Scrapbooking

A national forest resources assessment for Costa Rica based on low intensity sampling

Description
A national forest resources assessment for Costa Rica based on low intensity sampling
Categories
Published
of 15
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
Share
Transcript
  A national forest resources assessment for Costa Ricabased on low intensity sampling Christoph Kleinn a, *, Carla Ramı´rez b,1 , Peter Holmgren c ,Sonia Lobo Valverde d , Guido Chavez e a Chair of Forest Assessment and Remote Sensing, Universita¨ t Go¨ ttingen, Bu¨ sgenweg 5, D-37077 Go¨ ttingen, Germany b Consultant to the Pilot Inventory in Costa Rica, Biostatistics Section, CATIE, Turrialba 7170, Costa Rica c Forest Resources Development Service, Forestry Department, Food and Agriculture Organization of the United Nations, FAO, Rome, Italy d SINAC—Ministry of Environment and Energy, Section of Forest Fires and Payment for Environmental Services, SanJose, Costa Rica e  Assistant to the Vice Minister of Environment and Energy, SanJose, Costa Rica Received 7 May 2003; received in revised form 18 January 2005; accepted 7 February 2005 Abstract A goal of a National Forest Inventory (NFI) is the provision of information which is relevant and required for national leveldecision making and monitoring in forestry, but also for related sectors.Thispaperpresentsanddiscussesapilotstudy fromCostaRicawherein2000/2001alowintensitysamplingapproachwasused to generate national level forestry information. On a 15 km  15 km grid air photo plots were interpreted for forest andland cover type. Readily available 1997 aerial photographs were used that were, however, only available for about 70% of thecountry: of the 228 grid points for the whole country only 159 could be aerial photo interpreted. Out of the 15 km  15 kmbase grid of sample points, a 2  3 subset was selected for field assessment, resulting in a sample of 40 cluster plots,each comprising of four elongated rectangular sub-plots of 150 m  20 m located on the perimeter of a square of 500 m sidelength.Two novel components were integrated into the inventory: (1) the field plots were established on all lands, so that the treeresource was not only tallied inside forests but also on all other tree-bearing lands outside forests. (2) In addition to thebiophysical information gathered on the traditional field plots, interviews were carried out with forest owners on the site of thefield plots, in order to obtain data on the use of the forest resource.Fieldworkwascarriedoutby6fieldcrewsandtookaltogetherabout3months.Resultsweregeneratedfromthefieldsamplesfortheentirecountry.Aerialphotobasedareaestimateswerecomparedtothecorrespondingestimationsfromfieldsamplingforthe same area. According to the field sampling the forest cover for Costa Rica in 2001 is estimated to be 48.4% (simple standarderror percent 9.3%). An estimated 8.2% of the total volume (dbh > 30 cm, all species) is outside forest. www.elsevier.com/locate/forecoForest Ecology and Management 210 (2005) 9–23* Corresponding author. Tel.: +49 551 39 3473; fax: +49 551 39 9787. E-mail addresses:  ckleinn@gwdg.de (C. Kleinn), cramzea@intelnet.net.gt (C. Ramı´rez), peter.holmgren@fao.org (P. Holmgren),sonilobo@minae.go.cr (S.L. Valverde), guidocha@minae.go.cr (G. Chavez). 1 Present address: National Coordinator of the National Forest Inventory in Guatemala, FAO-Guatemala, 12 calle 1-67 zona 14, ciudad deGuatemala, Guatemala. Tel.: +502 363 55 60; fax: +502 263 55 50.0378-1127/$ – see front matter # 2005 Elsevier B.V. All rights reserved.doi:10.1016/j.foreco.2005.02.023  This inventory took place with support from Food and Agriculture Organization (FAO) in the framework of FAO – ForestResources Assessment ’ s(FRA) Program Support toNational ForestAssessments; it was carried outjointly by Sistema Nacionalde A ´ reas de Conservacio ´ n (SINAC), the Costa Rican authority responsible for forestry issues, and Centro Agrono ´ mico deInvestigacio ´ n y Ensen ˜ anza (CATIE), an international agricultural research center. Experiences of the study were subsequentlyused to implement similar inventories in three more countries (Guatemala, Cameroon, The Philippines). # 2005 Elsevier B.V. All rights reserved. Keywords:  National forest inventory; Trees outside the forest; Forest cover estimation 1. Introduction The principal goal of a National Forest Inventory(NFI) is the provision of information which isrelevant and required for national level decisionmaking andmonitoring inforestry,and also inrelatedsectors (Cunia, 1978). These large area exercises aresometimes referred to as forest inventories at the strategic scale (Schreuder,2001),contrastingthemtosmaller area forest inventories on a tactical scale suchas forest management inventories or operationalinventories. NFIs also provide