Resolved problems and new features
See also Distance 4.1 resolved problems.
Distance 5.0 Release 2 patch
Release date: 9 January 2008
MRDS Analysis Engine
355 Cannot deal with missing distances in interval data. Error is in create.interval.data function. [Fixed LJT 9 Jan 2008]
356 When running analyses using previous detection functions, get error “Could not find the method command in model definition”. [Fixed LJT 9 Jan 2008]
Distance 5.0 Release 2
Release date: 20 December 2006
- CDS/MCDS: Default behaviour changed when there is one transect in a stratum, or globally, or density is specified by sample. Previously in these cases, encounter rate variance was assumed to be Poisson. Now, encounter rate variance is assumed to be zero, and inferences should be restricted to density/abundance in the covered area, rather than the whole study area (if larger). A warning is issued, and appropriate documentation added to the manual.
- Import Data Wizard: Ability to choose type of decimal symbol added in Step 4 of the wizard. Default is to respect the operating system’s regional settings, and the alternative is to use the period – i.e., “.”.
MRDS Analysis Engine
- 349 When analyzing data in intervals, if also doing truncation get the error message “Error in process.data(data, meta.data, check = FALSE) : Error: width must exceed largest interval end point”. [Fixed LJT 21 Sept 2006 and JLL]
- See Fixes.txt in mrds library directory within R for details of other minor fixes (you need to run an MRDS analysis first to unpack the new library)
Distance 5.0 Release 1
Release date: 23 August 2006
This was not a public release.
- CDS/MCDS: Included a full list of warnings and errors produced by CDS and MCDS engine in online manual.
- CDS/MCDS: Added a warning when user does size bias regression using interval distance data (see known problem 344).
- MRDS: Now using mrds library version 1.2.7. Summary of new features since beta 5 (for a full list, see Fixes.txt in the mrds library folder – run an MRDS analysis first to unpack the new version):
- Various internal changes to streamline the program
- Adjustment terms added to ddf.df() – not yet available via the Distance graphical user interface as needs more testing.
- Various changes to statistics that are output
- Improved starting values and robustness of optimization
- Improved support for distance data in intervals
- Now works if there is only one stratum, and gives sensible output if Area = 0
- Improved support for single observer analysis (ddf.ds) with double observer configuration
- Improved detection function plots (although still further improvements to make – see Fixes.txt)
- MRDS: Added option under Help | Online manuals to open the mrds R library help.
- MRDS: R errors and warnings are now recognized in the analysis details log.
- MRDS: Added an option to get extended output from density estimates (Misc tab in model definition)
- MRDS: Added interface code that allows running of single-observer configuration surveys in this engine (but you have to use detection function method ds).
- MRDS: Added left truncation for single observer configuration
- MRDS: Added interval data, with fixed intervals for all observations (intervals that vary by observation is available in the mrds R library, but not yet in the interface). Note that you can’t do left or right truncation when there is interval data.
- MRDS: Correctly converts units of line length and perpendicular distance to units of area, but still ignores any unit conversion specified in the units tab of the data filter.
- Online manual updated to reflect the above changes, and other minor additions and improvements made.
