JapanJapanSociety Society ofForestPlanningof Forest Planning

63 ArticLe

Monitoring Deforestation and Forest Degradation

.m the Bago Mountain Area, Myanmar using FCD Mapper

Myat Su Mon'i, Tsuyoshi Kajisa'2, NobuyaMizoue"Z and Shigejiro Ybshida"2

ABSTRACT

Wliile deforestation in the tropics has been well mapped, there is Iittle infbrmation on the extent and changes of forest degradation. IIhis $tudy aimed te investigate teunporal and $patial patterns of both deforestation and forest

degradation in three re$erved forests of the central Bago mountain ranges, Myanmar, which have a long history of

selective logging activities. We used Forest Canopy Density (FCD) Mapper and Landsat images C1989, 1999, 2003,

and 2006) for the investigation, Over 17 years, more than 90% of the total area was kept as forests, deimed as FCD ) 10%, but closed canopy forests (FZ D z 70%) greatly decreased from 98% to 53% of the total area. Medium canopy forests (40% s FCI) < 70%) and open canepy forests (10% f FCD < 40%) increased. Gross forest degradation (ehange to ]ower FCD) was much larger than gross forest irnprovement (change to higher FCD) for all sites and periods, whereas differences in gross deforestation Cforest to non-forest) and gross reforestation (non-forest to forest) were relatively smal1. As a result, a high annual net forest degradatlon rate of 2.5% was observed, although the annual net

deforestation rate was relatively lew at O.2% betw'een 1989 and 2006. 0ur findings on higher forest degradation throw the question about the sustainahility of current harvesting levels of selective logging andlor extraction of non-wood forest preducts and shiding cultivation by lecal communities if no conservation or remedial measures are

taken. This study revealed the importance of monitoring deforestation and forest degradation as well as the

usefulness of the FCD Mapper.

Kdwo"ds:FCD Mappeg Iandsat images,deforestation, forest degradation, reforestation, forest improvement

for planning and sustainable management oi forests (PAwm et

INTRODUCTION al., 2008)

Deforestation is loss of forest cover by conversion of Tropical forests are critical to the balance of economical, forests to other land cover types (FAO,2000, 2005); it can ecological, and environmental factors on our planet because of be relatively easily identified, Howeve4 identifying forest

their high diversity. However, forests in the tropics are degradation, i.e., reduction in capacity of a forest or canopy depleting at an alarming rate, thus leading te forest cover and stocking within the forest to provide goocls and degradation and deforestatio,n that have been threatening services anO, 2000, 2005), is a subtle process (SouzA JR et al,, future sustainability CthiRozacE et al., 1999; RuDEL and RopER, 2003; P,4NTA et al., 2008). Forest degradation also represents

1997; I.AuRANcE, 1999, 2007; PANiA et al., 2008). Therefore, loss of various environmenta1 functions of the forest, thus

identifying forest degradation and deforestatiorr at different being simjlar to deforestation. While deforestatiun has been

spatial and tennporal scales could provide useful information widely emphasized and analyzed in a large number oi studies

'! Corresponding author: Myat Su Mon Faculty of Agriculture, Kyushu University

" Laboratory of Forest Maoagement, Department 812-8581, Laboratory of Forest Management of Forest and Forest Products Sciences, Graduate Kyushu University, 6-10-1 I'Iakozaki, Higashi-ku, School ef Bioresource arid Bioenvironmeiita1 Fukuoka,Japan Sciences, Kyushu University Tbl: +81-92-642-2867, Fax: +81-92-642-2869 Bmai1:[email protected],jp

J R)r.Hann,l5:63-Z2C201W NII-Electronic Library Service JapanJapanSociety Society ofForestPlanningof Forest Planning

64 fldon et al.

ttsjng satellite remote sensing (ARMENil]RAs et al,, 2006; IELE the tota] land area, in mainlancl Southeast Asia (FAO, 2005). and JosHI, 2008; PtFyRAIokuD, 2003), the extent and rate of forest These forests are important sources ef biodiversity Myanmar degradation have not been well documented (SiJNDEMjN and will faee serious problems in the very near future becau$e of REsosuDARMo, 1996; SouzA et al., 2003; JosHI et al., 2006; PAtwm its high deforestation rate, i.e., 1,3% per year for 1990-2000 and et al,, 2008). 1.4% per year for 2eOe-2oo5 (FAO, 2005). Forest cover

Remotely sensed estimation o'f forest canopy density assessment in Myanmar has been mainly based on combined

