AuvSurvey.xml

<!--
   The complete AUV survey scenario: a virtual ocean, a vehicle, an autopilot, a
   guidance stack and three scientific instruments, in one model.
 
   A survey AUV undulates between waypoints over a rolling seabed at about 140 m,
   sampling a stratified water column with a CTD, counting Calanus with a silhouette
   camera and writing a three frequency echogram to file. Everything in it is already
   covered by a focused regression case of its own under tests/in/Auv/; this file is
   the one place where the whole stack runs together, and it is also the regression
   input tests/in/AuvSurvey/AuvSurvey_in.xml, byte for byte, so the example and the
   test cannot drift apart. AuvSurvey_Test asserts that.
 
   ## The site
 
   About 68 degrees north off Lofoten, in outer Vestfjorden, in June. Every number in
   the environment below is only defensible relative to a stated site and season, so
   here it is stated: bottom around 140 m, a Norwegian Coastal Current surface layer
   over Norwegian Atlantic Water, a subsurface Calanus band in early summer, mackerel
   on their epipelagic feeding migration above it, and a demersal cod aggregation near
   the bottom.
 
   ## The fields, and where they come from
 
   Every scalar field here is ScalarField.<Name>.Model = "Parametric", read out of
   auv_ocean_fields.json beside this file. A parametric field is a depth profile plus a
   list of features - a Gaussian blob, a horizontal front, a depth band - so it is a
   function of north, east and depth and not of depth alone. The five entries are:
 
     SeaTemperature  9.5 degC June coastal water over a 20 m to 60 m thermocline down
                     to 7.4 degC and the 6.5 degC Atlantic core at 400 m, with a front
                     and a warm lens on top of it.
     SeaSalinity     33.8 PSU coastal water over Atlantic water above 35.0 PSU below
                     80 m, with the same front. The salinity front shares the origin
                     and the heading of the temperature front and carries the opposite
                     sign, so the cooler side is the saltier, Atlantic side: it is one
                     water mass boundary and not two unrelated ones.
     Calanus         A band centred at 28 m with a 9 m standard deviation, peaking near
                     2.2e-3 kg/m3, over a weak background, with two patches on top.
     Mackerel        Held in the upper 20 m by its background profile, with two loose
                     schools centred at 12 m and 10 m, peaking near 6e-2 kg/m3.
     Cod             Demersal, and the one field whose meaning is different in kind.
                     It has its own section below.
 
   The survey box is 300 m by 100 m at the origin, and the features are laid out over a
   far wider region than that: the Calanus patches are centred 600 m and 1500 m north,
   the mackerel schools 400 m and 1700 m north, the front 1000 m out. So for the most
   part this mission flies through the flanks of the structure and not through its
   cores. Inside the box the front is very nearly a constant offset, about -0.53 degC
   and +0.34 PSU, and the nearer Calanus patch adds at most some 1.5e-5 kg/m3 on top of
   the band, against a band peak of 2.2e-3. That is the intent and not an oversight:
   the file describes a region, this mission samples one corner of it, and a mission
   that ran north to 1500 m would fly through a patch.
 
   The one feature the box does reach into is the first mackerel school, 400 m north
   and 200 m east with sigmas of 150 m and 90 m: at the north-east corner of the track
   it still holds about 2.2e-2 kg/m3, a third of its peak, at its own depth. The school
   and the mackerel background stay above 20 m, clear of the Calanus band (FISH-0042).
 
   ## Cod follows the seabed
 
   The Cod entry sets "z_reference": "seafloor", so every vertical coordinate in it is
   a HEIGHT ABOVE THE BOTTOM, positive up, and the layer rides the rolling bathymetry
   instead of sitting at a fixed depth. It peaks near 1.2e-2 kg/m3 about 5 m above the
   bottom, wherever the bottom happens to be. That is what a demersal aggregation is,
   and it is precisely what a depth-only field cannot say: the previous version of this
   file pinned the cod into a fixed 126 m to 145 m band and called it a bottom layer,
   which it was only on average and never at a given point.
 
   All three species fields also set "mask_below_seabed", so their concentration is
   exactly zero under the bottom and the sounder does not integrate fish through rock.
 
