Specifications for a Predictive Ecosystem Mapping Standard 1. INTRODUCTION 1.1. OBJECTIVE The ob jectives of the PEM Stand ard ar e: 1. To en sur e that PEM data ar e deliver ed in a stand ard TEM- like fo rm that allo ws fo r mer gin g, integration, or co mpar iso n of multiple data sets . 2. To en sur e su fficient do cumen tatio n (QC /QA) to ev alu ate the accur acy of the pr oduct. 3. To pr ovide stability in the stand ard . To achieve stability , ch anges to the stand ard will be su bject to a ch ange man agement pr ocedur e that incor por ates an impact an aly sis of pr opos ed ch anges. 1.2. BACKGROUND The sp ecificatio n fo r a Pr edictive Eco sys tem Mapp ing (P EM) stand ard ar is es fr om fo ur pr ecedin g wor ks co mmiss ion ed by the Terrestrial Eco sys tem Mapp ing Alter nativ es Task For ce Res our ce In ven tor y Committee (R IC). These wer e: 1. Towards the Estab lishmen t of Pr edictive Eco sys tem Mapp ing Stand ard s: A White Paper , Keith Jo nes , R. Keith Jo nes & Ass ociates, Del Meidin ger , BC Min istry of For es ts, Res ear ch Br anch , Dav e Clark , BC Min istry of Env ir onmen t Lands & Parks , Res our ces In ven tor y Br anch , an d Fern Schultz, BC Min istry of For es ts, Res our ces In ven tor y Br anch ” 2. Pr oblem Analys is on Data Quality Ass ess ment Is sues by Dr. Dav id Moon , CDT– Core Decis ion Techno log ies In c., 3. Situation Analys is fo r Kno wled ge- Based Sys tems by Dr. Dav id Moon , CDT– Core Decis ion Techno log ies In c., an d 4. Pr oblem Analys is on Reliability, Quality Con tr ol an d Valid ation of Pr edictive Eco sys tem Mapp ing (P EM) by Dr. Richard Sims, an d Jeff Mathes on, R. A. Sims & Ass ociates. The white paper integrated elements of the su ppor ting studies to develop the ov erarch ing fr amewo rk pr esented in Figur e 1. The PEM Fr amewor k co mprises five main co mponents , each of which co ntain imp ortant elemen ts , features an d other related items . 1 The five main co mponents ar e — 1. Clien t Requ irement Ass ess ment 2. In put Data Quality Ass ess ment 3. Kno wled ge Bas e Ass ess ment 4. Output an d Reliability Ass ess ment, an d 5. QC / QA Pr ocedures an d Stand ard s • Figure 1 PEM Framework CLIENT REQUIREMENTS ASSESSMENT COMPONENT • Provincial Level Priorities • Regional-District Level • Project Level Requirements and & Integration Element Priorities & Integration Element Reliability Analysis & Planning Element INPUT DATA QUALITY ASSESSMENT COMPONENT KNOWLEDGE BASE ASSESSMENT COMPONENT OUTPUT & RELIABILITY ASSESSMENT COMPONENT • Thematic Elements • Spatial Elements • Procedural Elements • Meta Data Features • Evaluation Framework* • Documentation Element • KB Structure Element • Automated Inferencing Element • Field Checking Element • KB Validation Accuracy Element • Features: transparent, exportable, logic trace • Thematic Elements • Topological Elements • Positional Elements • Precision Elements • Evaluation Framework* • Sampling Element • Analysis Element • QC Linkage/Feedback QC / QA PROCEDURES & STANDARDS COMPONENT • PEM project planning • Evaluation Framework criteria, measures, thresholds • Map entity appropriateness • Input data quality catalogue • Options & protocols for input data quality improvement or alternate inventory approach • Training • Mandatory & preferred KB system elements & features • Documentation: KB & data structures, inferencing & PEM analysis procedures • Map unit rules & conventions • Recommended field checking protocol & KB validation accuracy thresholds • Training • Map-cartographic & legend conventions • Database & meta data requirements • Knowledge management • Reliability assessment protocols: sampling, analysis, reporting • Reliability thresholds, QC linkage • Training Comp onen ts 1- 4 have a nu mb er of definin g elemen ts an d features, which together ch aracterize their scope an d fu nctio n. The QC / QA Pro ced ures an d Stand ards co mponent 2 co ntains items that larg ely co rrespo nd vertically, fr om left to right, to the co re In put Data Qua lity As ses sment, Kno