(FCD) has been used to a$sess the level of forest degradation use of RS, geographical information systems (GIS), and

et al,, 1986; PklNcE, 1987; kNGRosE and ManIEsoN, CIt[wARI'Foimu] ground forestinventoryon natiollal and regional scales. 1986; and CAsEy, 1988; lg.EINN, 2001; Hiu)i et at., 2004; LEIMGRuBER et al. (2005) performed estirnation of forest cover Josm et al,, 2006; PANTA 2008), Measuring the extent of and focused en deforestation assessrnent throughout the

forest degradation is much more difficult than measuring country Howeveg there is dearth of detail and updated deforestation CDEFIREs et al., 2007). FCD Mapper (RIKIMARu,as$essment on forest degradation, which makes it difficult for 1996) i$ one of the rnetheds used to quantify FCD as a resource p]aTmers and decision-makers to implement effective

continuous variable from O to 100% UosHT et al,, 2e06), [[he conservation strategies for the remaining forest resources. results of FCD mapping give the stratification of iorest cover [Ihis study aims to evaluate spatial and temporal patterns of

and therefere otitputs is reliable to assess both forest forest degr'adatien and deforestation, using the FCD Mapper

degradation and deforestation. The main advantage of using and multi-temporal Landsat irnages in three reserved forests ef

FCD Mapper is to eliminate the need to undertake time the central Bago mountain ranges, Myanmar,

consuming of field measurement of FCD using densitometer

or hemispherical photograph (RIKJmxu et al., 2002; MATERIALS AND ME[VHODS CHANDRAsHEKHAR et al., 2005; BrwEs, 2007; AzrzI et al., 2e08),

which is necessary in other remote sensing (RS) technologies, Study Site

such as artificial neural network, multiple llnear regressien

and maximurn 1ikelihood classfication method. Altheugh FCD The study site was the central region of the Bago Mupper has been applied in dffferent forest types in tropical mountain ranges consisting of three reserved forests (Illls), regions, very few studies have attempted te monitor changes namely Khapaung, Middle Nawin, and South Nawin RFs,

'[he in forest degradation status using FCD Mapper (PANTA, 2003; covering an area ef 1,074km2, Bago mountain ranges Roy, 2004I JAMNIy3AI), 2004) . located in the eentral part of lower Myamnar (Fig. 1), running Myanmar still has the largest forest coveg almost 5096 of from north to south, have renowned natural teak (Tlectona "J."J:,,･f"glM"''i ;v" ?. l c' g './ Legendf'] t/1 ,N' 'iL'r /

v-f) I Myanmar ,/L,:i!1' i? ･I. -. ,hJ , {//{/i',l///･,/l/, Bago mountain ranges F.k,,yv'ynr;ill'/://i･ Study area =.-F Road

i I [fown iIIS11'i" e FCDmeasuredplots // t ", H 'X.h //･ W"E " ,ts l t tt tt(C) "t... "L .zh,"'t' k. "tst,

[[l},ungeoh ')'. "'X 1,, ,x, .e' ".e #'","k

tif i'""ts'-e"i'.7oaktwil\ d..s'' ntre..--.-X .et't //' j"' .,--i' l 3e ]5- o 30kmXk,} " l f.,.if.･n" e

hfi q'bi-"･.,..,.S lit :

Fig, 1 Location of stucty sites: (a) Khapaung, (b) Middle Nawin, and (cl South Nawin reserved forests and sample plots measured FCD

NII-Electronic Library Service JapanJapanSociety Society ofForestof Forest Planning

A4bnitoring Dojiorestation and Fbrest DagTadation in the BagoMountain Area, JV{yanmar ttsing FCD MtiPPer 65

grandis) bearing forests. Altitude ratiges from sc to 6oom Tl{EiN et ai,, 2007), above sea level, The $tudy areas mainly hacl an castern or a Over the last three decades, commercial forest plantations

western aspect. Occurrence of rainfa11 varies in intensity;have been established in the three RFs. However., some older clepending upon the time of year; it is different between plantations in the Khapaung ru-' have already been exploited

regions, According to some of the nearest meteorolegical because of the construetion of the Khapaung dam, In 2004, the

stations, ayerage annual rainfall was 1,917mrn for eastern parts Ministry of Forestry implemented a special afforestation and 1,O09mm for western parts during 1995 and 2006. Alluvial,project called the Bago Ybma Greening Project with the airn of red-brown forest. soil and lateritic soil are comrnon in the conserving and protecting the remaining forest resources. three RFs. Forest type occurred in the study area was commonly mixed deciduous forests, which are by iar the Procedures of FCD Mapper

