Full Article - Notornis and Birds New Zealand
Contents
1. Calculate Distance Time Interval and Speeds i dt R idea P ompare the present record with the previous unexcluded record Set thresholds for Flags Detect Flagged records Is the Flag Too fast Is the Flag Too soon H j IFYES Using LC IQ location Distance Speed IF speed reasonable in context IFYES see note and Flags exclude poor records THEN retain both records IF both records are LC gt 0 retain both See APPENDIX E ELSE exclude poorer record IF both records are equal to or better than LC IQ 0 44 and speed lt speed threshold and Time Interval gt 0 2 h Continue selection until no TEN EK both M o n i exclude poorer of two records excessive speeds remain IF LC of poorer record 0 lt 44 or A B Z Is there uncharacteristic THEN exclude this record Or excessive zigzagging inthe lt lt z S lat And Or Jon 2 Check no excessive speeds remain by plotting speed against time Is the LC A B Z IFYES I Consider excluding poorer record IF YES Review selection thresholds Consider excluding poorer record Weg d Map records and investigate on land and Set Reason see APPENDIX C for each excluded record any other improbable records Inspect recalculated Revised Time Interval Distance and Speed IF a record two or more records away from the present has now been excluded THEN reset Reas2. O Selected locations Excluded locations 200m Fig 3 Demonstration of selected and excluded Argos locations of a northern royal albatross Diomedea sanfordi moving along the coast of Chile achieved using selection procedure set out in the text PTTs of differing duty cycles and repetition rates are compared in Table 2 The most notable feature is the poorer quality of the material received from the sooty shearwater Nicholls et al 2007 However an explanation may be that the shearwater often dives and it has a less stable flight pattern in comparison to that of the larger albatrosses both of which would make receipt of its transmissions more difficult Table 1 and Fig 2 illustrate an example of our selection process applied to the tabulated spreadsheet information reasons for exclusions are shown The original and selected data are mapped in Fig 2 to show the results of the selection process 128 Nicholls amp Robertson Selected locations MM ic 3 281 BM c a Hl c o LC B amp Z Fig 4 Demonstration of selected Argos locations of a northern royal albatross Diomedea sanfordi moving along the coast of Chile see Fig 3 achieved using the selection procedure set out in the text Each selected location is surrounded by a circular buffer see text to represent the possible precision according to Location Class LC of that location point The locations of a northern royal albatross off Chile in comparison to the
3. x pI0231 ay IO U SOY SI Z UOYLIOT Jt 6 pue z u A1 q MPLA L 0 pasueyp 1 T enrur PZ e nuriod suonco o ay dew o eyo 92X4 KX ue ur q un 8o1 p sn vv pey snotaaid y pue nit Z Jo 2 8V TOD EM eNO p d j i d y st ZNO 1 J ss 2oid uono s Suump pasueyp aq Le Aremu 1NOT Jo Adoa FA e nurrol Daiutaiaid SI gZLV 1 Jt ssa00id uono s Suump pasueyp ad e AED 11Vy 1Jo Adoa Pia CNOT WOU amoge jeu z sn p r ddns jou emunoyg r qumu etutD p e se TNO Dia IV I wos le no e2 1snuu z sn parddns jou emwoyg 1 qummu eurD p e se ZIV 1 Dia INOT Woy amoge 1snuu Jost p r ddns jou e nurroT r udsruu r us 1s AA ut s n 8A 2AL Z N aquinu eurD p L se NOT aseqeyep eu1 1x 10 J uspe ds ur play LIN Wor le n e2 sn 3 sn p r ddns zou e nurroT r udsruu H uI ymog ur s n eA AneS N 1 qumu eurD p e se LVI DSIN AIWO E C6L Cp 90L OLT COL SP 610 0Z1 660 Cr 9L OLT lt 6Z Str 1X9 UP g 3X9 UD T 1X0 LP F I qumu da0 Jaquinu ddl Jaquinu dd 0 Ja8ayut Jaquinu gq jaquinu gq Jaquinu gq Jaquinu gq Jaquinu gq Jaquinu gq Jaquinu gq WeN x s 207 dds TIT p ds H Pagu aunty H a0uejsIq H NOSVia S Ado 2 HS UOT FT S SCHO S LV I S INOT S LLV 1 S IV HV OV dV HV dv OV adv VV A panuyuoy q xIpueddy 133 Valida
4. that summarise the variables affecting the Argos location calculation highest values indicate better transmitter performance c Plausibility Tests Passed NOPC grades the reliability of the record using tests for residual error transmission frequency continuity movement of the transmitter and plausible velocity from previous position Anon 1999 4 tests passed best lt 2 invalid The CLS Argos system provides a variety of data file formats and if requested can include the provision of Location Service Plus This auxiliary service provides all of their determinations of location not just those validated locations of high quality that are normally provided COM or PRV files This Location Service Plus facility is essential for researchers using low power PTTs depending on the economics and requirements of their study For each location determination by the system there are 2 possible solutions 1 on each side of the ground track of a satellite s orbit CLS Argos FAQ The user of a DIAG or PRV C file is responsible for selecting the correct solution from these 2 location sets The locations provided differ significantly in their accuracy depending on the quality of the transmission and its reception The Argos definition of the accuracy for the position ranges from lt 1 km 122 Nicholls amp Robertson H to indeterminate and unspecified The system provides diagnostic information e g IQ to help r
5. 1 LC at a time beginning with the least accurate LC Z progressing through to LC 3 2 amp 1 Initially with the full dataset Fig 1A there were speeds of gt 200 km h7 but as the arbitrary LC locations were removed the number of excessive speeds declined until there were none gt 50 km h At this stage only 26 of the original dataset only records with LC 3 2 1 Fig 1B remained b We applied the visual selection procedure described here to the same dataset and plotted the resulting speeds Fig 1C This process retained 94 2 of the original dataset and the maximum speed was reduced to 85 km h RESULTS The 2 procedures yielded broadly similar flight patterns The speeds waxed and waned over the deployment with bursts of high speeds sustained over several days as the bird migrated across the southern Pacific Ocean Nicholls amp Robertson 2007 However arbitrarily and progressively excluding locations with LC values of Z B A and 0 ultimately removed some 74 of the records whereas the visual procedure excluded 5 8 from the same dataset It is important to note that the range of speeds is significantly higher for the visual selection process As speed is a factor of time interval the fewer records from the arbitrary selection obviously ensure the resultant artefact of lower speeds The quality of the records received from CLS Argos over deployments for 3 species with Validating satellite locations 127