wledge Bas e As ses sment an d Output an d Relia bility As ses sment co mponents of the PEM fr amewo rk. The PEM co mponents closely parallel co mponents of the cu rrent TEM RI C stand ard s an d pr ocedur es. With PEM ho wever , the In put Data Quality an d Kno wled ge Bas e Ass ess ment co mponents ar e diff erent, bu t ar e so mewhat an alo gous to “Map pin g an d Field Sur vey Pr ocedures” section of the TEM RI C stand ard . The Output an d Reliability Ass ess ment co mponent of the fr amewo rk can be the same fo r bo th PEM an d co nven tional TEM. Pr esently this co mponent has no t been developed fu lly fo r TEM. The PEM stand ard will ad dress co mponents 2 – 5 an d will pr ovide bo th stand ard s an d recommendatio ns fo r each elemen t. It will also es tablish min imum levels of do cumen tatio n an d meta- data stand ard s to su ppor t ev alu ation of the accur acy an d utility of the PEM pr oduct. 2. ISSUES IDENTIFIED A two-day wor ksh op (J uly 6– 7, 19 99) identif ied a nu mb er of imp ortant PEM related issues. These issues related to required infr astructu re (I ), co ntr act sp ecificatio ns (C ), stand ard s (S ) an d gu idelines (G) . While all ar e imp ortant an d ar e identif ied in the table below, the stand ard will ad dress on ly those issues related to stand ard s an d gu idelines . Table 1 co mpiles the results of a wor ksh op session that identif ied issues related to PEM stand ard s an d identif ied fu nctio ns ad dressing the identif ied issues. The issues have been gr ouped into the fo llowin g categories : Mapp ing Entities, In for mation Management, Pr ocedures, Kno wled ge Management, an d Imp ortant issues beyond the scope of the stand ard . Each issue is flag ged with I, C, S, or G to indicate In fr astructur e, Con tr act, Stand ard , or Guidelin e. • Table 1 Issues and Functions Relevant to PEM Standards Issue Product definition Non-forested units Site-series generalization Information delivery Versioning Change management Function Mapping Entities Product definition provides documentation on the nature of the PEM being produced and should include: - Definition of the PEM or TEM classes used - Definition of the attributes used to define or characterize them - Rationale underlying choice of mapping and map entities - Relation to TEM classes “domain values” (explain) Establishment of standard classes (other classifications) for nonforested unit nomenclature Creation of more generalized TEM classes to accommodate areas recognizable as belonging to a group of more detailed classes but not recognizable as to specific class. Information management Specification of the form in which the data are delivered to the BC corporate data base (logical data model) Enhancement of the database to capture user-defined attributes Identification, labeling, tracking, reconciliation, distribution, documentation Establishment and implementation of change management procedures 3 S S S S S S - Projec t definition Attributes Data independence Accuracy Assessment Biogeoclimatic boundaries Cartographic output Software independence Knowledge base Input data Input data quality Knowledge base applicability Screening of proposed changes impact assessment notification of changes including what changed grandfathering projects and contracts timing for upcoming field season, project in progress Establishment of standards to define the geographic extent, resource discipline supported, strata used (TEM analogue), classes and attributes et cetera for the project. Develop and maintain a project management system to track ongoing and completed projects Documentation (rationale, definition, domain) - Core attributes - Interpretation specific attributes - User defined attributes - PEMplate of core attributes - See TEM table 4.1 and 4.2 Programs, validations, interpretations et cetera should be independent of hard coded parameters, parameter files, or other non-data base structures requiring specific non-domain expertise for maintenance. Procedures Specification of minimum requirements and characteristics