Above Ground Forest Phytomass Assessment in Southern Gujarat

Above Ground Forest Phytomass Assessment in Southern Gujarat
of 10
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
    R  E   V  I  S  E  D   P  R  O  O  F RESEARCH ARTICLE Above Ground Forest Phytomass Assessmentin Southern Gujarat Prashant Patil  &  Sarnam Singh  &  V. K. Dadhwal Received: 14 September 2009 /Accepted: 4 May 2010 # Indian Society of Remote Sensing 2011 Abstract  Spectral modeling of above ground biomass(AGB) with field data collected in 48 field sitesrepresenting moist deciduous forest in Surat district isreported. Models were generated using LISS-III andMODIS data. The plot-wise field data were aggregatedto MODIS pixel (250 m) using area weightages of forest/vegetation. The study reports that above ground phytomass varied from 6.13 t/ha to 389.166 t/ha whileAGB phytomass estimated using area-weights for sitesof 250×250 m, ranged from 5.534 t/ha to 134.082 t/ha.The contribution of bamboo in AGB has been foundvery high. The analysis indicated that the highest correlation between AGB phytomass and red band (R)of MODIS satellite data of October was (R  2 =0.7823)and R  2 =0.6998 with both NDVI of October data aswell as NDVI max . High correlation (R  2 =0.402) with IR  bandofFebruarymonthwasalsofound.Thephytomassrange obtained by using MODIS data varies from0.147 t/ha to 182.16 t/ha. The mean biomass is40.50t/ha.Totalbiomassis31.44Mt.ThemeanCarbondensity is 19.44 tC/ha in forest areas. The study isvalidation of region-wise spectral modeling approachthatwillbe adopted for mapping vegetation carbon poolof the India under National Carbon Project of ISRO-Geosphere Biosphere Programme. Keywords  Phytomass.Spectralmodeling.Mean phytomass Introduction The issue of climate change and its impact on thenatural ecosystems has drawn attention of the world.Therefore, attempts are being made to understand thedynamics of the atmospheric carbon, particularly gapsin Carbon released and Carbon sequestered for improved prediction of future atmospheric Carbon.The periodic assessment of phytomass and carbon inforest ecosystem particularly in tropics has beenemphasized recently for improving Carbon balanceunderstanding (Houghton 1991; FAO 1990). In global vegetation carbon pools, forest vegetation carbonconstitutes nearly three-fourth, therefore, it is impor-tant to understand vegetation carbon cycle, and also tomake an assessment of the past atmospheric carbonreleases (IPCC 2003). The sequestration potential of the forest varies greatly by types, site, environment,and human interference, hence is very dynamic.Climatically and ecologically India is very diverse(Dadhwal and Chhabra 2000; Chhabra and Dadhwal2004). Attempts to understand the role of terrestrialecosystem of India have been made for biomass and productivity using ecological methods (Chaturvedi J Indian Soc Remote SensDOI 10.1007/s12524-011-0121-3P. Patil : S. Singh ( * )Indian Institute of Remote Sensing (IIRS),Dehradun, 248 001, Indiae-mail: sarnam.singh@gmail.comV. K. Dadhwal National Remote Sensing Centre,Hyderabad, 500 625, India    R  E   V  I  S  E  D   P  R  O  O  F and Singh 1987; Rawat and Singh 1988; Haripriya 2002) and still our understanding on these forests is poor (Kale et al. 2002). Therefore, an attempt is beingmade in National Carbon Project (NCP) to estimatethe total phytomass and carbon density in India for  plants/trees in-side and out-side the forest through a project taken up by Indian Space Research Organiza-tion (ISRO), Government of India under its ISRO-Geosphere and Biosphere Programme in XI FYP.The numbers reported on mean phytomass carbon pool and density in India are quite variable, e.g.2,587 TgC and 49.2 MgC/ha (Hingane 1991),3,117 TgC and 60.2 MgC/ha (Dadhwal et al. 1998),4,017 TgC and 63.6 MgC/ha (Dadhwal and Shah1997), which is mainly due to different methodologiesfollowed. The study by Flint and Richards (1991) onhistorical forest/vegetation carbon pool in 1880 inIndia indicates that total carbon pool and density in phytomass was about 7,940 TgC and 77.3 MgC/harespectively with