- 347 General – when extra large fonts are used, some parts of the interface become hard to read as they disappear behind other controls [Leif Kastdalen 14 Aug 2006; Fixed LJT 18 Aug 2006]
- 348 General – the active tab appears at the wrong font size, depending on the font size setting in the Windows system. [LJT 18 Aug 2006] The problem is due to a bug in the tab control we use. [Workaround LJT 18 Aug 2006]
- 341 Main Window: When you have a multi-screen computer where Distance is open on the secondary machine, and you then unplug the second screen, Distance doesn’t move to the primary screen so you can’t see it! [EAR 14 April 2006; Fixed LJT 14 April 2006]
- 340 Model Definition Properties: CDS and MCDS – Estimate tab. When estimating density by sample, stratum and global, if Global density estimate is mean of stratum estimates, weighted by total effort in stratum, the stratum as replicates box becomes ticked when re-opening the properties, regardless of what it was when they were last closed. Doesn’t happen when not estimating density by stratum. [26 March 2006, Silka Kempema; Fixed in Beta 5]
Analysis Engines (problems affecting >1 engine)
- 339 MRDS and DSM engines: When the data structure is complex, with multiple substratum or subsample layers and there are covariates, can encounter an error “Error: Encountered a problem while executing the data creating query” (this problem was previously resolved in the CDS and MCDS engines). [Carter Watterson, 30 March 2006; Fixed LJT 30 March 3006]
MCDS Analysis Engine (includes CDS as well)
- 343 When using right truncation for cluster size (on Data Filter | Truncation) truncated observations were given a cluster size of 0 and included in the calculations, rather than being excluded. [29 June 06 DLB, Fixed 14 Aug 06 LJT]
MRDS Analysis Engine
- Various fixes made to the mrds library – see Fixes.txt for details (you need to run an MRDS analysis first to unpack the new library)
- 345 Summary info about GOF and Density estimates not displaying in results details – R code needs updating to use print rather than summary. [Fixed LJT 15 August 06]
- 342 Database Engine: When updating multiple records, under some circumstances the wrong records could be updated (when the ID order does not match the order of the records in the database table. [Carter Watterson, 20 April 2006; Fixed LJT 21 April 2006]
Distance 5.0 Beta 6
Release date: 20 January 2006
- This release, which was for internal distribution only, includes a preliminary version of the DSM analysis engine.
- MRDS: The way that field names are turned into variables to be included in the detection function formulae has been changed (to accommodate future analysis engines in R). The change only affects field names from layers other than the Observation layer. This means that formulae that use fields from layers other than the Observation layer as variables will need to be changed. For details of the new naming convention, see Chapter 10 of the online manual, under the topic “Translating Distance Fields into DS and MR Covariates”.
- 333 Analysis Details, Inputs tab. If you edit the data filter or model definition name, delete the name and then hit enter or click elsewhere, get a crash “Run-time error -2147214289 (80041c2f) Name cannot be blank”. [Mark Trinder 27 Sept 2005; Fixed LJT 28 Sept 2005]
- 334 Analysis Details, Log tab. If you don’t define the area units then you can get an message telling you to refer to the Misc tab of the Model Definition. This should refer you to the Units tab of the Data Filter. [Louise Burth 12 Oct 2005; Fixed LJT 12 Oct 2005]
- 337 Project Properties: For a non-geographic project, the geographic properties (projection system, etc) are set when the project is created, rather then being read from the defaults if a non-geographic project is set to geographic [LJT 18 Jan 2006; fixed LJT 18 Jan 2006]
- 338 Project Properties: The densification geographic property is set to be project specific by default, when the default should be to read it from the default settings [LJT 18 Jan 2006; fixed LJT 18 Jan 2006]
MCDS Analysis Engine
- 336 When you are fitting a detection function at one level (e.g. global) and estimating at a lower level (e.g., stratum or sample) and there are no observations at that level, the output gives the estimated f(0), ESW and p as 0.0. [30 Nov 2005 Mark Hulme; fixed 30 Nov 2005 - under these circumstances those estimates are no longer printed in the results]
MRDS Analysis Engine
- 335 When run under R 2.2.0, get the log error: Error in install.packages(“[directory here]\\mrds.zip”, : unused argument(s) (CRAN …). Workaround: MRDS is only supported with R 2.1.1; future versions will support 2.2.0 and later. For setup package of for the currently supported version, see the support page of the Distance web site [Louise Burt 23 Nov 05; Fixed LJT 24 Nov 05]
Distance 5.0 Beta 5
Release date: 30 August 2006
- MRDS: Now using mrds library version 1.2.2:
- Fixed a number of problems in beta (see resolved problems). Additional changes/new features listed below.
- 1) Prevent use of “observer” in model for trial data.
- 2) plot and gof functions now all contain a breaks= field and use it if provided.