most important forests in Myanmar. Under mixed deciduous '[he forest types, moist upper mixed deciduous forest (MUMD) FCD Mappeg a semi-expert system was developed by and dry upper mixed deciduous forest (DUMD) are widely the International Tropical Timber Organization and Japan distributed depending upon topography of the area. MUMD is Overseas Forestry Consultants Association alld it produces characterized by the presence of bamboos, Bamb"sa FCD with the range ef O to 100% (kKipsiARu, 1996; kKiMARu and 'lhis Polymompha and dmhatostacdyuTn Pergraciie. forest type Miyrvn4KE, 199n. Ihis comprises biophysical phenomenon occurs on well-grained slopes and goed quality soil, These modeling and analysis, utilizing data derived from four indices: forests contain the finest teak (Z grandis), usually associated I. vegetation index (VJ), selected from the normalized with pyjnkado oolia aylocampa) and DiPterocarPus species. difference vegetation index (NDVI), advanced DUMD is characterized by the presence of bamboo vegetation index CAVD or advaiiced normalized Dendrocalamus strictus and occurs on ridge tops and hot vegetation index (ANVI), aspects. Bambzasa tulda is also found in stiff $oil type of DUMD II, bare soil index (BSD, deRMoDE, 19ou), Characteristic tree species of DUMD are III, shadow index (SI), and very similar to that ef MUMD. Economically important IVI thermal index ffI), species such as teak (T grandis), pyinkado (X xylocampa), These four indices are calculated as new irnages from the raw padauk (Pterocamp"s macrocamp"s), tauklryan (1lerminaliaLanclsat bands 1-7. Based on the indices, FCD Mapper tomentosa), thitya (Slaorea obioprgij?)lia), and ingyin (Flantacmecalculates a vegetation density (VD) which includes grassland siamensts) are growing in two forest types of the st.udy area. and forest but excludes bare soil. Grass land is then separated Ihree RFs have been systematically managed under the from forest using a scaled shadow inclex (SSI) and finally FCD selective logging system for teak and other hardwoods, known is calculated for each pixel of forested land, FCD Mapper as the Myanmar Setection System (MSS). Under the MSS, a produces these indices and integrates them into an FCD as an felling series is divided into 30 blocks of approxiinately equal index ranging from O to 100 for each pixel of the final FCD yield capacit}L Selective logging is carried eut in one of these image. The relevant formula and algorithrns used in FCD

blocks every year and the whole forest is therefore worked Mapper are shown in TtLble 1. over in the fe11ing cycle of 30 years MAH, 2004). Since forestry [[he underlying principle for each of the above four main sector plays an important role in socio-economic development indiees is that the VI has a positive relationship with the of the nation, increasing populatioll andi growillg demand for quantity ef vegetatioll and therefore VI increases from timber, the study area was under shorter fe11ing cycle aess grassland to forest. If there is more tree vegetation, there is

than 30 years felling cycle) and subjected to repeated logging more shadow and SI also increases as the forest density

activittes (BRuNNER et al., 1998; FoREsT DEItA,RTMENI; 1995a, increases. If thcre is less bare soil or lewer BSI, there wi11 be a

1995b, 2006a, 2006b). Land use history Df shifting cultivation correspon,ding decrease in the TI. TI decreases as the practieed by ]ocal communities with iallow periods over a la vegetation quantity increases, i,e, TI is less inside the canopy year rotation was also observed CIZAKEDA et ai,, 2005) , Qf a forest due to blocking and abserption of the sun's rays and There is only one main road that passes through three because oi the coeling effect of evaporation from leave$, RFs and joins two towns, narnely Okktwin, Taungeo district Detailed methodology of FCD mapping is mentioned by and Paukkhattng of Pyay district (Fig 1). AIthough the road RiKii{ARu (1996), Roy et al, (1997), RiK[MJtRu et al. (2002), and condition is relatively good in Middle Nawin and Seuth Nawin CHANDRAsHEEa{,m (2005) and th,e FCD Mapper ver,2 user RFs, some parts are sti]l under construction in Kiapaung RF guide. Outputs of the FCD Mapper was then exported to unti1 2009. 0thers are logging rDads and tbat are especially ArcGIS 9.2 eESRI Inc, USA) environments and used as a grid used during summer for logging extraction purpose. [Il)e system of 30m × 30m for GIS analysjs.

periphery of the study area was more likely to face forest dep!etion because of its location surrounded by a densely 'fertile populated area, easily access, soil ior agriculture, and rich supplies of timber and bamboo (HTuN and MyiNg 2002;

J Fbr, Plann. J5:63-72(201a) NII-Electronic Library Service JapanJapanSociety Society of ForestPlanningForest Planning

66 Mbn et al,

'Ilib]e 1Formu]a and algorithms usecl to calculate indices in FCDMapper

Indices ITormula and algorithms - NDVIan = CBand 4 Bancl 3) 1 (Band 4+Band 3) =: [Band 4X (256TBand 3> × Band 4-Band 3) +1 ]'M, (Band4-Band 3) >O

ANVI This index is synthesized index fro]n NDVI and ANq by principal component analysis.

= L 5+Band 3) 1+ Band 4))1 Band + BSISITIwoSSIFCD[((Band (Band ({Band5+ 3) (Band 1+Band 4))] × 100+100

=: uBand -Band [(256- Band 1) × (256 2) × <256 3)]'" This index is calibrated from the value of thermal band. 'i]his index ls calculated frorn the fust principal eomponent of VI and BSI. Tliis index is calibrated shaaow index for the forest,ed 1and.