6. and after the target deployment Typically locations obtained during testing or preparation 9 Records with no location determined by Argos Generally LC Z and quality index 0 or 1 Records NOU WN L Records overland and rejected Appendix D Flow chart demonstrating the selection Appendix E Flow chart demonstrating the selection procedures for validating CLS Argos records exclusion decisions needed for validating CLS Argos explained in text DIAG locations demonstrated in this paper Note that Nicholls et al 2007 have noted a greater precision C gt D for locations with LC A as against those with LC 0 so the process may need to be modified to take this Convert ARGOS archival DIAG file into tab delimited text file into account depending on the precision required by the selector Add Unique ee ae Add line C D Add deployment details bird ID Mark pre amp post deployment records For ALL records but concentrating on the flagged records For each record confirm first location is the better of the two locations Sort by latitude Exclude Z IF NOT Resort by PTT Date or line Compare location with previous two and next two records and select a location from each pair giving the shortest distances between records IF second is location selected set Reason 2
7. to the bottom of the sheet e Records without a location LC Z latitude sorting and moving them to the bottom of the spreadsheet f A numeric field Reason with the value of 1 is added to all records indicating that the 1st location is selected The time interval distance and speed from the previous to the present position are calculated using great circle distance Appendix B g A calculated field Flag detects records that need visual checking because the speed is too high or the time interval between records is very short h Further descriptive character fields for the identification of the animal Species Location Sex Band Number Name are added according to need See Appendix B for the file specification and spreadsheet formulae used for the various calculations These steps may be achieved within a database where the records can also be identified by a unique number stored sorted and selections exported e g Prince et al 1992 We imported a tab delimited text file into a database calculated some fields and then exported into a pre constructed Microsoft Excel template spreadsheet The formulae of the calculated fields within the spreadsheet are applied to the entire and pre and post deployment records All criteria for the selection can be displayed in consecutive columns to facilitate visual comparison Originally we developed an entirely visual selection process but subsequent
8. 0 Speed kph 40 60 80 100 120 140 160 Days from 1 Jan 1998 Speed kph Days from 1 Jan 1998 Fig 1 Timelines of unselected flight speeds A comparing the effects of arbitrary removal of the poorest Location Classes LC 0 A B Z B and the selection of records using the process set out in the text C location for an albatross after it had left the breeding location rechecked A flow chart an example and a decision table of the entire procedure are provided in Appendices D F and Table 1 We used the deployment on a Chatham albatross that foraged at 2 locations and migrated across an ocean to illustrate a comparison between the 2 procedures a progressive arbitrary removal of the poorest LC of locations b our visual selection procedure for detecting out lying points in seabird flights determined from CLS Argos location files The characteristics of the dataset used are shown in Table 2 which shows the initial removal of the no location LC Z records before the selection processes began 177 176 8 176 6 176 4 176 2 176 175 8 175 6 44 2 44 4 B 446 E T S 448 45 45 2 Longitude 477 176 8 176 6 176 4 176 2 176 175 8 175 6 Latitude Longitude Fig 2 Mapped results of removing Argos location points selected using the process set out in the text see data in Table 1 a We excluded
9. 00 S CSI FP 090 9ZT I ZT L 69 rt 10 9Z1 ZS T I ZT I I CO9FP GCUO t I SC ce 897 10 9Z1 OS T Iv ZT A0 L 19911 ZEOLIT OLD 100d aytds 8 See ye mq Seu ou popnpxy S0 V v TCC T8 y Iert ZENIT S ZT 8 ev TY FCO OS T S ZI 8 IL PCP LET OLT y ZC 6 IC Ft 80 9LT 00 4 0c TZ Iv P PICT 8L79LT DULISIP Iva YIM EEN e S pesa ayids Sur ajou p pnpxq 00 4 I 0S TO Iv S CVC rr S6ZSZ L 0 SCC 0 Wty yC9ZL ZS 0 0 NC 0 L IV P7 9CC9ZL G TI Wr PEILI 8S T Pl I LEV PE 6C C9ZL IL CO YP Ft SC 9ZI 87 0 uoos ool TI CO L YVVYV O8C9ZL 9 0 4 YV Ft SC 9OLT 8S T 9 0 c L Ser rr Src 9ZI E Cc EFt TEOLT 00 4 80 L eerti C C9ZL c 80 lt eV tY GC9OLT 8S 7 EU c I WTP S8VC9ZL p ds TEAU OWI DULJSTP p stA suonisod p p ps OLDI Bejg pasdg au sourjsiq uose y yey suo apnyzyey usy pnir8uo ur suontsod lt y wy ut p ds YUL owy up UT saduR sIp Zapp We apnySuo BUON PIO JOLY JO POURI 8661 AON SL 9661 AON II TETTE Lid yueutAoydap vpust ayouvssvpoy ssomeqye ureyyeyD wo eq s nse1 paddew 103 z 211 os e s 4x9 ay ur p qu s p ssad0id uono s au leoIpur 0 SP1OIII UOHLIOJ Ld JO uono s payejouuy T lqer Validating satellite locations 125 Table 2 Examples of satellite tracking locations comparing the quality of records received from the CLS Argos system for different species for different PTT duty cycles and different repetition
10. Argos but as the off period in the transmission regime lengthens alternate locations are increasingly likely With 3 and sometimes 4 satellites available there are a few occasions when the consecutive locations are simultaneous exceptionally or only seconds apart Over such short time periods any error in the location creates impossibly high speeds We selected 0 3 h 18 min as the time interval to warn for this condition Stepping through the Excel file each present record being considered is compared with the previous selected record Initially all records are selected but as records resulting in excessive speed or distance are excluded each comparison is made 124 870 c YV Ft TEOLT Oct 80 c L SWT C9ZL 0 OC L YV YV GCO t OS T 0 Ka I I cerry ZVC9ZL OT S YV Ft EC 9LI OC LV I 87 I LEV tr TET OLT aide Suo Ayyenb 100d pue zrod I A acer pa ae 00 4 eptont H 90 oF S LY 688 9ZL E CU S Z7 9TOLZT Oe MOOS oog SI 0 S I estt O09 9ZI c 60 lt V TY PEILI 4S T 60 lt IL CCHFt che OLT 8T S I AC ZPP TEILT SS I 81 STI LC L Ec tt CTT OLT VCC IC CO Ft PO OT 0S 0 uoos ool TOIT T0 UC I ICT WOLT p109 1 1x u das Dapp 00 4 lt 9S S 008th IFC9ZL Aarenb 100d pio2 EE i i e sno uely nurs papnpxq 00 4 IL S 00 91 S ZAC C 96 SZ I EI ZT AC Ott L9ZT 87 0 tL OT C L SSY tt VEL 9ZT s lr 1es i iama p IA Z 4sey 00 p pn oxq 95 0 L TOT