   ## ConfigFile, and the working directory it is resolved from
 
   ScalarField.<Name>.ConfigFile is opened with the path exactly as written, relative to
   the PROCESS working directory and not to the directory this file sits in. Both ways
   the scenario is run start in playpen/bin - the example run below, and the ctest run
   of tests/in/AuvSurvey/AuvSurvey_in.xml, which is this file byte for byte and which
   CMake gives WORKING_DIRECTORY ${PLAYPEN_DIR}/bin - so one relative path serves both:
 
      cd build/Release/playpen/bin
      ./FhSim ../examples/input/fishery/AuvSurvey.xml
 
   That is why the path below points into the installed playpen tree and not into the
   source tree. Run it from anywhere else and the path has to move with you.
 
   A Parametric field must use ConfigFile and not an inline Config: every XML attribute
   in this tree is passed through the expression evaluator, which is not bracket aware
   and mangles a JSON array of numbers. See fhsim_environment issue ENV-0019.
 
   ## Watching it run
 
   The same file, run with FhVis instead of FhSim, draws itself:
 
      cd build/Release/playpen/bin
      ./FhVis ../examples/input/fishery/AuvSurvey.xml
 
   The vehicle is a yellow torpedo with fins that deflect to the commanded RudderDive
   and RudderYaw, the echosounder draws its beam as a translucent cone out to
   MaxRangeDown, the seabed is the SumOfSines surface the Cod field is
   referenced to, and three visualisers in the OBJECTS block below draw the ocean the
   instruments are measuring: the Calanus layer and the demersal cod as ray-marched
   volumes, and a vertical temperature section running 1900 m north through the water
   mass front. See the comment above those three objects for the sizing.
 
   The live view exists only because of <Visualization frameRate="25"/> in the OBSERVERS
   block. An OBSERVERS block is honoured exactly as written and there is one here for
   the FileOutput, so without that line FhVis would run this model headless.
 
   The camera starts at NED (-30, 0, 0) looking at the origin, which is beside the first
   waypoint and above the vehicle's 20 m start depth. Steer it with the arrow keys and
   the numeric keypad.
 
   ## Why the species are what they are
 
   The three species are chosen to separate by frequency, which is the whole reason a
   three frequency sounder is worth simulating. Each field names its Species, a key of the
   echosounder Parameter library (fhsim_fishery data/acoustics/parameter_library.json,
   copied from the echosounder sensor repository), and its mean LengthM; the library gives
   the target strength per frequency, the weight per length and the length spread.
 
   The consequence worth knowing before reading the echogram: at the design Calanus peak
   the layer alone is Sv = -101.0 dB at 38 kHz, -81.4 dB at 120 kHz and -73.2 dB at
   200 kHz. The modelled noise floor of this sounder reaches -101 dB at about 16 m of
   range at 38 kHz, so Calanus alone is invisible there beyond that, while at 200 kHz it
   survives far beyond the vehicle's reach. That contrast between frequencies, not blanket
   visibility, is what the echogram is for.
 
   ## The Calanus band holds Calanus
 
   The mackerel field stays above 20 m and the cod field below 80 m above the bottom, so
   the 26 m to 31 m band is Calanus at 120 kHz and 200 kHz (FISH-0042). At 38 kHz Calanus
   is below the noise floor and the tails of the mackerel schools, so the band's rise from
   38 kHz to 200 kHz, about 24 dB, is a little short of the 27.8 dB of Calanus alone.
 
   ## The seabed is pinned
 
   Environment.RandomSeed defaults to a time based seed, and the default bathymetry is
   a sum of randomly phased structures rather than a flat bottom. Both are pinned here,
   so the seabed is the same rolling surface on every run. See FISH-0036.
 
   ## Undulation envelope
 
   The dive control flies the vehicle between 10 m and 45 m rather than down to the
   80 m default, for two reasons. The band has to contain the Calanus layer at 28 m and
   the mackerel above it, which is what the instruments are here to find; and a
   commanded dive angle is not a flight path angle, so the vehicle makes only about
   0.36 m/s of vertical speed at the default 0.349 rad command. A 35 m band is about
   97 s per leg, so a 600 s mission is five legs. See FISH-0037.
 
