the canopy and are

unmatched with their efficiency to capture the

canopy dynamic behavior over space and time.

Thus, a vegetation index is an ideal tool for

effective characterization of ecosystem states

and processes for long term climate change

studies and near real time operation

applications.

Objectives

The aim of this study is to generate a high fidelity

global annual vegetation index data record useful

for global change research and for the analysis of

temporal and spatial land surface vegetation

anomalies. The specific objectives are:

• To analyze the 10-year MODIS Terra and Aqua

records

• Assess these records stability

• Combine these record into a single annual

reference temporal profile

• Assess the usefulness of this reference

record for the study of vegetation dynamic

Data and Methodology

We used all years Terra and Aqua global 1km and CMG MODIS 16-day VI records

• Terra : Feb. 2000-Feb. 2010

• Aqua : May 2002-Feb. 2010

• Because Terra and Aqua VI data streams are highly identical and produced 8 days apart at 16-day composite periods we combined

them into one record

This 10-year record (10 years Terra and 8 years Aqua) was preprocessed using the data reliability index (Didan & Huete, C5 VI product) to eliminate any less than ideal data (cloud, aerosol, shadow, extreme viewing geometry, etc…)

The resulting records were averaged into a one single annual cycle to serve as the reference for anomaly and change analysis

This reference record consisted of an 8-day 46 global VI maps

Conclusions

To accurately study change using Vegetation

Index data, or any other data for that matter, a

precise and reliable reference record is

required. Change could then be measured

against this reference record and accurately

ascribed to disturbances and/or climate drivers.

In this work we designed a methodology to

generate such a reference record using the

MODIS vegetation index 10-year data record.

Both NDVI and EVI reference records were

generated at an 8 day interval and different

spatial resolutions (250m, 1km, and 5.6km).

These records are expected to be particularly

useful for annual and inter annual change

studies, for long term trend analysis, and ready

for integration into phenological,

biogeochemical and global climate models.

Example reference VI profiles

These high fidelity reference profiles could be

used to drive phenology, biogeochemical, and

change detection models.

Aqua/Terra Reference EVI based Annual GPP estimated

following the GPP≈ f(EVI) tower empirical relationship

suggested by Huete et al (2008).

~ ~ ~ ~ ~ ~

~ ~ ~ ~ ~ ~

Terra & Aqua are highly

identical which justifies

combining their products.

0

0.1

0.2

0.3

0.4

0.5

Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Jan-10

Ve

ge

tati

on

Ind

ex

Terra_NDVI_Atacama_Desert_Dark

Terra_EVI_Atacama_Desert_Dark

Terra_NDVI_Sahara_Desert_Bright

Terra_EVI_Sahara_Desert_Bright

Aqua_NDVI_Atacama_Desert_Dark

Aqua_EVI_Atacama_Desert_Dark

Aqua_NDVI_Sahara_Desert_Bright

Aqua_EVI_Sahara_Desert_Bright

Terra & Aqua long term VI

stability over bare soil. After 10

years of operation both sensors

are performing well.

Resulting Terra MODIS 10-year Global NDVI temporal profile

compared to AVHRR GIMMS data record (1981-2003).

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1-Jan 20-Feb 11-Apr 31-May 20-Jul 8-Sep 28-Oct 17-Dec

ND

VI

Date

NDVI_Cascade mountains

NDV_Agriculture

NDVI_Taiga

NDVI_Desert

NDVI_Tapajos

NDVI_Amazon_Forest

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1-Jan 20-Feb 11-Apr 31-May 20-Jul 8-Sep 28-Oct 17-Dec

EV

I

Date

EVI_Cascade mountains

NDV_Agriculture

EVI_Taiga

EVI_Desert

EVI_Tapajos

EVI_Amazon_Forest

Jan. 1st

Jan. 9th

Dec. 19th

Dec. 25th

The difference between AVHRR and MODIS profiles during

the Second half of the year are a result of the differences

In their tropical forest NDVI signal (MODIS is more sensitive

because of the narrower bands).