- 3) predict functions now all contain compute and newdata argument and use them as specified.
- 4) if an object is not detected by either observer an error is issued and the analysis stops
- 5) if fields other than “observer” or “detected” are different between observers a warning is issued that it may cause analysis to fail.
- 6) bins used for an analysis are automatically used in plot and gof without specifying in the arguments. A binned analysis is easily created with a call to make.bins prior to the analysis.
- 7) Prevent use of “observer” in model for trial data
- 8) plot and gof functions now all contain a breaks= field and use it if provided.
- 9) predict functions now all contain compute and newdata argument and use them as specified.
- 10) if an object is not detected by either observer an error is issued and the analysis stops
- 11) if fields other than observer or detected are different between observers a warning is issued that it may cause analysis to fail.
- 12) bins used for an analysis are automatically used in plot and gof without specifying in the arguments. A binned analysis is easily created with a call to make.bins prior to the analysis.
- 13) modified ddf.ds to restrict exponent of haz rate to be >1
- 14) modified gstdint and tablecgf to use a standardized set of grid points (exponentially spaced with more points closest to 0) for computing integrals to create a spline for interpolation. calls to tablecgf in integratedetfct and flt.lnl were modfied.
- 15) ddf.ds and flt were also changed such that the option showit now has 3 levels: 0 - no messages; 1- basic messages about refitting; 2 - all iterations are shown
- 16) Also in ddf.ds the parscale command in optim is used to resolve problems with scaling continuous covariates - need to test this more but appeared to be working well. It uses the max value of each column in the design matrix as the scalar for the parameter. This puts each parameter on the same footing in the optimization.
- 17) added functions model.table, collect.models and collect.model.names; these are presently not documented and aren’t complete for all model types but does work for method=”ds”. It collects sets of results and builds a table for model selection.
- 18) dht.se was modified to return the component v-c matrices (vc1 for detection prob and vc2 for er).dht was modified to pass vc,vc1 and vc2 as a list in its return arguments. The list elements under vc are named total, detection and er. Also it computes the average detection probability and passes it back as average.p
- CDS/MCDS: Results stats file now gives parameter estimates, as well as number of key function, adjustment and covariate parameters. See “MCDS Engine Stats File” in the MCDS Engine appendix of the online help.
- Maps. Default colours are now chosen from a palette that ensures that (i) polygon lines contrast within lines/points; (ii) they are not too garish.
- 319 Analysis Details, Log tab. When running in debug mode, it reports the directories and filenames where it has saved command files, etc, in Windows short file (8+3) format. This can be confusing for users - would be better in long file format. [Jeff Laake 1 June 2005; Fixed LJT 3 June 2005]
- 311 Import Data Wizard: When click finish, get Error 3251: Operation is not supported for this kind of object, raised from D5IDWiz.ImportDataEngine.MakeImportStructure. Only occurs in versions of D5Beta4 downloaded between 8-14 April. Solution: Download a patched version of the import dll D5IDWiz.dll from the support page of the Distance web site, and use it to replace the original version. [14 April 05]
- 318 Model Definition Properties: CDS and MCDS – Estimate tab. If you select stratify and then detection function by stratum and then no stratification, you’re left with no detection function boxes being ticked. [Aditya Gangadharan 2 June 2005, Fixed LJT 2 June 2005]
- 320 Model Definition Properties: CDS and MCDS - Estimate tab. When selecting to estimate detection function by stratum and by sample, the interface can ignore the selections you make for estimating the global density estimate as a function of the stratum estimates. Can also get the message “The option to obtain global density estimates as the unweighted mean of stratum estimates has now been removed from Distance.” when it does not apply. [Aditya Gangadharan 15 June 2005; Fixed LJT 16 June 2005]
- 316 Model Definition Properties: MRDS. If you select stratification, or a sample layer other than the default, this is not remembered next time you open up the model definition. [Pierre Richard 30 May 2005; Fixed LJT 1 June 2005]
