Living Land

[

] 132

covered latitudes between 86°N and 86°S globally. The laser’s

return pulse was detected by a telescope on-board ICESat.

GLAS recorded the changes in laser energy intensity as a

waveform, and the waveform contained information on the

vertical structure of forests. At flat areas, the height differ-

ence between signal start and signal end corresponded to the

maximum canopy height within the footprint. If a waveform

was bulky at high elevation, the observation was considered

to have been made at a high-biomass forest because it meant

a large reflectance from the canopy level. In this way, many

studies have been conducted to analyse the GLAS waveform

to estimate canopy height and AGB.

We developed a methodology to estimate canopy height and

AGB using GLAS data, and evaluated its performance. A technol-

ogy to accurately measure canopy height or AGB using satellite

data will make it possible to acquire enormous amounts of data

on forest resources without labour-intensive in-situ measure-

ments. The study area was Borneo. The tropical forests in Borneo

have a considerable canopy height and are rich in biomass even

on a global scale, and also rich in biodiversity with many rare

animals and plants. However, the area of forests in Borneo has

been rapidly decreasing in recent years because of forest fires,

oil palm plantation development and so on. Therefore, Borneo

attracts attention from REDD+ implementations.

We collected field-measurement data of canopy height and

AGB in 37 plots which coincided with the GLAS footprints.

We used the data as reference data to examine the relation-

ships between some parameters of GLAS waveform shape

and canopy height or AGB. These relationships could be

used to estimate canopy height and AGB from GLAS data.

The estimation accuracies (root-mean-square errors) were

4 m for canopy height, and 38.7 t/ha for AGB. Among the

field-measurement sites, there was a rich forest where tree

height was 30 m and AGB was 300 t/ha. However, accuracy

remained high even in such a forest.

We applied the estimation models of canopy height and

AGB to the GLAS data in the whole of Borneo. After exclud-

ing data on cloud-covered measurements and non-forested

areas, we obtained 127,862 data points that were valid for the

analysis. Field measurements at 127,862 plots would require

a great deal of labour, so using satellite data is an efficient

method for collecting information on forest resources. The

results of estimating canopy height and AGB from the GLAS

data showed that the average canopy height was 17.3 m, and

the average AGB was 191.8 t/ha over Borneo. Furthermore,

we created AGB histograms for each province to compare

the distribution of forest resources. In Brunei Darussalam,

the forests were found to have high AGB because of the

many forest reserves. The forests in Kalimantan Utara and

Kalimantan Timur, in the eastern part of Borneo, also had

a high AGB. By contrast, in Kalimantan Barat, Kalimantan

Tengah and Kalimantan Selatan, in the southern part of

Borneo, relatively low-AGB forests were found. Finally, in

Sabah and Sarawak, in the Malaysian part of Borneo, forests

with a wide range of AGB were distributed.

Next, we divided the GLAS data into two groups accord-

ing to their observation period, and we estimated the rate

of forest loss between the two periods by identifying GLAS

data for non-forested areas (estimated canopy height < 2

m). As a result, the forest loss rate was 2.4 per cent per

year during three years from 2004 to 2007. The rate in the

Malaysian part of Borneo was 1.4 per cent per year and

A forest biomass map of Borneo (500-m resolution)

The distribution of GLAS footprints in Borneo

Source: CGER

Aboveground biomass

SRTM elevation

iving

and