Modelling For Prediction of Global Deforestation Based On Thebpopulation Growth

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

ISPRS Journal of Photogrammetry & Remote Sensing 54 Ž1999.

317–324

Modelling for prediction of global deforestation based on the


growth of human population
Krishna Pahari ) , Shunji Murai
Institute of Industrial Science, UniÕersity of Tokyo, 4-6-1, Komaba, Meguro-ku, Tokyo, Japan

Received 18 June 1998; accepted 16 February 1999

Abstract

Deforestation due to ever-increasing activities of the growing human population has been an issue of major concern for
the global environment. It has been especially serious in the last several decades in the developing countries. A
population-deforestation model has been developed by the authors to relate the population density with the cumulative forest
loss, which is defined and computed as the total forest loss until 1990 since prior to human civilisation. NOAA-AVHRR-based
land cover map and the FAO forest statistics have been used for 1990 land cover. A simulated land cover map, based on
climatic data, is used for computing the natural land cover before the human impacts. With the 1990 land cover map as base
and using the projected population growth, predictions are then made for deforestation until 2025 and 2050 in both spatial
and statistical forms. q 1999 Elsevier Science B.V. All rights reserved.

Keywords: forest loss; remote sensing; population density; population-deforestation model; global land cover change; AVHRR

1. Introduction is becoming increasingly important to make predic-


tions about the state of forest in the future under
Global land cover change, particularly from forest different scenarios to suggest appropriate policy
to other land cover types due to increased human measures.
activity, is one of the most important issues in global Even though significant progress has been made
change research. It has been especially remarkable in in global change research in recent years, the lack of
the last few decades, which witnessed an increasing a reliable spatial dataset on deforestation continues
rate of deforestation due to pressure caused by the to be a major obstacle for modelling global change
population growth. Since forest is so vital for the ŽMurai, 1995.. However, it is still possible to analyse
sustenance of the ecosystem to which we belong, it the trends of global environment, including defor-
estation with the existing satellites, meteorological
and socio-economic data. In this study, deforestation
is addressed relating it to population density and
)
Corresponding author. Fax: q81-3-5452-6408; E-mail: predictions are made about future deforestation, based
pahari@skl.iis.u-tokyo.ac.jp on a projected population growth.

0924-2716r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 4 - 2 7 1 6 Ž 9 9 . 0 0 0 3 2 - 5
318 K. Pahari, S. Murair ISPRS Journal of Photogrammetry & Remote Sensing 54 (1999) 317–324

2. Actual and potential natural land cover

It is now possible to monitor the global land cover


with satellite data, particularly NOAA Global Vege-
tation Index ŽGVI. data. Even though developing an
accurate global land cover map is still an ongoing
process by several researchers and projects, global
land cover maps are now available and provide a
good global overview. Fig. 3 presents an updated
version of the global land cover map from Murai and
Fig. 1. Historical trend of world population and projection Žsource:
UNPD, 1994..
Honda Ž1991. revised using the AARS 4-min grid
dataset on global land cover ŽAARS, 1997..
Since land cover change has been occurring over
a long time and satellite data for global monitoring
There are several factors that seem to be related to have been available only for the last 15 years, a
deforestation, namely population, GNP, government potential natural land cover map has been developed
policy, land ownership, etc. However, it has been to simulate the land cover map prior to human
found by the authors in the study that population is activities, based on climatic data. Fig. 4 shows the
the most significant factor in global deforestation. potential natural land cover map, based on De Mar-
Human population has grown significantly in the tonne’s aridity index ŽAI. and the classification cate-
last few centuries and has been especially alarming gories introduced in Table 1 Žbased on modified
over the last several decades. Fig. 1 shows the criteria by Murai and Honda, 1991..
historical trends of world population and future pro- The AI is defined as:
jections, based on UN medium variant population
scenario ŽUNPD, 1994.. Obviously, this projection, Pi
Ž AI . i s ,
if correct, is bound to have a tremendous impact on Ti q 10
earth resources, including forest.
Fig. 2 shows the general framework of methodol- where Pi is the total annual precipitation Žin mm.
ogy followed in this study. Details of the various observed in cell i and Ti Žin degrees Celsius. is the
steps are discussed in the following sections. annual sum of monthly mean temperatures of those

Fig. 2. Framework of the methodology adopted to predict future global deforestation.