total forest area of 102.68 Mha.Ravindranath et al. (1995) estimated the standing biomass (both above and below ground) of 8,375million tons for the year 1986 in India, of which thecarbon storage was reported to be 4,178 million tons.Most of the studies related to volume and phytomassat national level are based either on raw data fromstate forest department or growing stock estimates.Traditionally forest departments have focused oncommercial wood or timber species and therefore,have no data on plants with <10 cm diameter,however such plants contribute significantly. At patchlevel several methods have been proposed to estimateforest biomass (Whittaker  1966; Ovington 1968; Brown et al. 1991; Kale et al. 2002). Of these non- destructive or least destructive approaches of easilymeasurable plant parameters, such as diameter/girthand/or height and their relationship with plot or point volume and/or biomass are followed currently (Kiraand Ogava 1971; Tiwari 1992; Lodhiyal and Lodhiyal 2002; Roy and Ravan 1996; Kale et al. 2004). Remote sensing data has been widely used for  phytomass estimation (Sader et al. 1989; Tiwari1994; Kale et al. 2002; Foody et al. 2003; Lu 2005). Bamboo, a very high density biomass, is welldistributed in tropical to temperate forests. It growsvery fast and attains a height of 40 m in 3  –  4 monthsand can yield 50 t/year/ha (Vasishth et al. 2008).  Acacia catechu  Willd. is another economicallyvery important species and biomass ranges 29 to223.46 t/ha in plantations of Punjab (Rawat et al.2008). The biomass ranged from 39.4 kg tree-1 to738.98 kg tree-1 and from 77. in all these sites.Phytomass reports from semi-arid region of India aresummarized in Table 1. The implicit use to GeographicInformation System (GIS) based spatial modelingapproach uses spatial databases of climate, edaphic,geomorphological, vegetation indices and phytomassdegradation ratios as a function of population densityto improve the prediction of models.Medium to coarseresolution satellite data have been used for phytomassassessment in large landscapes (Muukkonen andHeiskanen 2007).Since India has a large climatic variability, it is planned to collect units from field data samples indifferent ecological zones. It is envisaged to developstrata-wise zonal models for each vegetation/forest type based on remote sensing data and average phytomass/ carbondensityin250×250mplots.Thepresentstudyisan attempt to validate the biomass estimates for semi-arid region of India. Material and Methods Study AreaThe study area lies in between 20°50 ′  02.57 ″  to21°33 ′ 04.62 ″  N latitude and 72°34 ′ 38.24 ″  to 74°21 ′ 00.48 ″  E longitudes in southern part of Gujarat statein India (Fig. 1). Topographically this region is mostlyflat and interspersed with isolated hillocks of 44.8 mto 578 m. The average annual rainfall is about 1,489 mm with a wide range of variation from995 mm to 2,481 mm. Soils in general are black cotton types with varying proportion of loam. Thearea has mixed deciduous forest of slightly moist todry Teak ( Tectona grandis ). Manvel bamboo is foundin few pockets. The characteristic species are teak andits common associates. The southern dry mixeddeciduous forests occur in drier places. Dry deciduousscrub has coppiced growth of teak,  Terminaliacrenulata,  etc. which is scattered throughout the studyarea.Data PreparationSix scenes of 8 day composite images of MODISSurface Reflectance (SR Product MOD09, path/row J Indian Soc Remote Sens    R  E   V  I  S  E  D   P  R  O  O  F 24(h)05(v), 06(v) and 25(h)06(v)) of 2007 belongingto February, March, May, October, November andDecember were down loaded from GLCF site (http:// ). Spatial resolutionof MODIS SR for Band 1 and Band 2 is 250 m,which is daylight data only. LISS III (spatialresolution 23.5 m and bands were green, red, infraredand short-wave infrared) data of dry and wet seasons(Table 2) were co-registered with the help of geotiff images of Landsat ETM + . Geo-referencing of satelliteimage and processing of data was done in ErdasImagine 9.1 software and ArcGIS 9.1 for creating thedatabase. The