=(VDxSSI+1)iM-1

notes:NDVI: normakzed diflerent vegetation index; l"il:z;advanced vegetation index; ANVI: advanced nonnalized vegetation index; BSI bare soit index; SI: shudow index; Tl: thermal index; VI): vegetation densitySSI: scaled shadow inclex; FCD: forest canopy density va

[[lable 2Images used for FCDassessment

No. Year Acquisition date Satellites

1234 1989199920032006 16-Ol-1989 Landsat TNT

30-12-1999 Landsat emM +

23-Ol-2e0315-el-2006landsat rrIM +'

Inndsat IifM +

Data Sources Image preparation was conducted in ERDAS IMAGINEG9.1 (Leica Geosystems Geospatial Imaging, USA) before FCD Seven bands of (.andsat data sets were taken as inputs for mapplng. the FCD Mappez [Iliis study used 13347 Landsat TM for 1989 and Landsat EI'M+ for 1999, 2003, and 2006 CIlable 2). Assessment of Forest Cover Dynamics "iatrelength and characteristics between Landsat TM and

MM+ are almost the sarlle. JA)cAiABAD and ABAR (2004), Roy Fullowing the forest definition of FAO (FAO, 2000, 2005), (2004) and C}mNDRAsHmcHAR et al., (2005) alse applied Landsat we classfied lancl into non-forest (NF) if FCD < 10% and forest TM and Erl]M+ in FCD mapping and used the outputs to study ii FCD ) 10% based on outputs ol FCD mapping. Based on the change detection of forest cover. URy[J et al. (2008), forests were elassified into tlte foUowing Data acquisition date is an important factor when using three categories for analysis of change detection matrices: the FCD Mappei- (ROY, 2004) because it models shadow, bare- open canepy forests (OCFs; 10% g FCD < 40%), medium soil and thermal indices for canopy cover estimation. Tb avoid canopy fore$ts (MCFs; 40% S FCD < 70%), and closed canopy seasonal variation in forest cover assessment, we selected four forests (CCFs; FCD 2 70%). Tb analyze and to cletect spatial Landsat data sets tliat were acquired during the same season, and temporal patterns of forest clistribution, we compared the It ls eictremely difffcult to obtain images with the same results between three RFs over ditferent periods: 1989-1999, aequisition date within the growing season of the study area, 1999-2003, 2003-2006, and 1989-2006. Table 3 shows the particularly in tropical regions where cloud cover is common definitions and rules applied in estimating net deforestation (MAs, 1999). However, the data sets used in this study were and net forest degradation ef the study area, acquired before leufoff season, and these can provide effective

results in estimating FCD, AccuracyAssessment The Landsat TM image of 1989 was geo-referenced using topographic maps of scale 1:63360, published in 1945 by the Ground data was collected from August to October 2007 Survey Department of Myanmar. Landsat images of 1999, and January to February 2009 Lo coinclde with the growing 2003, and 2006 were geo-referenced using an image-to-irnage season of forest vegetation and acquisition of images. A purely

registration method where the 1989 image was taken as the random sarnpling design was applied to establish sample plots. master imhge and the root mean square error was O,O05 pixels, Precise locations of the sample plots were reeorded using

NII-Electronic Library Service JapanJapanSociety Society ofForestPlanningof Forest Planning

Mbniton'ng Dqforestation and librest Dagrtzdationin the Bago Mb"ntain Area, Adyanmar tfsing FCD imPper 67

Table3 Definitions and rules te calculate deforestation and iorest degradatien

Change types Definition aiid calcu!ation of changes

complete conversion of forest to non-'[orest, Gross deforestation Le,, total number of pjxels changed from CCE MCE arid OCF to NF

re-establishment of forest from non-forest Gross reforestadon CPEI{A[TA and MlvrHER, 2000; NAGENDRA et al., 2006), Le., tota1 number of pixels converted from NF to OCF; MCE and CCF - Netdeforestation gross deforestation gross reforestation

change to lower FCD within a forest

C]ross forest degradationaiiy(SilxENA et al,, 1997; URtru et al,, 2008), i,e.,tota1numberofpixelsconyertedfromCCFtoMCForOCFaswe!1asMCFtoOCF

change to higher FCD within a forest, Gross forest improvementany i,e,, tetal number of pixels conver ted from OCF to MCF or CCF as well as MCF to CCF - Net forest degradationgross forest clegradation gross forest improvement

Annual rate of net netdeforestation 1 × × 100 deforestati,oii (%) total forest areas at initial year of assessment assessmentperiods

Annuat rate of net 1 etfotddti x 100 forest degradation (%)v total forest areas at initiul year of assessment assessmentperiodsx

notes: NF:non-forest; OCF: open eanopy forest; MCF: medium eanopy forest; CCF: closed canopy forest

Tab]e 4 Error matrix for measured and estimated classes of forest cover

Measured NF OCF MCF CCF TbtalUser's accuracy (Ve 4ooo4 2to2216 o722938 oo1110211361735113171 67596390 EstimatedNFOCFMCFCCFTota1

Moducer:s accuracy (%) 10081O,62 63 58 90 Overall accuracy <%) Kappacoefficient

notes: NF: non-forest;OCF: open canopy forest; MCF: medium canopy for¢ st; CCF: closed canopy forest