K. Pahari, S. Murair ISPRS Journal of Photogrammetry & Remote Sensing 54 (1999) 317–324 319

Fig. 3. Global land cover map for 1990 from NOAA GVI data, based on Murai and Honda Ž1991. and AARS Ž1997..

months with monthly mean temperature greater than resolution. for the last 30 years were obtained from
0, divided by 12, for cell i. the University of Tokyo.
Global rainfall and monthly mean temperature Based on the above analysis, Table 2 shows the
data Žinterpolated into a surface grid of 10 min results of potential natural land cover and actual land

Fig. 4. Potential natural land cover map, based on climatic data.


320 K. Pahari, S. Murair ISPRS Journal of Photogrammetry & Remote Sensing 54 (1999) 317–324

Table 1 Table 3
Aridity index ŽAI. and land cover types ŽMurai and Honda, 1991. Cumulative forest loss in selected countries Žforest cover in 1990,
Land Cover Types AI based on FAO, 1997.

Desert F5 Country Potential Forest cover Cumulative forest


Semi-desert 5- AI F10 forest in 1990 loss Ž%. Žcolumn
Ž%. Ž%. 2ycolumn 3r
Grassland 10 - AI F 20
Forest ) 20 column 2.
Brazil 97.54 66.68 31.64
Peru 91.95 53.63 41.67
Bolivia 92.39 47.22 48.89
cover Žfor 1990. on a global level. The table shows Ghana 100.00 42.23 57.77
Cameroon 97.88 43.50 55.56
that humans deforested 15.3% of the potential forest
Zimbabwe 74.84 23.16 69.05
area and increased the potential area covered by Bangladesh 100.00 8.10 91.90
deserts and semi-deserts by about 14.8%. Thailand 99.53 25.99 73.89
Malaysia 97.27 53.18 45.33
India 82.28 21.85 73.44
Nepal 83.81 37.25 55.55
3. Population-deforestation model
Former USSR 41.87 37.96 9.34
France 99.28 25.87 73.94
Out of several correlation analyses studied by the UK 98.98 9.63 90.27
authors Žsuch as GNP and deforestation, several
combinations of population and deforestation., it was
found that the correlation between the logarithm of
population density and the cumulative forest loss
In this study, the cumulative forest loss for each
computed from potential natural land cover and ac-
country was computed as the total forest loss ob-
tual land cover was most significant. More specifi-
served from the potential land cover map ŽFig. 4.
cally, the correlation of GNP per capita with forest
and the current forest cover for 1990, based on FAO
loss was very low in all tested areas.
forest statistics.
Due to the unavailability of detailed historical
A regression analysis was conducted using the
data about the global forests over a fairly long period
logarithm of population density as independent vari-
to conduct trend analysis and to make projections,
able and cumulative forest loss as dependent vari-
cross-country data obtained by grouping countries
able. The countries were grouped into regions, based
with similar ecoclimatic zones and similar stages of
on similar ecoclimatic zones and level of socio-eco-
economic development have been used to develop
nomic development.
the model introduced in this paper.
Table 3 presents the cumulative forest loss for
Although Fig. 3 provides a good global overview
selected countries. Table 4 gives a summary of the
of forest and other land covers and fits globally with
correlation index between natural logarithm of popu-
the FAO forest statistics, the authors concluded that
FAO’s statistics led to more reliable estimates of the
forest cover for individual countries.
Table 4
Correlation between population density and cumulative forest loss
Table 2 for different regions
Percentage of potential and actual Ž1990. global land cover Region Regression function R2
Land cover Potential Actual Change from
Tropical Asia 16.042 lnŽ x .y19.56 0.638
type area area original
Tropical Africa 15.206 lnŽ x .q7.8446 0.717
Forest 48.46 33.20 y15.26 Sahelian Africa 16.872 lnŽ x .q12.305 0.638
Grassland 34.27 34.73 q0.46 Tropical Latin America 16.896 lnŽ x .y7.020 0.672
Semi-desert 8.36 15.79 q7.43 Central America and Mexico 21.637 lnŽ x .y29.643 0.824
Desert 8.91 16.28 q7.37 Europe 14.719 lnŽ x .q0.728 0.523
K. Pahari, S. Murair ISPRS Journal of Photogrammetry & Remote Sensing 54 (1999) 317–324 321

lation density and cumulative forest loss for various density Žlogarithm. and cumulative Žor total. forest
regions. Fig. 5 shows the scatterplots for population loss for various regions.