- 317 Model Definition Properties: MRDS. Pressing Default causes the options available in the Sample layer combo box to be repeated, so that the same options become available twice. [LJT 1 June 2005; Fixed LJT 1 June 2005]
- 315 Setup Project Wizard: When you define a sampling fraction multiplier, the wizard sets the default action for this multiplier in the model definitions to be to multiply the density estimate, rather than to divide it. [Jeff Breeden 18 May 2005; fixed LJT 18 May 2005]
Design Engine and Design Properties
- 312 Equal spaced zigzag design sometimes doesn’t generate the first or last sampler. [Laura Marshall 1 May 2005; Fixed LJT 5 May 2005]
- 313 Equal spaced zigzag design sometimes fails on coverage probability simulation with error 5002 “Invalid array index.” in D5DesEng.CModDesignEngineTravelCosts.TotalZZLoopTravelDistance [Laura Marshall 9 May 2005; Fixed LJT 9 May 2005]
- 314 Overflow error in D5DesEng.ZigzagEqualSpaced.GenerateSamplers when generating coverage probability estimates using a very large number of strata and simulations [Len Thomas 9 May 2005; Fixed LJT 9 May 2005]
- 236 When running an analysis, occasionally get an error from D3NEH.NEngineHost.OpenProjectDatabase. Error number 3260, “Couldn’t update; the database is currently locked by user ‘admin’.”, source D4DbEng.ProjectSettings.Setting. Currently only reported on a Windows 2000 machine. [Charles Paxton 9 May 2003] [I think I’ve fixed this, but will leave as outstanding until confirmed, LJT 9 May 2003][Similar problem occurs with IsBeta setting - I think I fixed it, LJT 1 July 2005]
CDS Analysis Engine
- 321 When try to estimate density by sample and detection function global, get the error message “** Error: Density for each sample is unnecessary when detection and expected cluster size are estimated at higher levels **”. This is a nuisance because there are situations when you want the estimate by sample, for example for exporting to analyses outside Distance. [Aditya Gangadharan 15 June 2005; Fixed LJT 16 June 2005 - note though that stratum variances are different if you estimate density by sample than if you do not - I guess because it is using the Innes et al. type variance estimator when you estimate density by sample (see the mrds documentation for explanation of Innes et al. type variance estimator)]
- 325 Degrees of freedom reported incorrectly for detection function, cluster size and encounter rate estimates in bootstrap summary [LJT 24 Aug 2005; Fixed LJT 29 Aug 2005]
MCDS Analysis Engine
- 322 Can return f(x,z)=-1 under some circumstances (estimated key function scale very small or very large) [LJT 30 June 2005; Fixed LJT 30 June 2005]
- 323 When fails to converge, does not correctly retrieve results from previous iteration. Instead gets results from intermediate tries. [LJT 30 June 2005; fixed LJT 30 June 2005]
- 324 When fails to converge, often gives a misleading warning message about the reason why [LJT 30 June 2005; fixed LJT 30 June 2005]
MRDS Analysis Engine
- 320 Fixed error in binned data analysis; create.model.frame was changing binned to numeric 0/1 instead of T/F because it was thrown into a matrix. This caused indexing errors in accessing data and it screwed up fitting for binned data with covariates.
- 321 Ordering of partials and v-c matrix of exponent and scale parameters in hazard rate didn’t agree which created incorrect variance estimation for hazard rate.
- 322 If distance was not included in mrmodel for method=io.fi or trial.fi, the results were incorrect because the analytical integral for the logistic was not correct.
- 323 Made changes to gof, qqplot and plot routines to handle left truncation properly.
- 327 Fixed error in binned data analysis; create.model.frame was changing binned to numeric 0/1 instead of T/F because it was thrown into a matrix. This caused indexing errors in accessing data and it screwed up fitting for binned data with covariates. . [JLL May 05]
- 328 Ordering of partials and v-c matrix of exponent and scale parameters in hazard rate didn’t agree which created incorrect variance estimation for hazard rate. . [JLL May 05]
- 329 If distance was not included in mrmodel for method=io.fi or trial.fi, the results were incorrect because the analytical integral for the logistic was not correct. . [JLL May 05]
- 330 Made changes to gof, qqplot and plot routines to handle left truncation properly. . [JLL May 05]
Distance 5.0 Beta 4
Release date: 8 April 2005
- CDS/MCDS: When bootstrapping, progress in percent is shown in the interface.