¢ 100 GPSMAIpt 60CSx (Garmin Ltd, USA) in a Universal Transverse r -: Mercator coordinate system with common datum of Wl)rld ･nv . Geedetic System 84, Canopy density was measured at five --f.' peints, i,e,, four eorners and the center of eaeh plQts x 80 - (30m t -e- L1!Il11N--e ' `!!g!s: 30m, 171 points) (Fig. 1), using a eonvex densiometer - -.S . (Forestry Suppliers Inc, USA), Mean canopy density of each ...e 8aUinvotifittho . .r,' was calculated aceuracy assessment, mappillg plot for Overal1 60 .･ ee e i1 accuracy was evaluated by scatter plot ana an error matrix ./i.es. .d- e between measured and estimated FCD. ...t - e ･fi･ -"- .- t･ 40 .･ "-- RESUI:TS 1 1 ttt..tttt tes I . t...-..: I･ .t.t.. Aceuracy ef FCD Mapping : tt.tt y=O.8319x+7.6974 20 =: l ttt.. R2 e.8125 i .nf' shows relationship measured 1tt Fig.2 the betweenFCD /tt . using a cenvex densiometer in the field and FCD estimated r' 1oi---- ...... -.-...... -ll-...... 1...... /t----tL-] using FCD Mapper with re=O.81. The error matrix of

6o so toe rneasured and estimated classes of forest cover CTbble 4) o 2o 4o shows the accuracy of the FCD Mapper in elasstlying forest MeasuredFCD%

cover, i,e,, 138 of 171 observations were correctly classified

with an overall accuracy of 81% and a kappa eoefficient of O.62, Fig, 2 Scatter plot of measured and estimated FCD%

J Fbr.Aann.15:63-MceOJov NII-Electronic Library Service JapanJapanSociety Society of Forest Planning

68 Mon et al,

Forest Cover Change preminent in the Middle and South NaMrin RFs eompared te

the Khapaung RE In the former two RFs, CCF largely

Fig. 3 shows spatial distribution of four classes of forest diecreased to around 30% between 2003 and 20e6, whereas 'fable cover in each yeag and 5 indicates percentage$ ot forest MCF increased from 1% to approximately 60% between 1989 cover classes for three RFs. In 1989, almost all land 08%) of and 2006, On the other hand, Khapaung maintained a the three RFs was covered with CCF; whereas NF comprisecl relatively high percentage ef CCF (62%) in 2006, whereas the only 1% of the total area. Forest cover changes were more percentage of MCF was 30%, Changes in NF were relative]y

NW"E

s

gIur

Legend[

] reserved for igIEi$ non-forest

Fig.3 Forestcoverclassifiedby FCDof threeRFs for 1989, 1999, 2003, and 2006

Table5Percentage of four forestcover types in each year

Study sites year NF (%) OCF <%> MCF (%) CCF (%) Khapaung 198919992eo320061,O1,82,33,8 O,32.83,83.7 O,62.24.53e.598.193.189.462.1

Middle Nawin 1989199920032006o.s1,61,47.2 O.47,59,O8.0 l.O10.914,858.597,880.074.826.2

Soutli Natvin 1989199920032eo61.12.31.76.e O,57.510,77,81.29.820,255.6 97.280,567,330,7

'ft'ial 19891999200320061.01.92.14,6 O,34,15.54.9 O.74,58.138.098,e89.584.352.6

notes: NF: nonlorest; OCF: open canopyforest;MCF:'''mediu-m canopy foregt;CCF: closed canopy fere-s't

NII-Electronic Library Service JapanJapanSociety Society ofForestof Forest Planning

Mbniton'ng Dnjiomstation and Fbrest Dagraclation in the Bago imtrntaitt Area, toanmar using FCD immper 69

smal1 in al1 three RFs with a percentage of much less than 10% reforestation in most cases, thus resulting in net deforestation, in 2006, Among different periods, changes during the latest except for the period 1999-2003 in the Middle and South period, i.e,, from 2003 to 2006, were the most prominent, Nawin RFs, It should be noted that gross degradation was Since each forest cover map has four c]asses, sixteen much larger than gross deforestation in all periods and RFs.

different classes were involved in change detection matrix. Tables 7 and 8 show annual net deforestation and forest 'Ihe [[hey were categorized into five groups: ne change, gross degradation, respectively highest deforestation rate of forest degradation, gross deforestation, gross reforestation, 1.96% occurred in the Middle Nawin M-' during the period and gross forest irnprovement. Table 6 shows percentages of 2003-2006, followed by that of 1.45% in the South Nawin RE As five categuries of forest cover change for each RF and period. expected from [fable 6, there were two cases of negative net

Among the three RFs, the Khapaung RF had a 1arger number deforestatien rates, i.e., net deforestation in the 1999-2003

of no change areas and less gross forest degradation. For all periods. Forest degradation rates were always much higher RFs and periods, gross degradation was always much higher than deforestation rates; the highest forest degradation rates than gross forest improvement thus resulting in net forest were found in the latest period of 2003-2006, with a maximum degradation; differences between them were larger in later forest degradation rate of 16%,

periods. Gross deforestation was also rarger than gross

Table 6Percentage of forest cover change types for four periods

Gross forest Gross Grossforest Gross Study sites PeriodsNo change (%) degradation er) deforestation (%) irnprovement (wn reforestation {9D Mlapaung 1989-1999 92.586,865,162.74.76.727,733.7 1,61.83,O3.1 O.53.43.6O.2 O.7L3O.6O.3 1999-2003 2003-2006 1989-2e06