Fig. 5. Scatterplot between population density Žpersonsrkm2 , logarithm. and cumulative forest loss Ž%. for Ža. Tropical Asia; Žb. Tropical
Africa; Žc. Sahelian Africa; Žd. Tropical Latin America; Že. Central America and Mexico; and Žf. Europe.
322 K. Pahari, S. Murair ISPRS Journal of Photogrammetry & Remote Sensing 54 (1999) 317–324

4. Predictions for deforestation

Having established the population-deforestation


model presented above, predictions have been made
for the future state of deforestation. Deforestation for
years 2025 and 2050 was estimated for each country
by using the population-deforestation model and the
projected population density, based on UN medium
variant long-range population projections ŽUN,
1997., and then the annual deforestation rates d i and
the forest loss L i since 1990 were calculated by
using the 1990 forest cover and the predicted forest
covers for 2025 and 2050 as:
LŽ1990 – 2025.

s 100 = Ž forest 1990 y forest 2025 . rforest 1990,


Fig. 6. Ža. Historical trends of population density Žpersonsrkm2 .
and forest cover Ž% of total land. in Thailand Žforest cover, and
mainly based on Mather, 1990 and population density, based on
Statistical Year Books of Thailand.. Žb. Analysis of historical dŽ1990 – 2025.
trends of population density and deforestation in Thailand Ž1949–
1r35
1991.. s 100 =  1 y Ž forest 2025rforest 1990 . 4,
and similarly for 2050. Table 5 shows the results of
such predictions for different regions.
In order to see that such linkages between popula- It can be seen from Table 5 that the deforestation
tion density and forest loss hold true for time series scenario is most serious for Tropical Africa, fol-
analysis for a given country, this method has been lowed by Sahelian Africa, central America, Tropical
tested using time series data of population density Asia and Tropical Latin America. The African region
and forest cover of Thailand. Fig. 6a and b show the is also the area where the population is projected to
results of such analysis for Thailand linking popula- increase most rapidly in the coming years.
tion density and the total forest loss and it shows that The predicted deforestation map for any year, say
there indeed exists a high correlation Ž R 2 s 0.9736. 2025, is then simulated using the following steps.
between logarithm of population density and cumula- Ž1. The land cover map for 1990, as presented in
tive forest loss. Fig. 3, is used as the starting point.

Table 5
Scenario of deforestation from 1990 to 2025 and from 2025 to 2050
Region Forest coverage 1990 Ž%. Predicted annual deforestation rate Ž%. Forest loss since 1990 Ž%.
1990–2025 2025–2050 2025 2050
Tropical Asia 34.81 0.55 0.30 17.44 23.39
Tropical Africa 36.94 1.15 0.68 33.27 43.67
Sahelian Africa 13.52 1.00 0.43 29.62 36.87
Tropical Latin America 61.24 0.37 0.29 12.03 18.29
Central America 31.58 0.65 0.39 20.50 27.87
Europe 36.78 0.01 0.00 0.26 0.26
World 33.20 0.24 0.11 8.13 10.54
K. Pahari, S. Murair ISPRS Journal of Photogrammetry & Remote Sensing 54 (1999) 317–324 323

Fig. 7. Map showing prediction of deforestation from 1990 to 2025.