- CDS/MCDS: When bootstrapping, the bootstrap point estimates are now the mean of the estimates for each bootstrap replicate. (Previously used reported just the original, non-bootstrap point estimate.) Bootstrap CVs are bootstrap SE/bootstrap point estimate. (Previously were bootstrap SE/original point estimate.)
- CDS/MCDS: When bootstrapping, if errors occur in some bootstrap replicates such that density and other statistics cannot be obtained from those runs, Distance now reports the number of runs where the bootstrap failed.
- See also previous beta versions, below.
- 281 New version of Tab control used in Distance 5 causes a run time error when closing Distance 4: error 91, “object variable or with block variable not set”. This can safely be ignored!
- 297 After double-clicking on the Distance setup program, the setup program for another package (e.g., Mark) may start up. Solution: Make sure there are no other setup programs in the same directory as Distance. [Shane Wellendorf 11 Aug 04; Fixed LJT 11 Aug 04 – renamed the setup program from setup.exe to somethng that’s likely to be unique]
- 306 General - opening projects. Under some regional operating systems, double-clicking on a Distance project to open it cased an error 13 - unknown type. [Atanga Ekobo Aug 2004, Fixed LJT 2 Feb 2005]
- 300 Manual: Need to describe how to do smearing – in data filter you need to set automatic equal intervals, as choosing the intervals manually doesn’t use the NCLASS command in the analysis engine. Also point out that smearing not supported for MCDS. [LJT 27 Oct 2004]
- 134 Analysis Details, Results tab. When results pages are very long, and you scroll down to the bottom of one page, then when you click on next you are taken to the bottom of the next page. Should take you to the top of each page as you enter it [26 Sept 01, LJT]
- 22 Browsers/Analysis Components: When you delete a Model Definition in the Analysis Components window, the IDs of the model definitions are not updated in the Analysis Browser table. [2 Nov 00] Also happens when you delete an analysis set using the Analysis Browser toolbar – the IDs of the new set shown are not updated. This causes an error in the mouse over event of the Analysis Browser when it doesn’t recognize the ID – should error trap this, as well as fixing the source of the problem. [3 Nov 00; Fixed LJT 12 Jan 05]
- 25 Browsers: when you hit Enter after changing the name of an item, all items in that set become selected (only the one you were editing should be selected).
- 50 Properties: When you click on OK, the contents of the current tab are not validated before the contents are saved [7 Dec 00]
- 51 Preferences: Changing options do not seem to have any affect until you restart (e.g., changing default open path; changing timestamp and echo data options for analysis).
MCDS Analysis Engine
- 217 Smearing has not been tested for the MCDS engine – it may produce incorrect results. In any case, qq-plots for MCDS analyses with smearing have been disabled, as they don’t make sense when there are non-integer numbers of observations. [LJT: 3 Feb 03. Disabled smearing interface for MCDS analyses, and added a note to the manual, LJT 1 Jan 05]
- 305 When fitting detection function at a higher level and estimating at a lower one (e.g., fit globally, estimate by stratum) and doing a bootstrap, get the inappropriate warning message ** Bootstrap Warning: Degrees of freedom less than 1.0 for estimating confidence limits on density. Confidence limits not calculated. **. This message should not appear, since bootstrapping is the correct way to obtain variance estimates in this situation. [Tiago Marques 8 Dec 2004; Fixed LJT 6 Jan 2005]
- 308 Same as 305, but message also appears when cluster size is a covariate [Tiago Marques, 17 Feb 2005; Fixed LJT 17 Feb 2005]
- 307 When bootstrapping, with cluster size as a covariate, can get “Internal error 50, mismatched strata” errors, and similar. Problem is caused by Distance trying to report expected cluster size during bootstrap, when this is not estimated when cluster size is a covariate. [Tiago Marques Jan 2005; Fixed LJT Feb 2005]
- 309 When the data structure is complex, with multiple substratum or subsample layers and there are covariates, can encounter an error “Error: Encountered a problem while executing the data creating query”. [Dirk Burhans, 2 March 2005; Fixed LJT 3 March 3005]
Distance 5.0 Beta 3
Release date: 6 January 2005
- MRDS Analysis engine
- CDS/MCDS: When using multipliers, can now include degrees of freedom used to the multiplier SE (including cue rate), and these are taken into account when calculating the degrees of freedom for the density estimate (using the Satterthwaite formula – 3.75 in Introduction to Distance Sampling). When degrees of freedom are not specified, they are assumed to be infinity. Note that the calculated dfs (and therefore confidence limits) will be different in this version of Distance to previous versions as Distance previously ignored multipliers when calculating the df (see known problem 256, now fixed).