ground control points were well andevenly distributed and RMS error achieved was lessthan half pixel. Forest cover density map was procured from Forest Survey of India (FSI 2003)and used for sampling design. Forest cover type anddensity, NDVI of Landsat images and accessibilitywere used for site identification.MethodsField Data CollectionThe ground truth data was collected during the fieldwork for vegetation cover type mapping and samplingdesign for phytomass inventory. Two stage clusteringapproach had been followed for collecting field data.Four sample plots of 0.1 ha were laid at each site of 250×250 m which is equivalent to the size of MODISSR pixel with 250 m spatial resolution. A total of 48sample plots at 12 sites for tree enumeration werelaid. A sampling intensity of 0.00035% has beenachieved. The number of sites were distributedconsidering the probability proportion to its size of the forest type and density. The plot size of 0.1 ha was based on earlier surveys carried out by state forest department and Forest Survey of India. A format wasdesigned to collect the data on various parameters like S. No States Sites Biomass (t/ha) Author 1. Madhya Pradesh Madhav National Park 7.42  –  129.95 Roy and Ravan 19962. Madhya Pradesh Shivpuri district 17.6  –  36.5 Kale 20023. Madhya Pradesh Chindwara district 28.12  –  85.26 Pande 20054 Rajasthan Udaipur 28.2 Ranawat and Vyas 1975 Table 1  Above groundforest biomass estimates inwestern India Fig. 1  ( a ) Location of the study area, ( b ) study area false colour composite of IRS LISS III and PAN merged data with sample point locationsJ Indian Soc Remote Sens    R  E   V  I  S  E  D   P  R  O  O  F diameter at breast height (dbh) at 1.37 m aboveground, height of trees at ultimate forking, number of  bamboo rosettes and culms in 0.1 ha plot. The dbhand height measurements of the plants with  ≥ 3 cm to<10 cm diameter were also observed. In the case of  bamboo the number of culms from 6 to 7 rosettesfrom large, medium and small size was recorded in0.1 ha plot. The diameter of culms in diameter classes(thick, medium and thin in 4  –  5 different rosettes) of different clump size was noted.Phytomass EstimationSite-wise sample plot field data on plant species with ≥ 10 cm dia. and  ≥ 3 to <10 cm dia. were organized inspreadsheet. The data were converted into requiredunit as per the volume equations. The phytomass wasestimated for each density-wise forest cover types.(a) Plants with dbh  ≥ 10 cmWood volume of individual trees was estimatedusing local species specific volumetric equationsusing either dbh (diameter at breast height) and/or height of the trees. The volume was multiplied byspecies specific gravity to obtain the biomass. For thisa literature was surveyed to find out the site andspecies-specific volumetric equations and specificgravity from southern Gujarat published and compiled by Forest Survey of India (FRI 1996) and ICFRE(1996  –  2002). A total number of 61 sites/region specificvolumetric equations and species gravity of tree werefound and used. For the remaining species generalequations was used. The phytomass of the individualtreeswasaddedtofindthebiomass oftheeachplot.Themean site phytomass was obtained for each site.(b) Plants with dia.  ≥ 3 and <10 cmThis component of biomass has not been normallyignored inearlier studies, however, there contribution in phytomass is very significant, therefore, phytomass for this component has been estimated and which is usefulas a correction factor. At first basal area was estimatedforplantswith ≥ 10cmdia.andforallplants( ≥ 3cmdia.)within the plot. The phytomass (t/ha) estimated above(i.e. for plants with  ≥ 10 cm dia.) was regressed with basal area (m 2 /ha) of the plants with  ≥ 10 cm diameter using simple linear model. The regression coefficientsthus obtained were applied on basal area of trees with ≥ 10 cm dia. and all plants ( ≥ 3 cm dia.) within the plot to obtain phytomass using linear model of Y = α  + β *x, where, Y is phytomass,  α   and  β  are coefficients andx is the independent variable (i.e. basal area). To obtainthe phytomass of plants with  ≥ 3 cm to <10 cm dia., the phytomass of class with  ≥ 10 cm dia. was subtractedfrom the phytomass of all plants within the plot. Theseestimates werethenadded totheobservedphytomass of  plants with  ≥ 10 cm dbh (obtained using volumeequations and specific gravity) to get phytomass (t/ha)(Fig. 2).