Ž2. Based on the gridded population density map are signs that the population growth is now actually
of the world for 1994 ŽCIESIN, 1996., and using the slowing in many parts of the world, even though it is
UN medium variant population projections for each still going to be very significant. One exception to
country, a map is prepared for the predicted popula- this is Africa, where the population growth is still
tion density for the year under consideration Žhere, picking up and is projected to slow down only after
2025 and 2050.. the next few decades. That is why the deforestation
Ž3. From the map of 1990 land cover, the cells rate in Tropical Africa is likely to be most serious in
with the highest population density Žfrom step 2. for the next several decades.
each country are then assigned a new land cover While there is some hope in the sense that the rate
class from forest to non-forest such that the total of deforestation is expected to slow down in most
number of cells so converted is equal to the total part of the world, however, if we consider the avail-
predicted forest loss for that country. ability of forest resources for the increasing popula-
Fig. 7 displays the predicted map of deforestation tion, the situation looks very serious. Table 6 shows
for the period 1990–2025, generated by using the
above procedure. Table 6
Projected forest area per capita
Region Forest area per capita Žha.
5. Discussion of results 1990 2025 2050
Tropical Asia 0.19 0.09 0.08
If we compare the predicted rates of deforestation Tropical Africa 1.35 0.36 0.22
in the next five decades, it appears that deforestation Tropical South America 3.46 1.94 1.56
is going to be a significant problem especially in the Central AmericarMexico 0.58 0.29 0.22
developing countries of the tropical region. How- Tropical average 0.72 0.33 0.25
Sahelian Africa 0.74 0.20 0.13
ever, the speed of deforestation is likely to be less Europe 1.24 1.24 1.24
compared to the peak period of 1980s. This con- World average 0.82 0.48 0.41
forms with the trends of population growth, as there
324 K. Pahari, S. Murair ISPRS Journal of Photogrammetry & Remote Sensing 54 (1999) 317–324

the projected forest area per capita for different able and possibly incorporating other factors such as
regions until 2050. It is a matter of further investiga- GNP into the model, is being considered for future
tions as to what are the requirements of forest area research. Another topic of future works is the estab-
per person for healthy and environmentally sound lishment of some thresholds Žrefraining destruction.
living conditions. However, one thing that is very to minimise the projected tendency of forest loss.
clear is that the availability of forest per person will The authors are also working on other aspects of
decrease very drastically, especially in the develop- global environment related to global deforestation,
ing countries, which is likely to pose a major chal- such as carbon fixation, primary productivity, carry-
lenge for human well-being. ing capacity, etc.

6. Conclusions
References
The authors have established a model for linking
population density with the deforestation, to predict
the future state of deforestation. It is seen that even AARS, 1997. AARS Global 4-Minute Land Cover Data Set, Land
Cover Working Group of Asian Association on Remote Sens-
though the rate of deforestation is somewhat decreas- ing. CEReS, Chiba University, Japan.
ing, deforestation will continue to be a significant CIESIN, 1996. Gridded Population Density of the World. Consor-
problem in the next several decades, especially in the tium for International Earth Science Information Network, MI,
developing countries of the tropical region. Africa is USA.
likely to have the most rapid deforestation followed FAO, 1997. State of the World’s Forests 1997, Food and Agricul-
ture Organisation of the United Nations, Rome.
by Tropical Central America, Tropical Asia and Mather, A.S., 1990. Global Forest Resources, Belhaven Press,
Tropical Latin America. The situation looks very London, 25 pp.
serious in terms of forest resources available per Murai, S., 1995. Development of global eco-engineering using
capita in developing countries. remote sensing and geographic information systems. In: Mu-
Although forest loss is caused by various factors rai, S. ŽEd.., Towards Global Planning of the Sustainable Use
of the Earth Resources, Proc. of 8th Toyota Conference.
including the market of forestry exploitation, and Elsevier.
thus forest loss in certain regions may not be directly Murai, S., Honda, Y., 1991. World vegetation map from NOAA
related to the population increase, the above analysis GVI data. In: Murai, S. ŽEd... Applications of Remote Sensing
of the global trend as a whole, shows that in general, in Asia and Oceania, Asian Association on Remote Sensing.
UN, 1997. The State of World Population 1997. United Nations
deforestation is highly correlated with the logarithm
Population Fund.
of population density. UNPD, 1994. World Population Growth from Year 0 to Stabiliza-
Further work on global deforestation modelling, tion. Population Division, Department of Economic and Social
using more accurate datasets as they become avail- Information and Policy Analysis, United Nations.

You might also like