- CDS/MCDS: Added number of observations to list of columns available in the Analysis Browser. Number of observations, number of samples and total effort are now also summed over samples/strata.
- CDS/MCDS: Extended the number of columns in the stats file, to
accommodate very large number of strata and samples. New format is:
See online documentation for further details of the stats file and its uses.
- MCDS engine maximum number of factor levels per factor covariate increased from 50 to 200.
- MCDS engine: New diagnostic added to the bottom of the Parameter Estiamtes page – a table summarizing the distribution of the estimated probability of detection, given the covariate values. This is useful for diagnosing possible problems with the estimates due to low estimated detection probabilities. See the online help, Chapter 9, MCDS analysis guidelines for more information.
- Components: Actbar2.ocx upgraded from 188.8.131.52 to 184.108.40.206. Fixes some minor bugs in the toolbar.
- Components: Tab32x30.ocx upgraded from 220.127.116.11 to 18.104.22.168. Fixes some minor bugs in the tab control.
- Components: MapObjects GIS upgraded to version 2.3
- 261 Shortcuts to Distance and the help files are added to the current user’s start menu with no provision for adding the shortcuts to all users start menu. [Ed Debevec 11 Feb 04; Fixed LJT 13 Feb 04]
- 262 On multi-users systems, where users don’t have permissions to access the Distance program directory the instructions for moving DistIni.mdb to another folder and editing the HKEY_LOCAL_MACHINE section of the registry don’t work. You still get the “cannot open” error message. [Ed Debevec 11 Feb04; Fixed LJT 13 Feb 04]
- 296 When windows are maximized, opening a new window can cause the error message “Invalid bands collection index, Distance MdiMain.LoadChildMenu”. This is caused by a bug in the toolbar component – to fix it, download an updated version of the component from the Distance web site, under Distance 4.1 support. The updated component will be included in the next release of Distance. [Ken Burnham 10 Aug 2004; workaround LJT 12 Aug 2004]
- 271 Analysis Browser: When using point transcts, column summarizing effective detection radius still says “ESW” (as in effective strip width). [Steve Buckland 6 April 2004; Fixed LJT 9 April 2004 – solution: now says EDR/ESW. Applied similar solution to other columns that have different terms in point transect analyses]
- 254 Help: Using F1 to access help produces a Distance internal error “Could not find Help ID” for some windows (e.g., Analysis Browser). [8 Dec 03; Fixed 8 Dec 03 LJT]
- 259 Main Window: When choosing Windows | Cascade, Map windows do not arrange themselves in the cascade. [LJT: 5 Feb 04; Fixed LJT 5 Feb 04]
- 48 Map Browser: Sometimes when you try to delete a stored map, it claims that “No map selected” when you can see that one is [Internal note: this is a problem with the data control not being on the same record as the flexgrid – see CCtlMapBrowser.DeleteMap] [Fixed LJT 5 Feb 04]
- 260 Map Browser: Moving between lines in the Maps tab of the Project Browser can sometimes cause a “No current record” error [LJT 5 Feb 04; Fixed LJT 5 Feb 04]