Middle Nawin 1989-1999 79.Z74.638,127.017.914.449.065.71.5Ll7,76.8 O.88,64,7O.2 O.6L3O,5o.?, 1999-2003 2003-2oo6 1989-2006

South Nawin 1989-1999 79.6os.548.731.416,621.137.862.221.25.95.4 e,g7.47.2O.5 O,9L8O.4O.5 1999-2003 2003-2oo6 1989-2{}06 Tota1 1989-1999 88,882.558,953.28,39.732.242.2 1.61,64.14.0 O.64,742e,2 O.7L5O.6O,4 199Y20e3 2oo3-2006 1989-2006

Tlable 7Annual net deEorestation rate (%) of three reserved forests

Study sites 1989-1999 1999-2003 2003-200G 1989-2006

Mlapaung O,08O.08O.12 O.12-O.05-O.14 O.501.961.45 O.17O.38O,29

Middle Nawin

South Nawin Tbta1 O.09 O.06 O.85 O.21

Table8Annual net forest degradatien rate (℃ of three reserved forests

Study sites 1989-1999 1999-2003 2003-2006 1989-2006 }QiapaungMiddle O.421.721.58 O.821.473.51 8.9616.1011.351.993,883,66

Nawin

South Nuwin Tbtal O.76 1.27 10.35 2,50

J Fbx Hann. 15:63-72(201Q) NII-Electronic Library Service JapanJapanSociety Society of Forest Planning

70 Mbn et al,

-,rt.. 70 ,tu/g

oo -iolo.

50 l-iU4o'i

3e //'61>

20 -Irr Ig. iO 'i,

-,-' o tt'-'&"1'''''g'"tt"'6'""'S""."""as'""'nv.'-'-L""".-'--."""t7"'tt'"Nl'-''tt""di" gggggsgggsssss 8888 H------NN "" esa cy 6- ab th - th di g ab d6 "･ ab th . or aaaaoamaaoeoo s･ s g･ aaaaaaa en am= ¢ oo -- r- "t---" r.1 NNNN k88R

Fig,4 Ik}tal exploited timber volume (1,OOOmS) from three RFs (1990-2007)

Among different periods, forest degradation was very rapid

DISCUSSION dunhg 2003-2006, i.e,, 10% in totai, Changes in the harvesting

volume could be one of the reasons for this. Accorcling te 'Ihe FCD Mapper was designed for foresters and records of timber exploitation over 10 years, exploited timber resource rnanagers to easily collect informatien on forest volume was highest during 2003-2004 CFig. 4). Accordingly. canopy with reducing costs and saving time (kmMARu et al., conservation activities including checking annual allowable

1997). In the other words, FCD Mapper eliminates 1arge cuts for selective logging activities ior each management unit

amount of ground data which are not always easily available, should be implemented based en updated inventory data, and [[his study demonstrated to detect forest cover change, enrichment planting sheuld be conducted for open and deierestation, and forest degradation using FCD Mapper. Ihis degraded forest areas, study achieved an overall mapping accuracy of 81%, which is The Bago mountain ranges have been managed for an acceptable level of FCD mapping in comparison to uther timber production and non-wood forest products (NWFPs) by studies [71% (PANTA, 2003) and 50% CJosHI et al,, 2006) in the Forest Department (FoREsT DERitRTMEN[e 2006a, 2006b), Nepal, 83% gmmAi} and ABAR, 2004) and 84% (AzLzi et ai., Sustainable forest management for )ong-term benefit was 2008) in Iran, and 80% in India (C}{A/NDRAsHEKflAR et al., 2005) ]. attempted in the forestry sector by introductien ot a Performance of the FCD Mapper in forest cover assessment rehabjlitation project, the Bago Ybma Greening PrQject provided satisfactory accuracy, thus showing that it can be (FoREsc DE?AR'fivtENz 2004), In this project, both silvicultural used to assess forest degradation as well as deforestation. treatments in natural forests and socio-economic dovelopment During the 17-year peried, the tiiree RFs maintained a were considered in promotion of conservation activities. high percentage of forest cover (more than 90%) and the Assessment and monitoring of forest cover chmiges using annual net deferestation rate was O.2% in tota1, thus being remote sensing data with FCD Mapper us demonstrated in this

much smaller than the 1.3% or 1.4% values estimated by the study will be usefu1 to evaluate and improve effect.iveness of FAO (2005) and similar to the estimate (O.3%) by LEiMGRm3ER this kind of conservation aetivities. et al. (2005). On the other hand, we fou,nd that a high degree

of forest degradatien occurred in all study sites, with an annual CONCLUSION

net degradation of 2.5% in tota1. The high percentage of CCF in 1989 gradually decreased in all areas, whereas MCF NV℃ found that the FCD Mapper is useful technique for

signMcantly increased. 11iese findings irnply that forest inonitoring and assessing deforestation and forest degradb degractation may be more problematic than deforestation in tion, which are signifieant problems in planning and the study sites. [mplementation of sustainable forest management. Ihe study We found that there were diiferences in deforestation and area still had a large percentage o £ forest area, more than 90% forest degradation arnong different ]ll?s and periods. Among of the total area, over the 17 years cons{dered, Howeveg gross the three RFs, forest degradatiQn was larger in the Middle and forest degradation was much larger than gross forest