of procedures to assess accuracy (procedure neutral) - Thematic - Spatial Specification of procedures to refine small scale lines for application at larger scales Specification of minimum requirements and characteristics - Guideline on applicability of TEM Standard - minimum set (e.g. polygon id, schema for attribution) Creation of standard requirements for edge matching - When appropriate / mandatory - Within/Between projects PEM<-->TEM and PEM<-->PEM Development of specifications for exchange formats or standards Define minimum content and elements for knowledge base documentation Define entities and relationships Specification of thematic data requirements - Mandatory themes (e.g. TRIM, BEC lines) Mandatory IDQ report as early contract deliverable? - Appropriate thematic input - Rationale for non-mandatory thematic selection - Create an input database catalogue - Develop procedures to evaluate spatial congruence - Guidelines for the evaluation of thematic content and format Knowledge Management - Guidelines for determining the applicability of existing KBs to new areas - Guidelines for the specification of boundary conditions for a KB when created - project boundary or selected strata within 4 S S S S S S S S G G Validation Custodianship Responsibilities Contracting Confirmation of scientific soundness or to corroborate results against a standard - Knowledge base - Corporate validation routines - Input data - Define appropriate validation procedures Document the degree of validation effort - TEM validation rules should show strong relationship with PEM knowledge base Important Issues beyond the scope of PEM Standards Mandate, responsibilities (as contract specifications), roles (as standards), resources, intellectual equity Roles and functions (e.g., correlation, validation, certification) Templates - Schedules A and B - RFP contents Evaluation criteria for Responses to RFP - Implications on scheduling and turn around of enforced PEM standards (including limitations on government QC/QA resources, task dependencies) - Expected efforts/costs associated with normal PEM tasks as a guide to proponents and contract administrators Conflict of interest with QC/QA contracts Full costing of the PEM process to support benefit/cost analysis Documentation of time associated with steps in PEM, including unbilled time for public service employees G I I C S = standard, G = guideline, I = infrastructure, C = contract 3. PRINCIPLES The wor ksh op identif ied the fo llowin g pr incip les that the stand ard will attempt to fo llow. 3.1. CONFORMANCE TO EXISTING STANDARDS Wher ever po ssible, the PEM stand ard will co nfo rm to an d utilize ex istin g RI C stand ard s by referen ce or attachment to ex istin g stand ard s rath er than by cr eatin g new stand ard s requiring co rrelation or recon ciliation. The PEM stand ard will fo llow the gener al pr oto col es tablished by the Res our ces In ven tor y Committee fo r pu blished stand ard s. RI C recog nizes five co mponents of a stand ard . These ar e: 1. A discipline stand ard related to clas sif ications , co ncepts , an d en tities in es tablished disciplines su ch as For es t In ven tor y, Soil Sur vey , or Terrain Mapp ing which sh ould be fo llowed when co llectin g or us ing data to su ppor t other inven tor ies. For ex ample, the Terrain Class if ication stand ard “Terr ain Class if ication Sys tem fo r Br itish Columbia, Ver sion 2. (1 997) . D. Howes an d E. Ken k” sh ou ld be fo llowed when co llectin g or us ing terr ain data elemen ts in su ppor t of a TEM. 2. A data co llectio n stand ard related to the method of sp atial an d thematic data captu re. These may be field, of fice, or co mputer- bas ed method s. For ex ample, “Field Manu al 5 fo r Des cribing Terrestrial Eco sys tems ” is the stand ard fo r site data co llectio n when us ed as a co mponent of ecosy stem map pin g. 3. A digital data captu re stand ard sp ecifying en tities, attr ib utes, data stru ctu res an d fo rmats fo r the su bmis sio n of digital data to the go ver nment repos itory . 4. A stand ard fo r interpr etations an d interpr etive method s ap pro priate to the data so urce to which the stand ard ap plies . 5. Stand ard s fo r co mmo n ou tpu t pr oducts . Gen erally, these ar e sp ecificatio ns fo r stand ard an alo gue map or repor t pr oducts gener ated fr om the digital data. 3.2. SOFTWARE INDEPENDENCE Wher ever po ssible, the stand ard will be indep endent of pr opr ietar y so ftware or pr opr ietar y so ftware co nstructs . Wher e the min is try respo nsible requires data in a vendor depend ent fo rmat or co nstruct the stand ard will en deavor to pr ovide stand ard co nver sion pr ocedur es or vendor indep endent fo rmats which can be imp orted into the min is try sy stem. 3.3. PROCEDURE INDEPENDENCE Wher ever po ssible the stand ard will be pr ocedur e indep en dent. For ex ample, the stand ard may require es timated accur acy pr ovided as the attr ib ute(s) being es timated, the es timated value, an d statis tically defined co nfiden ce limits rath er than requiring that a sp ecific pr ocedur e be fo llowed . In other cases it may pr ovide a min imum accep tab le pr ocedur e bu t allo w fo r other mor e so phisticated pr ocedur es that pr ovide ad ditio nal infor mation . 4. FRAMEWORK FOR THE PEM STANDARD In keeping with the white paper fr amewo rk, the PEM stand ard fr amewo rk will co nsist of the fo llowin g elemen ts . 4.1. RELATIONSHIPS BETWEEN PEM AND TEM STANDARDS The stand ard will identif y ar eas of ex act co rrespo nden ce, similar ity with, an d diver gen ce fr om ex istin g TEM an d, where relevant, other RI C stand ard s. This will include discipline stand ard s, map pin g stand ard s, digital captu re stand ard s, an d repor ting stand ard s. 4.2. MAPPING CONCEPTS The stand ard will identif y an d sp ecify stand ard s fo r the identif ication, do cumen tatio n, an d rationalizatio n of map pin g en tities an d map en tities. Mapp ing en tities ar e the basic ecosy stem elemen t being map ped e.g., Biogeo climatic zo ne, su b-zone, su b-zone ph ase, su b-zone variant, or site series, site series stru ctu ral stage, site modifier , site series seral stage. Map en tities ar e the labeled po lyg ons on the map an d may represen t diff erent co mpositions of map pin g en tities. The stand ard will also mak e recommendatio ns or set stand ard s fo r do cumen ting the relationsh ip between TEM an d PEM map pin g an d map en tities. 6 4.3. MAPPING CONVENTIONS The stand ard will identif y an d mak e recommendatio ns on the desig n an d imp lementation of co nven tions fo r the display of PEM infor mation in map fo rm. This may co nsist of sy mbol, label, an d line co nven tions . It will also mak e recommendatio ns or set stand ard s fo r the do cumen tatio n of the relations hip between PEM co nven tions an d TEM co nven tions . 4.4. QUALITY CONTROL / QUALITY ASSURANCE 4.4.1. Quality Control Quality co ntr ol stand ard s will co nsist of three co mponents . The firs t is a rigor ous , welldo cumen ted pr ocedur e that if fo llowed will pr od uce co nsistent results . The secon d is as sur ance that the pr ocedur e has been fo llowed . Notif ication is required if significan t deviation is co ntemplated. The thir d is the deter min ation an d monito ring of qu ality resulting fr om the ap plied pr ocedur e. The stan dar d will develop do cumen tatio n requirements fo r PEM pr ocedur es. The do cumen tatio n will pr ovide su fficient detail fo r the pr ocedur e to be co nsistently replicated an d monito red . Quality co ntr ol pr ocedur es sh ould be developed fo r ap plication to the deter mination of : 1. In put data qu ality . This will be ap plicable on ly if new input data ar e being pr oduced fo r the PEM pr oject. Wher e ex istin g thematic map s ar e being us ed, qu ality as sur ance pr ocedur es will be ap plied to deter min e the qu ality of the input map s. 2. Pr eparatio n an d co mpilation of the input data base. 3. The validity of the kn owledge base. 4. The imp lementation of the kn owledge base ag ain st the input data. 