11. F 1x9 DI SHA a Z 1o g V 0 1 TE s n eA Sse 92 uoneoo 0 YO Su oO da ssw HH AN WIG leq 98 91 90 66 20 90 1X9 P ST aea SHA O so81y STO Aq parddns LL4 Jo uone gnu prt nbiun T80 Z oam Lld SHA 8 Sp1OdaI 0 Surr J Sunaos pre 0 pray Ieuondo Iz 1a3a UI oul v s aoujd putto p aq u011292 9s Su1unp u ppni Ajwsauas H ss 3204d uoysajas sof jvijuassa S INJ DVI sosupy WOU owp HA x emunoy uoneue dx pue uonduos q n eA eotd Ay d L ureN PPH 109 1x 1 UL p ure dx spio2 JO uon s 107 ainpasdoid y 107 pasn jaayspeaids aoxq ue 104 suoneogro ds uumo2 pue sly g xipu ddy Nicholls amp Robertson 132 u ur o d p 10 pue eurrue 103 aureuyoTu JO sey x s IOJ poO ups yu ur o d p Jo uonooj ay pue s p ds ay 103 apOd Aaya NOJ LLd Mau e 10 qQu ur o d p mau e Jo uoneoo ISI 107 0 J3 sur Ayjenueyw bZ N PA PAV at esau urr1 5ue stq euog uoneo2o 1Ju s id y 0 uonsoo snorA 1d y woy ydy ut p ds FT RINUIIO LL Mau 10 qu ur old p Mau Jo uonc2oo ISI 107 0 Masur Aypenueyy pao2 1 yuasaid pue snoraaid usamyeq sinoy ur IWI L LLJ Mau qu ur old p Mau Jo uonoo ISI 10 0 Mast ATTenueyy O8T OId Z SOD 08T IdsFZ SOD 08T OIds W V V V SOD 08T OId Z NIS 08T OId 7Z NIS SODV 8790 9969 muog yuasaid ou 0 uoe snotaaid woy wy ur a0ue sTp pI SS T SIDOT gt xtpueddy sapod aag p pn
12. I pio5 q S L s1940 pue Apa yo aseyjoa Arayeq amprodwa 3a SULINSeaU z 0 sI qunu no art sIaquimnu yq g oz ATTensqQ laer 1 d lur PP SHA S aInpRJnueU L Ld pue nu o s s N sosry s I 1939 UI ep SHA a payeulioy pue pamn3yuoo ATsnorea are eyep pue siosu s f 1 8 lur zp SHA re sIOSUdS W017 JUSWaINsesUt au Modat sptatt M047 1X N GZ 1939 uI Ip SHA d ZHI ur LL ou Jo Aouanbayy orpes p le no e 2 so8rv 9 1F08S910F jaquinu gq I Aouanbary SHA o Kail asuey pio32 Jo Ayyiqerpas o pmn3 v passed s s Atttdtstppid Jo aquiny z 1 d lur DdON SHA N spUodas ur s 8ess u jet pue SIT UsaMjaq uut ZL 1939 uI uoyemqgsseg SHA W eam EET 0 Suons SZI SI SLLd Did Woy unSu ns asuei Jeord4 gp spun ssed y ut eusis 1s S3uomns y Jo unSu ns eUsIS 8ZI 1a3a UI eusis 1s q SHA 1 P OTT Uey 1 1e 3 u18u ms eUSIS e U1LA soS1v Aq p At 2 1 SasessauU Jo JaquInN 0 1 8 lur Suoms SHA x g 1 BUL ssed ay ut so3rv Aq paatadar sasessaul Jo Jaquiny f 1 d lur sasessoul SHA sol iv Aq p le no e2 uoyrsod puov s ay Jo pnir8uo AA610 0Z1 3X9 ZNO SHA I so8ry Aq p le n e3 uontsod puooas ay Jo apnyyeT S660 Zr 1x3 zivi SHA H so8ry q p le n e3 uontsod Te y Jo pnit uo A9PLOLL PO INOT SHA 5 so8ry Aq p le n e3 uontsod e ay Jo apnyyeT St6L SP 7X2 LLV1 SHA 1 Ajaatpadsar sassed aqTTJoyes u A1 q pue umnprA soUeUTIOFJIEd I Ld SUIqLOsap siaquinu apoo OMI O
13. Name Species Sex Optional Copy formulae for all fields marked as S in Excel spreadsheet for all records Appendix B Set speed threshold in formula for warning flag eg suggested Diomedea 60 km Ir Thalassarche 40 km bi Set short time interval value in formula for warning flag eg suggested Time Interval 0 3 h Copy formulae from template down the Excel spreadsheet for all records Initialise first record for each PTT animal Distance 0 Time Interval 0 Speed 0 and Set Reason 1 The Spreadsheet automatically calculates and inserts Distance Time Interval and Speed to each new point with warning flags to mark records that are improbably fast or too close together in time For ALL records consider the locations lat and long Concentrate on the flagged records IE the wrong mirror image location lat and lon is given by CLS Argos THEN select the better point from LAT2 LON2 in the context of surrounding records THEN copy to lat and long THEN set Reason 2 ELSE proceed to next step Indication distance for consecutive records is very large and dissimilar compared to adjacent points This process of comparison of the present record with the previous selected record continues throughout the file If the previous record is marked for exclusion then the selection comparison process must backtrack to the last earlier selected record and the comparison repeated As a record is marked for excl
14. Notornis 2007 Vol 54 121 136 121 0029 4470 The Ornithological Society of New Zealand Inc Validating locations from CLS Argos satellite telemetry D G NICHOLLS 31 Northcote Road Armadale Victoria 3143 Australia C J R ROBERTSON P O Box 12397 Wellington 6144 New Zealand cjrr wildpress org Abstract Satellite tracking with the CLS Argos system has provided enormous benefits to wildlife studies especially for oceanic bird species The system provides 2 locations 1 from each side of the satellite orbit but they are irregular over time and of variable accuracy Procedures are described here to identify outlier locations and retain the maximum number of valid observations from DIAG files thus producing a more homogeneous data set from which to map distributions track movements and investigate behaviour while determining the rate and direction of travel Nicholls D G Robertson C J R 2007 Validating locations from CLS Argos satellite telemetry Notornis 54 3 121 136 Keywords Argos telemetry animal tracking oceanic birds flight speed albatross shearwater bird distribution INTRODUCTION Satellite tracking using the CLS Argos system has provided enormous benefits to wildlife studies especially for oceanic bird species The system links alocation and data collection receiver aboard NOAA satellites and miniature low powered transmitters platform transmitter terminals PTT attached to wild animals It use