(c) Bamboo Phytomass EstimationTo estimate the phytomass of bamboo-culms quarter girth formula (girth/4) 2 *length) has been used to obtainvolume; and to obtain phytomass, volume was multi- plied with specific gravity (ICFRE 1996  –  2002). Theculms of bamboo were categorized into thick, mediumand thin based on girth class in different clumps within0.1 ha plot. Mean culm-phytomass was estimated for each girth class. It was then multiplied with number of culmstoobtainmeanphytomassoftheobservedrosettes.Mean rosette-phytomass was multiplied with number rosettestoobtainthebamboo-phytomassin0.1ha,whichwas then added to plant biomass, discussed in para (a)and (b) to get the phytomass of 0.1 ha plot.(d) Weighted Area Phytomass EstimationMODIS SR 250 m data have been used to createvector-boxes around the 12 sampling sites in Arc GIS,whose coordinates were taken using a GPS during fielddata collection. These vector-boxes of 250×250 m werethen overlaid on LISS III and PAN merged data. Visualinterpretation of merged data within each vector-box for mapping for forest cover type/land use and forest cover density maps was carried out. Forest type and densitymaps were intersected in GIS domain to obtain density-wise forest cover type map. The area (in ha) of eachforest type was multiplied by respective phytomass(t/ha) to obtain total phytomass for that type and density Table 2  Details of satellite data (IRS LISS III)Sl. No. Path Row Wet Season Dry season1 93 57 19 October, 2006 10 April, 20062 94 57 24 October, 2006 09 May, 20063 94 58 24 October, 2006 15 April, 20064 95 56 10 October, 2005 20 April, 20065 95 57 10 October, 2005 20 April, 2006J Indian Soc Remote Sens    R  E   V  I  S  E  D   P  R  O  O  F within the vector-box. Proportionate area of each landuse/land cover class occurring within the vector bound-ary was obtained by taking the ratio between the areaoccupied by the respective class within the pixel andtotal area of the MODIS pixel i.e. 6.25 ha. The classessuch as water body and settlement were not considered.Subsequently, area weights based on per cent areaoccupied in the MODIS pixel by each land use/landcover class was multiplied with the corresponding phytomass to obtain the total biomass in the vector-box.Areaweightedphytomass was obtainedfor eachsamplesite.(e) Spectral ModelingThe spectral modeling to correlate phytomass withreflectance of multi-season MODIS data and toextrapolate the phytomass in non-sampled areas wasdone. Four regression function such as linear, loga-rithmic, exponential and power were tried to find the best model with data-sets such as red and infrared bands, NDVI, NDVI Max , NDVI Mean , NDVI Min,  and NDVI Median  and NDVI Amplitude  with weighted area phytomass based on clustered sampling as well as phytomass of individual plots. Correlation coefficientsof best fit models thus obtained were used to model phytomass for the entire area/region case by case. Results and Discussion Estimation of Phytomass for Trees ( ≥ 10 cm dia.)Theplot-wise(  N  =48) phytomass ranges from 6.13 t/hain open degraded ecosystems to 389.16 t/ha in densemixed moist deciduous forests including bamboo phytomass. The phytomass of bamboo varied from0.947 t/ha to 186.166 t/ha, which is significant. Thecontribution of bamboo phytomass is quite high. Theculms are either hollow or solid but   “…  bamboo is verydense wood and its density runs 2 to 3 times that of the Fig. 2  Approach for estimating biomass for plants  ≥ 3 and <10 cm dia. class y = 5.3612x - 1.1414R 2  = 0.90610501001502002500510152025303540 Basal area (m 2  /ha)    T  r  e  e   A   B   G    B   i  o  m  a  s  s   (   t   /   h  a   ) Fig. 3  Correlation between phytomass and basal areaJ Indian Soc Remote Sens
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks