Close Window

LC-23 Abstract

Quantifying the Accuracy of MODIS Fire Products and Establishing Their Relationship with Land Cover Dynamics

Jeffrey Thomas Morisette — NASA (US-PI)
Joao Antonio Raposo Pereira — MMA-IBAMA (SA-PI)
Wilfrid Schroeder — University of Maryland (SA-PI)

Why is it important to monitor fires in the Amazon? 

The role of fires in tropical ecosystems has been established (Goldammer,

1990).  Throughout the 80s and

early 90s, constantly increasing human forcing and climate anomalies made

forest fires in the Amazon Basin a major environmental issue. 

Federal efforts have been directed at early detection and management of

uncontrolled wildfires (Pereira et al., 1999) and fires play a role in the

regional carbon cycle (Potter et al., 2001). 

However, not all fires are equal. 

Some fires result in a major conversion of the landscape. 

While in other locations, fires are necessary to maintain the

landscape (Eva and Lambin, 2000).  As

land management and carbon science efforts continue to expand, it will be

important to “filter” fire maps to differentiate between conversion and

maintenance fires.  To develop

such a filter, research is needed on the relationship between land cover and

fire data.  Due to the large

extent of the LBA region and the current use of remote sensing for regional

fire monitoring (Pereira et al., 1999), this research needs to be based on

land cover and fire products derived from satellite data available throughout

the region. 



Existing work on fire and land cover maps indicates

there is a relationship between fire occurrence and land cover dynamics, but

the form of that relationship depends on the initial and subsequent

land cover.  However, there were

areas for improvement in the data quality cited in that work (Eva and Lambin,

2000).  To address these

shortfalls we propose to investigate the relationship between land cover and

fire maps within the context of validation of new MODIS fire products.  This will allow us to leverage off existing equipment and

infrastructure within the region and take advantage of the new capabilities

offered by MODIS.  With this, we

focus on two separate, complementary, components: 





1-    

Quantifying the uncertainly in satellite derived fire and burn scar

products, concentrating primarily on new MODIS fire products and



2-     Establish techniques to distinguish between fires that cause land-cover

change and fires that maintain a state of equilibrium









The first component will provide validated products

to quantify the extent and timing of burning in the Amazon. 

NASA has an interest in having its latest tools and data products

utilized to the furthest extent possible. 

By teaming with the IBAMA, the Brazilian agency responsible for

national fire monitoring, the proposed effort will help integrate new MODIS

fire products into the region and help establish the use of these products

into the operational fire monitoring and managemnent of IBAMA. 

The geolocation accuracy (~50m) and multiple overpasses (4 per day with

Aqua and Terra) imply fire products from MODIS offer enhanced locational

accuracy and the ability to capture diurnal characteristics. 

However, before these products are put into operational use, they will

need to be validated in a rigorous way; where uncertainties in the product are

quantified.  The first component

of our effort will meet this need.





The second component will address the management and

carbon cycle question on the interaction between fire and land cover dynamics. 

Using the validated products from the first component, we will build on

the integrative nature of the LBA program by coupling IBAMA’s fire products

with existing and forthcoming land cover maps developed by the LBA program. 

By coupling the best available regional fire products with the latest

land cover maps the second component will determine fire’s influence in

answering the questions posed in the LBA solicitation.



§        

How are global ecosystems changing? 

(V3)





§        

What changes are occurring in global land cover and land

use, and what are their causes?  (F2)





§        

What are the consequences of land cover and land use

change for the sustainability of ecosystems and economic productivity? 

(C2)







Overall scientific and technical expertise will be

realized by the collaboration between IBAMA and the MODIS fire product team. 

IBAMA is responsible for operational fire monitoring in Brazil. 

To date IBAMA has used AVHRR and GOES data and is now interested in

using MODIS data.  The

investigators on this proposal have extensive experience with MODIS data and

the validation of MODIS products.  We

are also involved with international fire monitoring efforts that will help

place this work within the context of the Global Observation of Forest Cover (GOFC)

program and the Committee on Earth Observing Satellites (CEOS) Land Product

Validation (LPV) subgroup.





The proposed plan includes unique and innovative

methods by coupling coincident MODIS and ASTER data with airborne campaigns. 

The soundness, logic, and practicality of the proposed techniques have

been demonstrated through pilot work in Southern Africa (Justice et al., in

press) and existing research (Eva and Lambin, 2000). 

Leveraging off existing IBAMA, the MODIS-Fire team, GOFC, and CEOS

creates a high likelihood of achieving the objectives. The proposal utilizes

existing equipment, infrastructure, data sets, and analytical tools. 

The proposed work will integrate with the LBA program by providing

validated fire and burn scar maps to the LBA team. 

We will look to existing and proposed work to provide high-resolution

land cover products (namely, work of Skole et al, LBA team “LC-10”); which

will augment the land cover maps available from MODIS.





The training and education component includes annual

workshops in Brazil on the use of remote sensing for fire and burnt area

monitoring and management.  In

addition, the proposal requests funding for in-region graduate student (or

students). 





Close Window