15. coastline and bathymetry are plotted in Fig 3 This species is not known to fly over land after it leaves the breeding area All except 1 of the improbable over land locations were identified for exclusion by the speed quality criteria of the selection process After the selection process the refined at sea locations showed a strong relationship to the bathymetry The selected records shownin Fig 3 areexamined in more detail in Fig 4 in which the quality LC of each of the selected records is surrounded by a buffer defined by Nicholls et al 2007 at a distance of lt 2 5 km for records of LC 3 2 or 1 15 km for LC A and 25 km for LC 0 Location records with LC B Z were allocated an arbitrary error distance of 56 km The levels of quality give another indication of information lost if various classes of record are arbitrarily removed as in Fig 1 The selection processes identified the poorer Locality Classes as the location points most commonly associated with improbable or excessive speed However some of each of the poor quality LCs need not be excluded even for the poorest LC Locations of LC Z B and A are often single points deviating only a short distance from the trend of direct flights and if this poor quality coincided with a long time interval a normal flight speed resulted This situation existed for LC 0 observations as well although less frequently Where a high speed was flagged and the consec
16. esearchers interpret the accuracy of each location determination The solutions developed by previous workers that are used presently by researchers to extract as much valid location data as possible include using only locations of known accuracy e g LC 3 2 1 to establish the locations used to map the routes taken by albatrosses Diomedeiidae Poorer class records were clustered along the lines demonstrated by the quality records Weimerskirch et al 1992 Filtering procedures were perfected for slower moving species using a single maximum speed of travel 1st for seals McConnell et al 1992 and later other speeds were used for studies on penguin and albatross Brothers et al 1998 Klomp amp Schultz 1998 used 2 different speed thresholds for the short and long time intervals between successive locations to select points for their studies of shearwaters Other workers Anderson et al 1998 have arbitrarily discarded the poorest accuracy Locality Classes e g Z An alternative method Nicholls et al 1994 used arunning average of 3 points weighted for the LC and the current and adjacent locations later allowing for the difference in time between the 3 locations Freeman et al 1997 This method used all the locations provided and generated a possible flight path smoothed between the location points As our pelagic seabird tracking and validation experiments 1992 2001 proceeded we progressively lengthened the in
17. eviously excluded records Speeds greater than the predetermined mark set by the researcher e g 60 km h for northern royal albatross are considered particularly carefully The quality of the record the speed before and after and the trend in the rate of change of latitude and longitude are considered when determining which of the 2 locations has the poorer quality If the speeds adjacent to the interval being considered are high or increasing and the quality LC and IQ of the records is good the records are retained Where 1 or other of the poorer quality locations has an LC of A B or Z then 1 is generally marked for exclusion When 2 records for comparison both have LC 0 the speeds to nearby locations are considered and the quality indices IQ compared If the speeds of the adjacent records are similar the high speed is accepted and both locations are retained However if the speed is greater than 40 km h depending on species and greater than twice either of the previous or the next records 1 is usually excluded If both records are poor i e LC A B or Z the poorer record is excluded Appendices D P The selected dataset locations should then be mapped and any locations over land other than breeding sites of pelagic seabirds an improbable Nicholls amp Robertson 126 Speed kph 40 60 80 100 120 140 160 Days from 1 Jan 1998 200 160 120 8
18. h and eventually selected 60 km h as the warning flag threshold for 3 species of Diomedea and 50 km h for 1 Thalassarche albatross Setting the threshold too high can include probable outlying points in the final data set setting it too low can result in too many flagged points having to be considered and valid locations being wrongly excluded Selection To aid inconsistently finding possible outliers see Appendices B F Table 1 each record is automatically flagged if the speed exceeds a predetermined value suchas 60kmh fornorthern royal albatross Diomedea sanfordi 50 km h for Chatham albatross Thalassarche eremita and 40 km h for sooty shearwater Puffinus griseus or the time since the previous record is lt 0 3 h The speeds and time are chosen as a warning they are not in themselves sufficient reason for marking a record for exclusion Each flagged observation is first checked to determine if the alternate location is the more likely location position Excessive speed e g gt 200 km h and a large distance immediately followed by a record with a similar distance are clues that the alternate location may be preferable Location 2 is then substituted as the selected location and time the Reason is marked as 2 and a revised distance and speed are automatically recalculated Where the transmitters are operated continuously the correct location is almost invariably selected by CLS
19. ly developed the automatic flagging procedures shown in Appendices B F and in Table 1 In this process each record is considered in relation to the locations immediately before and after that record When determining if a record is valid we consider LC IQ time interval between locations speed and its rate of change along with the trends before and after in latitude and longitude We experimented by excluding suspect locations and then comparing the changes in the time intervals distances and speeds that were recalculated when changes are made in the Reason field The results of any change could be seen immediately in the revised distances times and speeds Appendix B The Reason for the change is recorded e g if the 2nd location was the more Validating satellite locations 123 likely correct location or the record was improbable because of poor LC and IQ using the classification scheme and guides in Appendices C F After the 1st pass through the file marked records are excluded other than those where the 2nd locations were used Reason 2 and the process is repeated for any flagged records that remain Others have independently measured or used the occurrence of similar speeds to determine arbitrary selection Weimerskirch et al 1992 Brothers et al 1998 After inspecting the frequency distribution of our calculated point to point speeds we explored different maximum speed thresholds of 40 120 km