4.4.2. Quality Assurance Quality as sur ance tests the final pr oduct to en sur e that a sp ecified stand ard of qu ality has been achieved . The test is indep endent of the method s us ed to pr oduce the pr oduct bu t it may be ph ased to test diff erent stages of pr oduct co mpletion. Pr ocedures will be developed an d do cumen ted to ev alu ate input data qu ality (thematic an d sp atial) , the kn owledge base us ed to pr edict PEM un its, an d the imp lementation of the kn owledge base ag ain st the input data to pr edict PEM un its. Thematic input qu ality on ex istin g thematic map s will of neces sity be qu alitative since field-tes ting will be pr oblematic. 4.5. DOCUMENTATION STANDARDS The PEM stand ard will es tablish min imum levels of do cumen tatio n an d meta- data pr ovided to ev alu ate the qu ality fo r the input data, pr edictive pr ocedur es, an d ou tpu t pr oducts of PEM. In the initial version of the stand ard s the ev alu ation may have to be qu alitative bu t the do cumen tatio n sh ould be su fficient to begin refinin g pr ocedur es an d to ap ply mor e qu antitative ratings to input an d ou tpu t data qu ality . 7 4.5.1. Data input The stand ard will identif y meta- data that sh ould be as sociated with input data su ch as fo res t co ver , terr ain, so il, an d biogeoclimate. The meta- data will ch aracterize co ntent, qu ality , an d fo rmat of bo th the thematic an d sp atial data. 4.5.2. Data input quality The stand ard will sp ecify meta- data relatin g to the deter min ation of thematic an d sp atial qu ality an d of the ap plicability of the input data to PEM. The stand ard will also referen ce qu ality as sur ance pr ocedur es developed fo r utilizing su rro gate infor mation to pr ovide qu alitative es timates of data qu ality in the ab sen ce of ap pro priate meta- data. 4.5.3. PEM process The stand ard will sp ecify the mater ials neces sar y to do cumen t the PEM pr ocess in su fficient detail fo r the pr ocess to be ev alu ated fo r su itability an d ap plicability. The stand ard will also sp ecify do cumen tatio n of the pr ocess in su fficient detail fo r it to be replicated, monito red , an d validated. 4.5.4. Analytical procedures The stand ard will sp ecify requirements fo r do cumen tatio n of an d the rationale fo r the an aly tical pr ocedur es us ed in the PEM. It will require su fficient detail to ev alu ate the ap plicability of the pr ocedur e an d to replicate an d validate the results . 4.5.5. Knowledge base and algorithms The stand ard will sp ecify do cumen tatio n requirements fo r the kn owledge base us ed in kn owledge- based pr edictive ecosy stem map pin g. The do cumen tatio n will include the logic, pr ocedur es an d as sumptio ns to facilitate meanin gfu l Q/C, as well as the sp ecificatio n of en tities, attr ib utes, ru le bases , belief co eff icien ts , an d algor ithms us ed in the pr edictive pr ocess. . The do cumen tatio n sh ould also include the pedig ree of the KB - the ex istin g KBs that fo rmed the basis of the su bject KB, the natur e of su ccessive iter ation s, co rrelation pr ocedur es to which it was su bject ... 5. PRIORITY ASSIGNMENTS Teams have been cr eated to pr epare an d distribute fo r discu ssion the fo llowin g items by Aug ust 15 , 19 99: 5.1. MAPPING CONCEPTS A on e-d ay wor ksh op was held to es tablish stand ard s fo r the definitio n, rationale, an d do cumen tatio n of PEM map pin g en tities an d map en tities. Pr ovisio nal map pin g an d map un its will be pr opos ed fo r discu ssion. Asp ects of section 5.2 will be influen ced by this wor ksh op. 8 5.2. RELATIONSHIP TO TEM STANDARDS 5.2.1. PEM Data Architecture The ex istin g TEM data ar chitecture an d TEM stan dar d fo r digital data captu re will be reviewed fo r ap plicability to PEM ap plications. The sp ecific issues of min imum co re data, the ab ility to ad dress no n-s tan dar d data, an d no n-f orested ecosy stem data will be ev alu ated an d recommendatio ns su bmitted. (Darr en McKellar et al., Jo hn Jo hnso n, Car men Cad rin, Dav e Clark .) 