data for sub-nationalgeographical or political units and are an input toglobal forest assessments and other internationalprocesses in the context of sustainable managementof the natural resources, such that there is aconsiderable and generic international interest inNational Forest Inventories and the information theygenerate.Some countries have no or no reasonably up-to-date NFI. Persson and Janz (1997)  fi nd that in manyNFIs the overall reliability is low in what refers toresultsandapproacheschosen.Aninterestingquestionis what the reasons may be that various countries donot yet have a comprehensive NFI in place althoughamong forest and natural resource managers the needfor up-to-date and high quality forest resourceinformation on the national level is usually clearlyrecognized. Major reasons for not having NFIsimplemented are probably of budgetary nature andinsuf  fi cient political prioritization.Field work is one of the major cost items of NFIs  – and a crucial element at the same time. Most of theforestry-relevant variables can only be assessed withreasonable accuracy in the  fi eld. Reducing theintensity of   fi eld sampling while maintaining statis-tical soundness is a step to make NFIs  ‘‘ affordable ’’ . If the major goal is to generate and update forestinformation  on the national level , NFIs may be carriedout with a lowintensity sampling strategyas presentedand discussed by Thuresson (2002). That approachmakes information available in a relatively short timeand at relatively low cost so that an NFI is notprohibitive from a budget point of view.The Forest Resources Assessment Program (FRA)of the United Nation ’ s Food and AgricultureOrganization (FAO) has proposed to investigatethe virtues of this approach in the context of their initiative to assist countries to gather reliablenational forest information (Saket, 2002). A corre-sponding pilot study was carried out in 2000/2001 inCosta Rica where details of the approach, itsimplementation and design were developed andtested. This paper describes the approach andpresents some experiences, results and conclusionsof this pilot study. The complete report of the resultsis in FAO (2001b). 2. Some forest information for Costa Rica Costa Rica has an area of about 51,000 km 2 and avery diverse topography and vegetation, and was oncealmost 100% forested (Keogh, 1984). The period of most severe deforestation was between 1950 and1980, and principal causes were reported to be thedemandforlandrather thanforwood (Hartshorn etal.,1982). Costa Rica was at that time among thecountries with the highest deforestation rate world-wide: Leonard (1986) reported a deforestation rate of 3.9% per year for 1950 – 1984. Today, Costa Rica hasan ef  fi cient conservation system in place, and about25% of the national territory are protected areas, mostof it covered by forest.Much has been published on the forest area of Costa Rica. FAO (2000a) compiled an annotatedbibliography on forest cover change in the country.Kleinn et al. (2002) compared published forest cover C. Kleinn et al./Forest Ecology and Management 210 (2005) 9  –  23 10  fi gures from the past 60 years (Fig. 1): while a cleartrend isvisible,the highvariationaround thetrend lineis striking. Those forest cover  fi gures are publishedand cited  fi gures, yet only few  fi gures srcinate fromsrcinal studies, and many are citations and modi fi ca-tions of earlier studies. Some publications did notreveal clearly where the  fi gures came from. In mostcases the underlying de fi nition of   ‘‘ forest ’’  was notgiven nor were the forest types considered speci fi ed.Statistically based forest inventory data for the entirenationalterritorywerepublishedforthereferenceyear1967 under an FAO project (Sylvander, 1981). For the1980s very low forest cover  fi gures were reported(e.g Sader and Joyce, 1988), down to less than 20%.In later studies, forest cover was determined withconsiderably higher values.The more recent studies are satellite imagery basedforest mapping studies. A recent one was carried outwith Landsat 7 TM imagery for the year 2000 byEOSL, CCT and FONAFIFO (2002). There, a forestcover of 46.3% was found, which includes plantationsand mangrove forests, and a forest de fi nition with aminimum crown cover of 80% was used.In the 10-year forest development plan for CostaRica, enacted in March 2001, the relevance of up-to-date information on the state of the forests isrecognized and the development of an informationsystem recommended (MINAE et al., 2001). TheForest Resources Assessment Program of FAO offeredtoassist implementingapilot study for a nationallevelforest inventory. 