20. n Appendix F Continued For the present and the nearest previous unmarked location IF both records are LC 3 2 or 1 THEN retain both locations unless records are almost simultaneous and result in excessive speed IF the LC of the two records are unequal see Note in Caption THEN Mark the poorer of these two records where the poorer record is the record lower in the order of LC 0 A B Z from good to poor set Reason 3 4 5 and 6 respectively for poorer record APPENDIX C IF both records LC are equal compare first component of IQ THEN mark the record with the lower component Use Reason 3 6 as above IF records LC and first component of IQ are equal compare IQ s 2nd component THEN mark the record having the lower IO component Use Reason codes as above IF records LC and both IQ components are equal THEN mark the record having the greater speed or distance deviation or the present record Use Reason codes as above Subroutine B A test for excessive speed in relation to adjacent locations Seabirds do not fly continuously nor do they fly at a uniform speed Speed needs to be judged in relation to the context of each record Criteria considered should include IF the speed to the present record is gt 40 km Da AND greater than twice the speed in the previous record and the next record THEN SPEED EXCESSIVE IF the speed to the present record is gt 40 km ht AND less than
21. on s of intervening records and reconsider next record from last excluded MAP Set Reason for these excluded records SUMMARISE and Consider NEXT record UNTIL end of file ANALYSE selected records See pseudo code for a definition of excessive speed in Appendix F and text See pseudo code for subroutine in Appendix F to determine poorer record Validating satellite locations 135 Appendix F Listing of pseudo code for the selection procedure with 2 subroutines for validating CLS Argos DIAG locations explained in textr Note that Nicholls et al 2007 have noted a greater precision for locations with LC A as against those with LC 0 so the process may need to be modified to take this into account depending on the precision required by the selector The investigator before beginning the selection process sets Instructions in italics Convert CLS Argos DIAG archive file into tab delimited text file Create a template Excel spreadsheet with the layout according to Appendix B Import tab delimited file into the template Excel spreadsheet Sort data by PTT and date time chronology Label each record sequentially with Line Select and mark all pre and post deployments for each PTT animal Set Reason 8 Select and mark records without a location ascending sort by latitude Set Reason 9 Sort ascending retained records on field Line Identify each bird deployment with
22. probable speed records Review selections or thresholds if excessive speeds remain At the end of the selection sort Ascending the file on the Reason field Separate the records with Reason gt 2 These are now the excluded records In a spreadsheet this separation is easily done by inserting blank new rows and annotating a row ahead of the excluded records Re sort the selected records in chronological order use line as a simple quick solution At the end of the selection again plot speed of selected records against Time Interval to confirm there are no improbable speed records Map retained locations Mark consider and remove improbable locations that are overland provided the bird is not near its nest Mark any records that do not belong to the dataset set Reason 7 Very occasionally CLS Argos radio reception wrongly identifies the PTT and allocates PTT locations from another program to your dataset but typically these are single message records mostly Z Occasionally PTTs with the same PTT ID are apparently being briefly tested by a manufacturer and will be wrongly attributed to your program They are usually identified in the selection process because of an exceptional deviation or they appear as a discontinuity in list sorted by location e g latitude Set Reason 7 Subroutine A A routine for determining the poorer record for each of previous and present locations 136 Nicholls amp Robertso
23. ra carriage returns from the multi line record Aword processor or editor allows the global selection and replacement of the field names e g Date Lat 1 with a tab and the carriage return between lines by replacing the carriage return and the 1st field name of the next line with a tab This process can of course be automated by use of a macro not described here Others have written customised programs McConnell et al 1992 Brothers et al 1998 We then stored the data in a Excel database arranged to calculate some fields Julian day numeric decimal versions of LAT1 LON1 LAT2 LON2 and these calculation formulae have not been included in the procedures described here Preparing the records Appendix B for selection a Each record is labelled with a unique serial number b Locations from text fields of geographic coordinates e g 30 5 N or 55 S must be converted into numeric values where the northernand southern latitudes are respectively positive and negative 30 5 55 0 respectively and eastern and western longitudes e g 170 E or 175 W respectively changed to positive and negative numeric values 170 0 175 0 respectively The latitudes and longitudes for each of the 2 possible Argos locations are converted into numbers and a copy made of the 1st nominated location Appendix B c Calculate the Julian day from DateTime d Markandremovethe pre and post deployment records
24. rates of transmissions Records accepted rejected records accepted or rejected during selection process during selection process Percentage distribution percentage distribution for number of messages received satellite pass for each PTT Diomedea sanfordi Thalassarche eremita Puffinus griseus PTT ID deployment 23738 2 23081 6750 4 Location start 43 5988 176 861W 44 4538 176 315W 45 534E171 010E Location finish 34 7515 113 846E 6 3965 81 323W 42 930S 179 114E Date start 28 Jan 1997 17 Feb 1997 13 Oct 1999 Date finish 15 Aug 1998 7 Jun 1997 29 Oct 1999 Active deployment days 564 110 16 Duty cycle h 9 on 135 off 3 on 3 off Continuous Repetition rate s 90 77 85 Total records received 577 480 189 Records accepted rejected 461 116 404 76 103 86 accepted rejected Location class 3 0 7 1 0 1 9 2 5 0 4 0 0 2 1 9 1 16 3 0 2 17 9 8 5 0 45 4 3 6 41 9 1 7 27 0 5 3 A 5 7 2 1 9 8 0 6 13 2 5 3 B 5 0 1 6 8 3 1 5 18 0 9 0 Z 1 6 1 4 1 3 1 3 Z 11 3 10 6 25 9 Percentage distribution 12 0 2 11 1 7 10 0 5 4 4 9 3 8 0 2 7 3 8 8 8 1 0 9 4 0 8 7 14 2 1 6 9 2 0 6 3 7 0 5 6 13 3 1 9 10 4 0 6 7 9 0 5 5 13 7 1 4 10 2 0 4 11 6 2 1 4 14 6 1 2 12 9 0 6 14 3 3 2 3 5 9 2 6 10 2 1 9 7 9 5 8 2 5 0 1 9 8 3 1 5 9 0 10 6 1 8 3 9 4 22 7 between the present and the last selected record If this earlier record needs to be excluded then the comparison begins again after including the intervening but pr