South Nawin RFs for al1 periods. As mention in study area, one improvernent, thus resulting in high net forest degraclation,

'IIhese reasen might be because the two RFs are more accessible in especially tn recent years. findings cal1 for conservation terms oi road conditions and distance from the road and activities to improve current land use practices, such as human settlements in compare with Khapaung RF (Fig. 1). commercial legging and shifting cultivation practices in the [fhese RFs are therefore more susceptible to fuelwood study area. Ihis study provides baseline information related to

collection and encroachment by surrounding communities. dynamics ot forest cever. Future studies should be conclucted

NII-Electronic Library Service JapanJapanSociety Society ofForestof Forest Planning

Mbniton'ng Daj?]mstation and Fbrest Dagradation in the Bcago Mbuntain Area, ltlyanmar usiug FCD imPPer 71

to identify factors causing deforestation and forest degrada- HTvN, K. and Miwi; W:, (2002): Tradhional and Current Imncl Use in tion, and find remedial measures for future sustainable forest South Nawin Watershed in Paukkhaung Model Forest, Myanmar. management. Newsletters: News on the Modei Forest Approach to Sustainable Forest Managernent. Food and Agriculture Organizadon of UnitedNations LITERALTURE CITED JAMAm]AD, M. S. and AIIKAR, AA, (2004): Forest Canopy Density Monitodng using Satellite Images. Geo-Imagery Bridging imiEN'i'ERAs, D,, RupAs, G,, RoDruGvEz, N,, SuA, S. and RoMERo, M., Continents XXth ISPRS Congress, fstanbu], 1hrrkey (2006): Patterns and Causes of Deforestadon in the Colombian Josm, C., I.EEtT"L J.D., SKIt}MoRE, AK, DimEN, I,C.Vl and OosfEN, Arnazon, Ecological indicators 6: 353-368 H,V,, (2006): Remotely Sensed Estimation ef Forest Canopy Azlm, Z., NAJA", A. and SoHRABI, H., (2008): Forest Canopy Densily Density: A Comparsion of the Performance of Four Methods. Estimading using Satellite lmages, Tlie International Archives of International Juurnal ef Applied Earth Observation and the }'hotograrnmetry, Remote Sensjng and Spatial Information Geoinformation 8: 8495 Sciences. Vb!. XX)IVII. Part B8. Bejing KERMoDE, D.WD., (1964): Some Aspects oE Silviculture in Burma BancEs, J., (2007): Using FCD Mapper Software and IAndsat Images Forest, Forest Department, Rangoon, Myanrnar to Assess Forest Canopy Density in Landscapes in Australia and KullNN, C., (2001): A Cautionary Note on the Minimurn Crown the Phl]jppines, Annals of Tropical Research 29: g20 Cover Criterion in Ferest Definitions, Can, J. For, Res, 31: 350-356 BRuNNEH, J., TZALBoTl; K, and E!-N, C., (199ee: Legglng Burma's LAvRAr"cE, WE, (1999): Refiections on the Tropical Deforestation Frontier Forests: Resourees and the regime. Wbrld Resource Crisis. Biol. Conserv 91: 109-117 Institue. Washington, DC. ISBN: 1-56973-266-3 LiLvRANcE, l4LE (2007): Forest Destruction in Tropical Asi}L Special C}IAwoRAsHEK}IAR, M.B., SARAN, S,, RAJu, RL.N. and kog PS., (2005): Section: Asian Biodiversity Crises, Current Sciellce 92: 1544-1550 Forest Canopy Density Stratdication: How Relevant is Biophyslcal inEllyaG]cuBER, R, )inLLN, D.S., S/rEIMNGER, M.K., BI{tptNl& J., MuEJ.1.ER, Spectral Response Modeling Approach?, Geocarto International [[: and SoNGER, M,A., Forest Cover Change Patterns in 20: 1521 (2005): Myanrnar 1990-2ooO. Environmenta1 Conservation 32: DAH, S, E,, (20e4): Tk}ak and Forest Management in Myanmar: (Burmu) 35&364Im.}/, Myatifnafs Natural Teak Forests are being Supplemented N. and Josm, PK., (2008): Analyzing Deforestation Rates, Increasingly by Plantations, rrl'O Tropical forest Updated 14: 12- leDEmas, Spatial Forest Cover Changes and Identitwng Critical ftreas of Forest Cover Changes in North-East India 1972-1999, Environ R, AcHARD, E, BRoMN, S,, HERom, M., Mulv]tyltRso, D., Monit Assess. DOI. 10. 10071s 10661-OOge472-6 ScHiA/uADiN(;ER, B, and SouzA Jr, C, D,, (2007): Earth MAs, J.E, (1999): Monitoring IAnd-oover Changes: A Comparison of Observations for Estimating Greenhouse Gas Emissions ftom Change Detection Tlechniques, int J. Remote Sensing 20: 139-152 Deforestation in Developing Countries. Environ. Science and NA[;ENDReL, H., }IARErri'H, S, and GFvLTE, R,, {2006): People Within Prak- Poli¢ y: 385394 forest Villages, Land-cever Change and Landscape Fragmentation EAO., (2coO): Forest resources assessment 2000. Food and 'Fadoba in the Andhari Tiger Reserve, India. Applied Geography Agriculture Organization of United Nations, Reme 26: 96-112 FAO., (2005): Forest resources assessinent 2005. Food and PANTvt, M., Analysis of Forest Canopy Density and Factors Agricu1ture Organization ef United Nations, Rome (20e3): affecting it, using RS and GIS Tleehniques (A Case Study from FoRD, J.R and CAsEy D.J., (1988): Shuttle Radar Mapping with Chitwan I)istrict of Nepal). H]C Master lhesis Diverse Incidence Angles in the Rainforest of Borneo. Int. J. PANTn, M,, }CM, K. and C., 1lemporal Mapping of Remote Sen$ing 9: 927-943 JositE,(2008): Deforestation and Forest Degraclation in Nepal: Application to FoREsr DEiiARTiLiENz (1995a): Forest Management Plan (1996-1997 to Forest Collversation. For, Ecol. and Manage, 256: 1,587-1,595 2005-2006), Pyay District, Bago Division. Myanmar a'ext in PE]iAum, Myanmar) R and MATIIEe R, (2000):1trilinalysisof Deforestation Patterns in the Extractive Reserves of Acre, Arnazonja form FoREsi' DERARTME)q C1995b): Forest Management PIan (1996-1997 Satellite Imagery: A Landseape Ecological Approach. Int. to 2e05-2006), Taungoo Disnict, Bago Division. Myanmar Cl'ext in J. Rernote Sens. 21: 2,5552,570 Myanmar) PruNcE, SD., <1980: Measurement of Cnnopy Interception of Solar FoRE/sc DEpARTMENz (2004): Bago Ybma Greening Project C2004-2005 Radiation by Stands of Trees in Wboded thvanna to 200&20co), Forest Department, Myanmai- CIbxt in Myanmar) sparsely int. Rernote Sensing S: 1747-1766 FoREsT DEEAwi'MENg (20e6a): Forest Management Plan (2006-2007 to (Sudan).J. PirfiiAwNuD, J,R, (20e3): Standardizing the Calculation of the Annual 2015-2016), itsray District, Bago Division. Myanmar CIlext in Rate of Deforestation, For. Ecol. Manage. 117: 593-596 Myanmou) 'llM k-MARL/･, A., (1996): IAMDSfYl' Data Processing Guide for Foiiis'r DEpAwrMENT; (2006b): Forest Management Plan (2006-2007 forest Canopy Density Mapping and Monitoring Model. ITI/O to 20152016), Taungoo District, Bago Divisien. Myanmar CTIext in Myamnar) workshep en Udiization of Remote Sensing in Site Assessment and Planning for Rehabilitation of Logged-over Forest 30- HN)I, E, W[KANTIKiL, KL and SurmITD, I., <2004): Implementation of July August 1, Bangkok, Ihaffand Forest Canopy Medel to Monitor Ferest Fragmentaden in Mt. RI)aNLARu, A. and M]yKrAKE, S., Development of Forest Simpang and Mt. Tilu Nature Reserves, West Java, Indonesia. 3rd (1997): Canopy Density Mupping and Monitoring Model usillg Indices el FIG Regional Cenferenee Jakarta, Indonesia, October 3-7