3. Methods: the inventory design 3.1. General characteristics  —   population and sampling frame The population of interest for the biophysicalinventory was all tree cover in Costa Rica. Thetopographic map of Costa Rica was used as apreliminary area sampling frame, updated by morerecent aerial photography for most areas (see below).The islands were not included as they represent a verysmall percentage of the national territory.The population was deliberately chosen to be notonly forests, but also including the tree resourceoutside forests (TOF) which is increasingly recog-nized as an important landscape element and aresourceprovidingmanyenvironmentalandeconomicservices and bene fi ts (FAO, 2000b; Kleinn, 2000;Sadio et al., 2002). This meant that, unlike intraditional forest inventories,  fi eld plots and aerialphoto interpretation did not stop at the forest boundarybut extended into all other land uses, except thosewhere natural conditions do not allow tree growth C. Kleinn et al./Forest Ecology and Management 210 (2005) 9  –  23  11Fig. 1. Forest cover  fi gures for Costa Rica as published in 54 sources (modi fi ed from Kleinn et al., 2002, Fig. 2 with kind permission of Kluwer AcademicPublishers). The dottedstraightlinecomesfrom fi gurespublishedin the FORSTAT databaseof FAO. Thereis a clear downwardtrenduntil about to the mid-1980s.After that an upward trend is visible, thoughwith a relatively high variability. It should be noted that the  fi gures arepublished  fi gures and are not necessarily srcinal  fi gures or  fi gures based on srcinal forest inventories.  (such as waters, barren land, areas above the timberline).A second featurerelatively novelfora tropicallargearea forest inventory was the inclusion of interviewswith land owners. The objective of thiscomponentwasto obtain information about past, present and projectedfutureuseoftheforestandtreeresourcesthatgobeyondwhat can be directly observed in the  fi eld. Thecorresponding population of interest was de fi ned forthis study as the  ‘‘ community ’’  of land owners in CostaRica. This population constitutes only a part of theforest users, with a speci fi c expectation towards theirforest areas (the  ‘‘ owner ’ s perspective ’’ ); however, anexpansion of the interviews to the group of all forestuserswasnotfeasibleinthisstudy:whilethepopulationoflandownersisrelativelyeasilyde fi ned,itwouldbeamuchlargerandmethodologicallyverycomplextasktodesignandimplementaninterviewsurveyforallgroupsof users. Interviews, workshops, meetings and groupdiscussions are common techniques in many commu-nity forestry projects; interviewers may spend con-siderable time with the stakeholders, build upcon fi dence and get a direct insight into the dynamicsof the community. This is one of the major differencesin large area forest assessments and a major imple-mentation challenge at the same time: interviews mustbe more focused and usually restricted to a short visit.Field sampling was a major component of theinventory providing direct and  ‘‘fi rst-hand ’’  observa-tions on a series of relevant forest variables with ameasurable (estimable) precision. To increase preci-sion of area estimations of forest types and land useclasses, aerial photograph interpretation was inte-grated. Aerial photography from 1997 was availablefrom a  fl ight campaign (Proyecto Terra) carried out tocompilenewtopographicmaps.Becauseofpermanentcloud cover in the northern part of Costa Rica, thismost recent available aerial photography covered onlyabout 70% of the country ’ s area. This study did nothave the resources to commission new imagery; so, 3 – 4 years old photographs were used, even though itwouldalwaysbepreferabletoobtainimageryclosertothe inventory date. Though it was not possible toderive estimates for the entire country from thisimagery, the aerial photograph component wasintegrated into this pilot study for methodologicalreasons, and the results obtained were compared to theresults of the corresponding  fi eld plots. 3.2. Land use classes The system of land use classes used (Table 1)was based on the FAO classi fi cation details of whichare in FAO (1998). Obviously, not all classes couldbe equally well distinguished in the aerial photoplots and in the  fi eld. The upper level classescorrespond to the FAO classi fi cation, the moredetailed classes are in part nationally adapted. Gallery forests , for example, elongated narrowforest strips along waters in otherwise non-forestedlandscape was speci fi cally introduced for this study,as it was known that this type of forest is typical inmany regions in Costa Rica. Secondary forest is aforest type which bears, depending on the develop-ment stage, some challenges in what refers todistinction to other forest types. Secondary forestgrows on land which had been under different landuse before. They are recognized above all by aspeci fi c species composition which can be assessedin the  fi eld; in the aerial photographs crown structureand contextual information were used to distinguishit from other forest types. A set of ground checkedexamples (interpretation key) served as referencefor aerial photograph interpretation. However,observation errors are present for all attributesassessed in forest inventories, and interpretationconfusions of land use classes can never becompletely eliminated. 