5.2.2. PEM Mapping Conventions The ex istin g TEM map pin g co nven tions will be reviewed fo r ap plicability to PEM map pin g en tities, map en tities, an d map pr oducts . If ap pro priate, a PEM sp ecific stand ard will be cr eated to be distributed indep endently or to be ap pend ed to the TEM stand ard . (B ob Maxwell, Ted Lea) 5.3. QUALITY CONTROL / QUALITY ASSURANCE 5.3.1. Input data quality Req uir ed elemen ts fo r do cumen ting an d ev alu ating the qu ality of thematic an d sp atial data inputs will be pr epared an d distributed fo r discu ssion. In ad ditio n, a pr ocedur e to deter min e the sp atial qu ality of input data map s will be pr epared an d distributed fo r discu ssion. (Dave Moon , Mar vin Eng , Br uce Enn s, Ev Ken k) 5.3.2. Biogeoclimatic Unit scale refinement Pr ocedures fo r localizin g Biogeo climatic bo undaries fr om smaller scale map s to the scale of the PEM will be developed, do cumen ted , an d distributed fo r discu ssion. (Marvin Eng , Del Meidin ger ) 5.3.3. Output data quality A pr ocedur e fo r deter min ing ov erall PEM ou tpu t qu ality fo r a map will be pr epared an d distributed fo r discu ssion pr ior to the secon d wor ksh op. (Del Meidin ger ) 5.4. DOCUMENTATION / META-DATA REQUIREMENTS Pr oject definitio n (Darr en McKellar, Dav e Moon ) The PEM Pr ocess (Keith Jo nes ) In put Data (Dave Moon ) Kno wled ge base (Dave Moon ) Pr edictive Algorith ms an d Imp lementation (Dave Moon ) Quality Con tr ol / Quality Ass urance (Dave Moon , Del Meidin ger ) 9 5.5. CHANGE MANAGEMENT 5.5.1. Versioning and version control A pr ocedur e an d stand ard fo r version man agement, co ntr ol, release, an d distribution will be pr epared . (S heila Jeck) 5.5.2. Change Management A requirements an aly sis an d su gges ted pr ocedur e to screen, as ses s imp acts, track versions , time releases, an d co mmu nicate ch anges to the PEM co mmu nity will be pr epared an d pr esented fo r review. This will fo llow the cu rrent RI C po licies fo r ch ange man agement. (S heila Jeck) 5.6. CREATION OF A PROTOTYPE ARCHITECTURE TO SUPPORT THE STANDARD A pr oto typ e ar chitecture to or gan ize, po pulate, man age, an d review the stand ard will be cr eated fo r us e du rin g the secon d wor ksh op. The pr oto typ e will man age the co mponents of the stand ard s, their relationsh ips to each other , an d identif y the imp licatio ns of ch anges to on e or mor e of the co mpo nents. (S heila Jeck an d Dav e Moon ) 6. APPROACH AND SCHEDULE The developmen t of the stand ard s will pr oceed in five stages. Stage 1 (J uly 7) Stage 1 co nsisted of a two-day wor ksh op to define the ob jectives of the stand ard , the issues to be ad dressed, an d the elemen ts of the stand ard needed to ad dress the issues identif ied an d meet the defined ob jectives . Stage 2 (J uly 20 ) The results of the wor ksh op wer e sy nth esized into a sp ecificatio n fo r a stand ard that identif ies the ob jectives , issues ad dressed, stru ctur es an d fu nctio ns that the stand ard mus t su ppor t. This repor t is pu blished on the web an d yo u ar e no w reading it. Stage 3 (Au g 15 ) The team as signmen ts identif ied in Section 5 - 0 - ar e su bmitted fo r po sting to the web site. Stage 4 (S ept 8) Follo wing review of the team as signmen ts an d their incor por ation into the pr oto typ e kn owledge fr amewo rk, a secon d wor ksh op will be held to review an d ap pro ve the co ntents of the fr amewo rk fo r pr eparatio n as a 1st ap pro ximation PEM stand ard . 10 Stage 5 (Oct 1) Decis ion s taken at the secon d wor ksh op will be incor por ated an d a dr aft stand ard cr eated fo r web distribution to the TEM an d PEM co mmu nities fo r co mment an d review. The review period will close October 15 . Stage 6 (Nov 9) Comments an d reviews received fr om the TEM an d PEM co mmu nities will be ev alu ated an d, where ap pro priate, be incor por ated into the pr opos ed stand ard . The stand ard will then be su bmitted to the Res our ces In ven tor y Committee fo r accep tan ce. 11