- 282 Model Definition Properties: CDS and MCDS, Variance tab. When you are using stratification, if you select to bootstrap only by observations, it saves options for bootstrapping by both samples within strata and observations. [David Borchers 26 May 2004; Fixed LJT 27 May 2004]
- 284 Preferences: Survey design. Changes to coverage probability map number of classes are not saved when you click OK [3 June 04 LJT; Fixed 3 June 04, LJT]
- 52 Survey Properties, Data Fields tab. Not updating properly when survey methods options chosen. (For example, change from line transect to point transect, the perp distance field is still white and the radial distance field greyed out, rather than visa versa). [Fixed LJT 6 April 2004]
- 53 Survey Properties: When you change the properties of a survey that has status run, the status is not reset or results deleted [12 Jan 01; Fixed LJT 6 April 2004]
Design Engine and Design Properties
- 258 For line designs, realized sampler area covererage and proportion of stratum sampled were calculated incorrectly in Design engine output page (usually as <0.001) [Srinivas V, 12 Jan 04; Fixed LJT 3 Feb 04]
- 264 Results not displayed correctly on systems where decimal place is denoted by a comma [Gjermund Bøthun 8 March 04; Fixed LJT 18 March 04]
- 267 Under some circumstances, the design engine can lock up – occurs when buffering a complex study area. [LJT 14 March 04; Fixed LJT 15 March 04] [Internal note – created the function GetCombinedExtent to get around this – LJT]
- 283 In an equal spacing zigzag design, if the design axis is at 90 degrees, get a subscript out of range error [David Borchers 27 May 2004; Fixed LJT 27 May 2004]
- 285 When run a design using a projected coordinate system, get “Object variable or with block variable not set” in D5DesEng.GeneralDesignProperties.InitDesignObject [Pierre Richard 20 May 2004; Fixed LJT 4 June 2004]
- 286 Projection parameters not being used when running designs [LJT 4 June 2004; Fixed LJT 4 June 2004]
- 287 Design properties: Projection not being saved correctly when user asks for same coordinate system as stratum, and stratum is projected with parameters [LJT 7 June 2004; Fixed LJT 7 June 2004]
- 288 When the new sample layer is to be projected, sample units are not saved projected onto it, so are in completely the wrong locations [LJT 7 June 2004; Fixed LJT 7 June 2004]
- 290 In some cases for zigzag designs, does not create a convex hull for a stratum and so survey effort is not clipped correctly. [Tiago Marques and LJT 12 July 2004; Fixed LJT 13 July 2004]
- 291 For zigzag designs where projected units are small numbers (e.g., projections in kilometers) can fail to correctly create a convex hull. [Tiago Marques and LJT 12 July 2004; Fixed LJT 13 July 2004]
- 292 Design Properties: Systematic Random Line Sampler: When you specify total line length, number of samplers and sampler spacing is not updated and remains at 0. When you then generate a survey from this design, you get “** Warning: No effort. Unable to generate design. **”. [Tiago Marques 12/7/04, Fixed LJT 13/7/04].