f Fox Aann. 15:62-72201di NII-Electronic Library Service JapanJapanSociety Society of Forest Planning

72 Mbn et al.

Vegetation, Bare Soil and Shadow. 18t:' Asian Conierence on of Deforestation in Indonesia: towards a Resolutien of the

1temote Sensing, October2a24,Malaysia Arnbiguities. Center for International Forestry Research. kKJMARu, A., Roy, RS. ancl MTyTyiAlcE, S., (2002): Trepical Forest 0ccasional paper no. 9, ISSN 0854-9818 Cover Densaly Mapping, Tropical Ecology 43: 39-47 TAKEI)A, S., SuzuK[, R. and T}IEIN, H.M., (2005): 11iree Years R;NGRosE, S. and MKrHEsoN, W, (198di: Monitoring Desertification Monituring of Shifting Cultivation Fields in Karen Area, the Bago 'raikai in Botswana usiig Landsat MSS Data: with Consideration as to Mountain$, Myamllar. Nihon Sh{nrin Gakkai Koen

the Nature of the Infra-red Paradox. In: Millington, A.C., Mutiso, Ybshishu (CD-ROM) S.K, Binns, JA

Putterns Tropical Destruction:Cross National of Deforestation, Ui"yu, Y/, et al., (2008): Deforestation, Furest Degradation, 1975-1990. Wbrld Dev 25: 53-65 Biodiversity Loss and C02 Ernissions in l

5Q6SuNDERIJN, (Received 18 May 2009) WD. and REsosuDARMo, I.AR, Rates and (1996): Causes (Accepted 5 September 2009)

NII-Electronic Library Service