   DepthHysteresis is 4 m, not the 1 m default. One metre is the minimum for the
   switching logic alone; it is not enough once the loop is closed, because the vehicle
   makes 2 to 3 m of depth while the pitch loop swings over, and the depth then leaves
   the commanded band at every turn.
 
   ## The gain signs
 
   Heading positive, pitch negative, speed positive: each gain takes the sign of the
   plant it drives. PidController.cpp forms error = measurement - reference and
   sfh::filters::PID then forms -kP*error, so the two inversions cancel and the
   composition is the standard convention. See FISH-0035.
 
   SpeedPid.InitOutput is a feed forward and not a tuning knob. 30 W is the cruise
   power of this parameter set at the commanded 1.5 m/s; at a different commanded speed
   it must be changed with it, or the loop spends tens of seconds unwinding an
   integrator that started in the wrong place.
 
   ## Neutral buoyancy on purpose
 
   AverageDensity is 1025, the same as Rho. Depth hold in this stack is pitch hold, so
   at the trimmed default of 1020 the vehicle rises at about 0.115 m/s regardless of
   what the pitch loop does, and the undulation envelope would not hold.
 
   ## A calm sea and no current
 
   Waves.NumWaves is zero and the current is zero, which makes this the clean case. A
   sea state is a one line change, and a shallow vehicle then sees the wave orbital
   velocity through the hydrodynamics and the autopilots see it through Speed and
   Pitch; that is intended behaviour and not a defect. StepMax of 0.0125 s resolves a
   7 s peak period 560 times over.
 
   ## StepMax divides every instrument period
 
   0.0125 s is not the 0.05 s the plan suggested, and the difference is not about
   accuracy. A sensor fires on the first accepted step at or after its due time, so an
   integrator step that does not divide a sample period makes the samples late by up to
   one step: at 0.05 s the CTD, which samples at 0.0625 s, is up to 37 ms late and
   SampleClock says so once, in a warning naming the StepMax it wants. 0.0125 s divides
   0.0625 s, the SilCam's 0.25 s frame period and the sounder's 1 s ping alike, so
   every instrument fires on its own grid. It is also the number a co-simulation host
   should pick for communicationStepSize.
-->
<Contents>
   <OBJECTS>
      <Lib
         LibName                       = "environment"
         SimObject                     = "Environment"
         Name                          = "Env"
         RandomSeed                    = "20240601"
         Waves.Model                   = "Component"
         Waves.NumWaves                = "0"
         Current.Model                 = "Constant"
         Current.Vector                = "0, 0, 0"
 
         Bathymetry.Model              = "SumOfSines"
         Bathymetry.Depth              = "140"
         Bathymetry.NumStructures      = "10"
         Bathymetry.StructureLengthMin = "150"
         Bathymetry.StructureLengthMax = "600"
         Bathymetry.StructureMaxHeight = "6"
 
         ScalarField.Names             = "SeaTemperature, SeaSalinity, Calanus, Mackerel, Cod"
 
         ScalarField.SeaTemperature.Quantity   = "Temperature"
         ScalarField.SeaTemperature.Unit       = "degC"
         ScalarField.SeaTemperature.Model      = "Parametric"
         ScalarField.SeaTemperature.ConfigFile = "../examples/input/fishery/auv_ocean_fields.json"
         ScalarField.SeaTemperature.ConfigKey  = "SeaTemperature"
 
         ScalarField.SeaSalinity.Quantity      = "Salinity"
         ScalarField.SeaSalinity.Unit          = "PSU"
         ScalarField.SeaSalinity.Model         = "Parametric"
         ScalarField.SeaSalinity.ConfigFile    = "../examples/input/fishery/auv_ocean_fields.json"
         ScalarField.SeaSalinity.ConfigKey     = "SeaSalinity"
 
         ScalarField.Calanus.Quantity          = "SpeciesConcentration"
         ScalarField.Calanus.Unit              = "kg/m^3"
         ScalarField.Calanus.Model             = "Parametric"
         ScalarField.Calanus.ConfigFile        = "../examples/input/fishery/auv_ocean_fields.json"
         ScalarField.Calanus.ConfigKey         = "Calanus"
         ScalarField.Calanus.Species           = "calanus_finmarchicus"
         ScalarField.Calanus.PropertyNames     = "LengthM"
         ScalarField.Calanus.PropertyValues    = "0.0028"
 