- 256 Degrees of freedom for density estimate calculated using Satterthwaite formula (3.75 in Introduction to Distance Sampling) ignore any multipliers, including cue rate multiplier. Reported df and confidence limits on density and abundance are therefore (slightly) wrong. [Tiago Marques 20 Jan 2004; Fixed LJT Jan 2004]
- 257 When cue rate has an SE, this is not reported in the results summary, although this SE is correctly incorporated into the CV of the final density estimate – just a formatting problem [LJT: 2 Feb 2004; Fixed LJT 2 Feb 2004]
- 270 Interface allows up to 30 intervals, but analysis engine gives an “array out of bounds” error when using >25 intervals [M Obbard 5 April 2004; Fixed LJT 6 April 2004]
- 272 Cue rate parameter rho not defined in list of parameters on first page of output [Steve Buckland 7 April 2004; Fixed LJT 10 April 2004]
- 273When there is only one sample, encounter rate cannot be estimated empirically. When the user asks for an empirical estimate, a Poisson distribution of counts is assumed, but no warning is given. Need to add`a`warning. [Tiago Marques 1 April 2004, Fixed LJT 10 April 2004]
- 277 When estimating density by stratum and globally, but not by stratum, can get internal error 50: mismatched strata. [David Hanni April 2004; Fixed LJT 28 April 2004]
- 279 When estimating density and abundance by sample, abundance is estimated for the area of the stratum that the sample is within. Would be better if it was abundance in the area covered by the sample. [LJT 29 April 2004; Fixed LJT 29 April 2004]
- 280 When estimating density globally from the mean of stratum estimates, weighted by area, if area if all strata is 0 Distance incorrectly uses the sum of stratum estimates rarther than a simple mean [Tiago Marques 5 April 2004; Fixed LJT 5 April 2004]
- 294 With objects in clusters and very large number of observations, can get integer overflow error from line 254 of CMOD function [Martin Raphael 12 July 2004; Fixed LJT 14 July 2004]
- 295 Under some circumstances, if you try to post-stratify by a field in the observation layer, you get the error message “** Error: Encountered a problem while executing the data creating query. **” [Paul Lefort 6 August 2004; Fixed LJT 10 August 2004]
- 303 Can get “Internal Error 102: Could not evaluate area under CDF” when the detection function is quite spiked. This error prevents the engine from producing a sensible QQ plot. [LJT 30 Nov 2004; Fixed LJT 30 Nov 2004 – upped the number of iterations of the Simpson’s rule integration evaluator from 10 to 20. If problem re-occurs, will need to up further]
CDS Analysis Engine
- 268 When bootstrapping to obtain variance, if you use mean cluster size as estimate of expected cluster size (rather than using size bias regression), get Internal error 53 and 54 (mismatched stats and mismatched modules) and no bootstrap estimate of variance of mean cluster size. [LJT 17 March 2004; Fixed LJT 17 March 2004]
- 269 Under some circumstances can diagnose convergence failure when none occurred. [LJT 17 March 2004; Fixed LJT 17 March 2004]
MCDS Analysis Engine
- 278 When estimating by sample or strata, the Density Estimates page of output has the total number of observations, effort and number of samples rather than the number for the appropriate sample/stratum. [LJT 28 April 2004; Fixed LJT 29 April 2004]
- 293 Upper bound on second parameter of hazard rate key function is too high. [Jeff Stratford and LJT 8 July 2004; Fixed LJT 13 July 2004 – set to 20 as with CDS analyses]
- 157 GOF P-value not saved to stats file [4 Dec 01, LJT; Fixed LJT 2 Dec 04]
- 298 Plot of factor levels with two factors (and probably more) shows repeats of the same combinations, and misses out other combinations (see det.func.zip project, analysis 24) [Rob Williams and LJT: 25 Oct 04; Fixed LJT 1 Dec 2004]
- 299 On a hazard-rate analysis, starting off from the constrained ML estimate for the CDS analysis, didn’t converge (but no warning) and the LnL estimate is way off (see det.func.zip project, analysis 73). [Rob Williams and LJT: 25 Oct 04; Fixed LJT 2 Dec 04 – set Simpson’s rule max iterations to 20 (from 10)]
- 301 Analysis 78 in det.func.zip project crashed in labels.for [LJT: 27 Oct 2004; Fixed LJT 1 Dec 04]
- 302 Can get incorrect parameter indexing on some output when fit multiple key functions and use a selection criterion to select the best [LJT: 30 Nov 2004; fixed LJT 30 Nov 2004]
- 303 Analyses with cluster size as a covariate produce mismatched strata internal errors [LJT: Summer 2004; Fixed LJT 30 Nov 2004]
- 304 Doesn’t calculate qq plot after convergence failure [LJT 2 Dec 04; Fixed LJT 2 Dec 04]