25. ring based on a single speed threshold have the advantage of simplicity and reduced processing time but we consider the loss and simplification of the original data to be too severe In the processing procedure reported here it should be emphasised that although records are removed the time intervals distances and speeds are recalculated within the revised set of records The sum of the time intervals between records therefore equals the time between the 1st and last records The sum of the distances is progressively reduced as the number of records is reduced and the exceptional speeds are eliminated A chain of records is maintained and removing poorer combinations provides slightly different representations of the flight progressively straightening therepresentation of the flight as more and more records are not used The final representation should meet the criterion that there are no biologically impossible points Our close inspection of the pattern of locations indicated that the poorer quality records of LC 0 A B and Z generally result from poor satellite reception However there are other instances where the reception was good strong signal strength many messages but the signal was unstable One interpretation is that a series of poor locations result when a bird flies fast with many changes in direction e g when it is foraging over a small area Such flights could affect the Doppler shift measured by the CLS Argo
26. rocedure may assist other researchers in improving the use of their satellite tracking data for similar studies ACKNOWLEDGEMENTS Satellite tracking experiments producing the data for developing this technique were funded by the New Zealand Department of Conservation the Australian Research Council the Ian Potter Foundation the World Wide Fund for Nature Australia Environment Australia the W V Scott Estate and private donors La Trobe University supported many of these tracking studies for more than 20 years G P Elliott and M Harger provided assistance with the great circle calculations M D Murray provided critical advice throughout the study Southlight farm and S McGrouther generously provided the facility and accommodation on the Otago Peninsula where much of the process was refined over 3 summers 1998 2001 Chisholm Institute has generously supported the salary of D G Nicholls while CIR Robertson was employed by the New Zealand Department of Conservation until May 2000 and by Wild Press subsequently Without the support and encouragement of these organisations this work would not have been completed LITERATURE CITED Anon 1994 CLS New Argos location News Flash CLS Toulouse France July 1994 Anon 1999 The Argos user s manual Maryland USA Argos CLS Landover Appendix A Typical CLS Argos DIAG file record Program 0137 Anderson D J Schwandt A J Douglas H 1998 Foraging ranges of waved alba
27. s receiver and the resulting location would be calculated incorrectly CLS Argos FAQ Nicholls et al 2007 An apparent flight path of 20 100 km point to point zig zags must not be taken literally The location measurements are made during the 3 16 min of a satellite pass during which an albatross could indeed have turned 180 or it could have flown more than 10 20 km a distance considerably greater than the specified accuracy of LC 3 2 or 1 If these flights also include a landing on the sea and 130 Nicholls amp Robertson submergence with associated temperature shocks to the transmitter it would cause an unstable signal that would distort the Doppler calculation Inspection of the full diagnostic information DIAG file provided from CLS Argos during the consideration of the calculated distance and speed produced a homogeneous set of locations There were no excessive speeds between successive locations The visual selection procedure described here retained a higher proportion of locations than other procedures already described without compromising the validity of the tracking The selection procedure we have applied to CLS Argos location data has been tested against tracking studies for 2 genera of albatrosses and 2 shearwaters Puffinus spp The consistent identification and exclusion of improbable locations helps in relating a bird s position and movements to atmospheric and oceanographic features We believe the p
28. s the Doppler effect i e the change of a PTT s radio frequency to determine the location The satellites are in polar orbit each passing overhead every c 3 h and up to 4 satellites are available to locate any PTT tagged animal The satellite orbits are not equally spaced relative to each another and so the resulting locations are not spaced regularly throughout the day The frequency of reported locations is highest at high latitudes and decreases with latitudes approaching the Equator Up to 20 location records of variable positional accuracy Anon 1999 CLS Argos FAQ day can be obtained from a tracked animal The challenge for the researcher is to retain the maximum number of observations after using explicit and objective criteria to eliminate outliers points to produce a homogeneous data set from which to map the animal s distribution trace movements or investigate behaviour and to determine the rate of travel CLS Argos grades the location records according to 3 criteria according to the CLS Argos Argos assessment of the animal being tracked e g Received 2 Aug 2006 accepted 31 March 2007 Corresponding author speed of movement Anon 1999 a Location Class LC indicates accuracy LC 3 2 1 have an accuracy of lt 1 km LC 0 gt 1 km and A B Z each have an unspecified accuracy b Quality Index IQ has 2 components based on frequency and signal stability number and timing of the received signals
