Forests for People

[

] 289

Time-space consistent global PALSAR data are

acquired using Fine Beam Dual (FBD) polarization

from June to September and Fine Beam Single (FBS)

polarization from November to January. ScanSAR

is used frequently for registered areas, where the

reference observation rule is defined in the Basic

Observation Scenario. Frequent ScanSAR observa-

tion becomes especially important to the process of

monitoring seasonal and annual forest changes in

spite of lower resolution. ScanSAR observation was

directed at limited areas of the Amazon, Southeast

Asia, Central Africa and Siberia for deforestation and

wetland monitoring.

SAR is an oblique observation system and the geome-

try and radiometry are easily distorted by terrain height

and spatial distribution. Correction of these distortions

using a global digital elevation model (SRTM-DEM)

is essential to the generation of accurate high-level

products (ortho-products) and for reliable forest clas-

sification. Under this project, we have produced global

10-m resolution SAR mosaics and forest/non-forest

maps. The resolution of the data is 10-m and it is possi-

ble to track the decadal or annual change of the forest

status in that geospatial scale.

Frequent ScanSAR observation over Brazil is one

representative operational example using PALSAR.

Since ScanSAR has a wide swath of 350 km and

the equivalent revisit time is shortened to nine

days, frequent ScanSAR imaging of the Amazon has

contributed to decreasing its deforestation. ALOS data

acquired over the Amazon was recorded and sent to

the Earth Observation Research Center of JAXA in two

days via the Data Relay and Tracking Satellite. The

data was processed using high-level processing and

distributed to the Brazilian Institute of Environment

and Renewable Natural Resources (IBAMA) via FTP

transmission. IBAMA compared the ‘before’ and ‘after’

The final report meeting concluded that the most effective and

important applications for research using L-band SAR would be

forest monitoring and interferometry.

Based on these scientific episodes, JAXA initiated two consecu-

tive forest research projects, the Global Rainforest Mapping Project

(GRFM: 1995-1998) and the Global Boreal Forest Mapping Project

(GBFM: 1998-2002), collaborating with approximately 10 interna-

tional forestry researchers. These two projects targeted observations

of the pan-tropical region and the boreal forest regions. Although

JERS-1 carried a data recorder, limitations in data recording and

linking with ground stations limited the collection of pan-tropical

and boreal forest region coverage during its mission life. Even

with its limited processing capability and the unavailability of a

digital elevation model, however, JERS-1 succeeded in creating the

world’s first 100 metres per pixel (100-m) resolution SAR mosaics

for the Amazon, Southeast Asia, Central Africa, Central America

and Siberia. These products were freely distributed throughout

JAXA and the data were used for forest monitoring in early 2000.

As an example of the product the occurance of fish-borne defor-

estation can be clearly identified using L-band HH polarization

SAR images.

From these scientific results, the L-band SAR was recognized as

a very effective tool to monitor the forest status (deforestation and

land cover changes) and developments occurring over time.

ALOS/PALSAR: a view below the canopy

L-band SAR is more effective than higher frequency SAR for forest

monitoring because the electric signal partially penetrates the forest

canopy and reaches the forest trunks, which contain large amounts

of biomass, and because the clear-cut regions can easily be seen as

they appear darker than the natural forest regions. This is because

the surface roughness is relatively smaller than the wavelength, a

phenomenon known as the Kirchhoff condition.

Based on experience gained from JERS-1 and its science project,

we established larger scale international collaborative forest research

projects such as the Kyoto & Carbon Initiative in 2002. This inter-

national programme is supported by 19 forest institutes.

Global forest/non-forest dataset for 2009