3.3. Sampling design For the establishment of   aerial photo plots systematic sampling was used and a basic squaregrid with a 15 km side was laid out over the countrywith a grid orientation de fi ned as north – south and thestarting point being  x  = 289500,  y  = 895500 in therectangular geodetic grid Lambert Sur. A total of 228points fell onto the territory of Costa Rica (light graydotsinFig.2)outofwhich n a  = 159pointsarecoveredby aerial photographs (square frames in Fig. 2).For the  field plots  a subset of the aerial photosample points was selected. Due to budget con-straints a maximum of 40  fi eld samples could beestablished, so that a 2 (east – west)  3 (north – south)subset of the basic 15 km  15 km grid was chosenresulting in a grid of 30 km  45 km (bold dots inFig. 2). C. Kleinn et al./Forest Ecology and Management 210 (2005) 9–23 12  3.4. Plot design Fig.3showstheplotdesignfortheaerialphoto plotand for the  fi eld plot. The  aerial photo  plots forinterpretation were established on the aerial photoclosest to the sample point. The plot center was closeto the center of the photograph, keeping the level of geometric distortions relatively low. The technicallybetter alternative of image recti fi cation was beyondthe projects resources and ortho-recti fi ed or geo-referenced photographs were not available at thattime. Because the size of all aerial photo plots is  fi xedat a square of 9 cm  9 cm on the contact print of theaerial photograph, the corresponding plot size in the fi eld will actually vary as a function of topography andimage characteristics. Among the 159 aerial photoplots the actual scale varied between about 1:32,000and 1:53,000, thus the resulting size of the plots in the fi eld varied between about 2.7 and 4.5 km. None-theless, the size of the  fi eld plots is  fi xed. Field plots  were established on a 2  3 subset of the aerial photo plot locations. The center of theclusters offour sub-plots corresponded to the center of the aerial photo plot. In order to make the per-plotinformation content high  –  i.e. to keep intraclustercorrelation low  –  relatively large sub-plots weredesigned, and the distance between sub-plots withinthe cluster plot was kept relatively large. Those twofeatures, of course, had to be de fi ned within the limitsofwhat was practically feasible. Forthe square shapedclusters a side length of 500 m was de fi ned whichappeared to be the maximum possible under the forestand topographic conditions in the country. Around theperimeter of this square four rectangular sub-plotswith side length of 150 m were located. The strip-shaped sub-plots had a width of 20 m, thus allowingrelatively good visibility of 10 m to the right and leftwhile walking on the central track. Each of the sub-plotscoveredanareaof0.3 haandthetotalclusterplotan area of 1.2 ha. The sub-plot area was determinedaccording to prior information on estimated treedensity stemming from earlier forest inventories inCosta Rica. In each of those major sub-plots smallersub-plots were nested for observation of smaller treediameter classes (Table 2).Outside forest, nested plots were not installedbecause of the expected low density of smallerdimension trees. There, trees with dbh > 10 cm weremeasured and registered on the entire area of the sub-plot.Photographs of the surroundings of the  fi eld plotswere taken to document current land use and are partof the inventory documentation. C. Kleinn et al./Forest Ecology and Management 210 (2005) 9  –  23  13Table 1Land use classi fi cation used in this studyClassForestPrimary forest ClosedMediumOpenYoung secondary forest ClosedMediumOpenAdvanced secondary forest ClosedMediumOpenForest plantation ClosedMediumOpenGallery forest ClosedMediumOpenOther wooded landOther land without trees ShrubsFallowsTOF * land Woody grass land(5 – 10% crown cover)Other landOther land without trees Barren landNatural Grass landWoody grass land( < 5% crown cover)TOF land: land withtrees outside forestsAnnual cropsCultivated land Perennial cropsRange landBuilt up areas Built up area with blocksBuilt up area — no blocksInland waterOther non interpreted(in aerial photography)The density sub-classes  ‘‘ closed ’’ ,  ‘‘ medium ’’ ,  ‘‘ open ’’  refer tocrown cover percentages of  > 70%, 40 – 70%, and 10 – 40%, respec-tively, following FAO de fi nitions. * TOF stands for trees outside the forest. The area on which thesetrees are found is named  TOF land  .
Search
Tags
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