29. tended duration of deployments up to gt 2 years for some deployments and the transmitters were operated intermittently with duty cycles of 3 23 h on and 3 160 h off to prolong battery life Each transmitter had a pulsed signal with a repetition rate of 60 90 s that operated during the on periods The error of individual locations has been measured for stationary transmitters Weimerskirch et al 1992 Anderson et al 1998 Brothers et al 1998 Nicholls et al 2007 but it was previously suggested to be 10 km for birds at sea Prince et al 1992 More recently the development of improved satellite receivers smaller PTTs revised and additional since mid 1994 LCs 0 A B Z with anew method of calculating the location by CLS Argos Anon 1994 have all improved the quality of positions obtained from Argos The accuracy of locations have also been measured for moving transmitters on small ships a car circling a proving circuit at variable speeds and on a transcontinental train Nicholls et al 2007 Between 1998 and 2001 we analysed albatross and shearwater flight data using DIAG files collected from latitudes between the Equator and the Antarctic using various PTTs over a wide range of transmitting regimes and with 2 4 satellites determining locations a combination of factors not generally applied in tracking programs We considered it was desirable to judge each location in its context rather than to appl
30. ter sustained speeds were observed For example few useable data would remain for the sooty shearwater Table 2 The data give a consistent impression that an increased number of improbable points are observed after a substantial change of direction at the Ist point of a long flight after local foraging and with the Ist location point after a long period of no transmissions Nicholls et al 2007 A mixture of locations of LC 0 IQ 40 46 and an occasional LC 1 location may indicate that a bird was resting on the sea or proceeding in a series of slow local flights Likewise a set of locations of LC 0 A Validating satellite locations 129 B or Z may occur when the bird is flying fast or quartering a local area DISCUSSION Researchers have used various methods to select records from CLS Argos datasets Of these methods progressively removing the poorest Locality Classes did not remove improbable points until only the best LC 3 2 1 remained The residue is usually a small minority of the original records The time intervals between the few remaining locations are thus usually relatively long and the resultant flying speeds are biased and certainly under estimated As changes in the flight speeds vary widely depending on the time interval a single or limited number of threshold speeds for determining improbable speeds are also insensitive criteria Simple exclusion rules setting aside the poorest locations or filte
31. ting satellite locations suon s UOTalaI 10 uonu l euond ox 104 uoneue dxg PLV PSV E gt POVG AI emuuol ueq are spio32 1 papnypoxy DIS NOSV 4 ur s 8ueup Aq p 1 S83u ATUO SUOTLIOT p 123 s 10 LAIOJU SUIT AdY pue 52ue sIG ADY WO p le no e32 10 pomo ATTeuoytpuod ATTeoyeurone p ds pastaayy CLV PAV PAV E gt COVS AI MUO PIY NOSVAA au ut asueyp e Aq Data UOTeIOT p l5 s JSEJ AY WOI STeATAJUT SUIT Axne nurn2 woy poeme T e neuurolne LAIAJU BUILT POSAY O8T OIds OV SOD 08T OIdxFOV SOD 08 0 Ids dV ydV SOO 081 014 OV NIS 081 14 FOV NIS SOOV 8F90 99 9 emwIog prey p le no eO PIY NOSV 4 au ur a8ueyp e Ad p s 383rn uoneoo p l 2 s 1se pue sty Woy p le n o e2 1 T eoeuuol ne 52uelstGq COV PZ E gt POV A eInurroq Play st ot pardo st play sty Jo pao3 snorA 1d y 2 p 13 s 1se u y z lt uose y PIODAI p pnpx ue J Dia Sty 0 pardo st 21 ay prOdaI paydaooe ue FI play p le n eo EdV PVV E gt POV I eImurol pat sty olur pardoa st pray sty Jo poder snotaard ay uo p lo l s 13se Udy z lt UOSedY PIOIAI papnyoxo Up J Dia sty 0 patdoo st ot au p1o32 paydaooe up It pay POPOP sts eue 10 Surddeur ut paqe e se pase 7 P FO CIN AN TVA e nuoq py 2q Woy yyuour BUYLI play po RMPI ta yG 1eu leouoO Hmo made Surddew e se pue ssad01d uono j s
32. trosses in the eastern tropical Pacific Ocean pp 180 185 In Robertson G Gales R ed Albatross biology and conservation Chipping Norton Australia Surrey Beatty amp Sons Brothers N Gales R Hedd A Robertson G 1998 Foraging movements of the shy albatross Diomedea cauta breeding in Australia implications for interaction with longline fisheries Ibis 140 446 457 CLS Argos FAQ _ http ftp cls fr himl argos general fag_ fr html Freeman A N D Nicholls D G Wilson K J Bartle J A 1997 Radio and satellite tracking Westland petrels Procellaria westlandica Marine ornithology 25 31 36 Klomp NI Schultz M A 1998 The remarkable foraging behaviour of short tailed shearwaters breeding in eastern Australia Ostrich 69 373 McConnell B J Chambers C Fedak M A 1992 Foraging ecology of southern elephant seals in relation to the bathymetry and productivity of the Southern Ocean Antarctic science 4 393 398 Nicholls D G Murray M D Robertson CIR 1994 Oceanic flights of the northern royal albatross Diomedea epomophora sanfordi using satellite telemetry Corella 18 50 52 Nicholls D G Robertson C J R 2007 Assessing flight characteristics of the Chatham albatross Thalassarche eremita from satellite tracking Notornis 54 168 179 Nicholls D G Robertson C J R Naef Daenzer B 2005 Evaluating distribution modelling using kernel functions for northern royal albatrosses off So
33. twice the speed in the previous record and the next record THEN SPEED NOT EXCESSIVE OIF the speed was lt 40 km h AND both records were of LC 0 lt 44 or better THEN SPEED NOT EXCESSIVE IF the speed was lt 40 km ht Caption AND one or other records were of LC A B or Z THEN VERY PROBABLY EXCESSIVE consider speeds of nearby records IF the speed was lt 40 km hi AND the distance travelled was less than 25 km THEN PROBABLY NOT EXCESSIVE see Note in
34. usion a reason for the exclusion is allocated set Reason 3 to 6 Appendix C If the Reason gt 2 then the Distance Time Interval and Speed are automatically recalculated in the spreadsheet Appendix B These revised figures are then also used to determine if a better choice can be made IF the time interval from the previous location is very short resulting in an excessive speed Flag Too Soon THEN apply following tests and actions IF both locations are LC 3 2 or 1 excellent known accuracy records THEN consider retaining reasonable speed IF both locations are LC IQ 0 44 or better AND speed lt speed threshold AND time gt 0 2 h i e moderate quality records reasonable speeds THEN retain both records ELSE mark poorer record using SUBROUTINE A below IF the speed to the present point was excessive see below compared to adjacent points Flag Too Fast exclude the poorer point using SUBROUTINE B below IF LC A B or Z i e a very poor quality record see note in caption THEN consider marking for exclusion Reasons for excluding include an isolated change in speed or excessive zig zag locations assuming IF rate of change in lat and long is uncharacteristic or excessively zig zagging THEN inspect adjacent records to detect cause and consider marking the poorer quality record At the end of the selection plot speed of the selected records against date time Confirm there are no im
35. uth America Notornis 52 223 235 Nicholls D G Robertson CIR Murray M D 2007 Measuring accuracy and precision for CLS Argos satellite telemetry locations Notornis 54 137 157 Prince P A Wood A G Barton T Croxall J P 1992 Satellite tracking of wandering albatrosses Diomedea exulans in the South Atlantic Antarctic science 4 31 36 Weimerskirch H 1998 Foraging strategies of Indian Ocean albatrosses and their relationship with fisheries pp 168 179 In Robertson G Gales R ed Albatross biology and conservation Chipping Norton Australia Surrey Beatty amp Sons Pty Weimerskirch H Salamolard M Jouventin P 1992 Satellite telemetry of foraging movements in the wandering albatross pp 185 198 In Priede I G Swift S M ed Wildlife telemetry remote monitoring and tracking of animals Chichester United Kingdom Ellis Horwood 02222 Date 02 01 99 00 39 45 LC TQ 50 Larl 42 762S Lon 164 245W Lat2 31 473S lon2 110 717W Nb mes 008 Nb mes gt 120dB 0 Best level 124 dB Pass Duration 2555 NOPC 2 Calcul freq 401 6650276 4 Hz Altitude 000 m 12 255 136 45 131 Validating satellite locations aseqeyep eu 1x 10 Joayspeards ut poy ayeq woy m ysnut Joer p r ddns jou emwog read ay Jo Arenue se App woaz Suyunos Aep Jo uo sey euDap pue sAep Jo Jaquinyy 6297 6E Jaquinu gq F Aeq unt S n Aaen Aq partddns qumu uoncounu pir nbrumnq LPEL66 1 d lur d
36. utive locations were both LC 0 with IQ values of both records equalling 40 44 46 and less often 48 either of the pair of records was often plausible and the choice was consequently difficult It was generally determined on the trend of latitude and longitude selecting for the smoother flight direction and speed or arbitrarily excluding the 2nd of the 2 equal quality locations Nicholls et al 2007 note the different precision of LC A as against LC 0 Researchers may need to modify their selection process accordingly to reflect their own requirements As a result of this visual selection procedure the calculated point to point speeds of remaining records were all less than 100 km bh There are still many records with speeds over 50 km h Fig 1C These records would have been excluded if only LC 3 2 or 1 were used We considered them to be valid observations because with many records especially on fast migrations the distances travelled are so great that the location error is inconsequential Moreover for birds that fly very long distances the high speeds can be sustained for days Speeds determined from records of LC 3 2 and 1 only produced biases in the sample resulting from over representation of long time periods between the infrequent retained observations The paucity of records of LC 3 2 or 1 was exacerbated when long or intermittent duty cycle transmission regimes were used to study migration when fas
37. y ur pes Ol pue ar YI Sunceu leouo5 preg paepe SMO u1 UMOP p rdoo pue jrenueur jos st Op ye Jas aray pJoys Iyy p ds AL LLAM N ep 00L OF lt PAV AI UOOS ool g 0 gt FHV 4D g rg 4I e nurrol paads Atss 2x IO s eA l1u uur HOYS Suyajap Dia p leln eo Jaquinyy pueg 1x3 1 qumu I dado Jaquimu SU dd I 1 qumu I dado Jaquinu 6Z Sr daz 1 qumu OPO daz z 1 8 lur 00 4 4X9 YD F 3X0 UP OT SPETTA 1X9 YD g yu ururoo p ds Pastady IEAIaIT au P S a0uejsiq p stA q Je jeet UOT ISP yuon Oro 1281 oNpueg S H SV Uv OV dV OV NV WV TV MV v panuyuoy q xIpueddy 134 Nicholls amp Robertson Appendix C Selection criteria codes for marking CLS Argos records during procedure for selecting records Code Reason Location 1 selected Location 2 selected Improbable record of LCs 3 2 1 or 0 Improbable record of LC A Improbable record of LC B Improbable record of LC Z location index gt 10 Record in wrong hemisphere Presumed to be a transmission reception error in the PTT identification or an illegal transmission of an unregistered PTT with the same identification Typically there are only 1 or 2 messages and the location may coincide with the location of a manufacturer Not used for species that normally travel in both hemispheres 8 Pre or post deployment locations Locations obtained before
38. y an automated filter originally designed for seals McConnell et al 1992 or to use a small set of fixed criteria Although time consuming during the development of the technique the individual assessment procedure resulted in the retention of a higher proportion of the observations and a better understanding of the quality of the data and of other behavioural implications Nicholls et al 2005 Data preparation CLS Argos provides a record with a pair of locations and diagnostic information about their determination in the DIAG file Appendix A The record is obtained from either real time downloads using the DIAG command or the equivalent DIAG archival file which contains the complete dataset supplied by CLS Argos This file typically has a 6 line record for each location determination Appendix A The fields include PTT number date LC and IQ in the 1st line and the 2 alternate locations on the next line Other diagnostic data follow in the next 3 lines with the measurements from the PTT sensors or extra messages in the last line or lines The complete record is described in the CLS Argos manual Anon 1999 In this DIAG format researchers can read location records easily but these need to be converted into a spreadsheet or database format to aid sorting and to assess select and map the useable locations Concatenating the 6 line archival data into a 1 line record is achieved by stripping the field names units and ext
Download Pdf Manuals
Related Search