         ScalarField.Mackerel.Quantity         = "SpeciesConcentration"
         ScalarField.Mackerel.Unit             = "kg/m^3"
         ScalarField.Mackerel.Model            = "Parametric"
         ScalarField.Mackerel.ConfigFile       = "../examples/input/fishery/auv_ocean_fields.json"
         ScalarField.Mackerel.ConfigKey        = "Mackerel"
         ScalarField.Mackerel.Species          = "mackerel"
         ScalarField.Mackerel.PropertyNames    = "LengthM"
         ScalarField.Mackerel.PropertyValues   = "0.35"
 
         ScalarField.Cod.Quantity              = "SpeciesConcentration"
         ScalarField.Cod.Unit                  = "kg/m^3"
         ScalarField.Cod.Model                 = "Parametric"
         ScalarField.Cod.ConfigFile            = "../examples/input/fishery/auv_ocean_fields.json"
         ScalarField.Cod.ConfigKey             = "Cod"
         ScalarField.Cod.Species               = "cod"
         ScalarField.Cod.PropertyNames         = "LengthM"
         ScalarField.Cod.PropertyValues        = "0.60"
      />
 
      <!-- ConfigKey is given on every field even though it defaults to the field name,
           because one JSON document holds all five entries and the mapping from an
           XML field name to a key in that document is worth being able to read here.
           The Cod entry is the seafloor referenced one; see the header. -->
 
      <!-- ================================================================
           The environment, drawn.
 
           Three fhsim_environment visualisers, all of them standalone: they have no
           ports and no connections, they find the Environment through the shared
           "Environment" resource rather than by its name, and they find their field by
           the name it is listed under in ScalarField.Names. That lookup and the
           time-invariance check happen in VisualBakedScalarField::FinalSetup, which
           runs in the HEADLESS binary too, so a typo in FieldName below fails the
           regression run and not only the live view. That is deliberate.
 
           Each one bakes its field once, at startup, into a texture and costs nothing
           per frame afterwards, so the resolutions below are a startup cost and not a
           frame cost. They are kept modest all the same - about 200 k cells per volume
           against the 3.5 M of fhsim_environment's own ScalarVolumeBlobs demo - because
           the bake is CPU work and this scenario is usually watched over software GL.
 
           ReferenceLength is the path length AlphaRange is quoted over. The default of
           1 m makes any ray crossing more than a few metres of visible field opaque, so
           it is set here to the vertical scale of each structure: 20 m for the Calanus
           band, 10 m for the thinner cod layer.
 
           Mackerel is deliberately NOT drawn. Its field fills the upper 20 m, directly
           above the Calanus box over its entire footprint, so seen from the side the two
           ray-marched volumes would blend into one wash and neither would be readable.
           Adding a third <Lib> for it is a copy of the Calanus block with FieldName,
           ColorRange and VisibleRange changed, if that is the picture wanted.
 
           Two occlusion artefacts of VisualScalarVolume are documented and accepted
           upstream and are not defects: a ray that passes through the seabed still
           blends its below-seabed samples over the seafloor unless the field masks them
           (all three species fields here set mask_below_seabed, so they do not), and
           while the camera is INSIDE a box the fragment is depth-tested at the far box
           face, so the seafloor erases the volume across its whole footprint. Backing
           the camera out of the box restores it.
           ================================================================ -->
 
      <!-- The Calanus layer over the survey box. The band is centred at 28 m with a 9 m
           standard deviation, so a 0..60 m box holds all of it; 1.25 m of vertical
           spacing resolves the band, 5 m horizontally is plenty for a field that has no
           horizontal structure worth the name inside this box. VisibleRange starts at
           3.0e-4, just above the 1e-4 to 3e-4 background, so the ambient water is clear
           and only the layer is drawn. -->
      <Lib
         LibName         = "environment"
         SimObject       = "VisualScalarVolume"
         Name            = "CalanusVolume"
         FieldName       = "Calanus"
         BoxCentre       = "150, 50, 30"
         BoxSize         = "420, 260, 60"
         Resolution      = "84, 52, 48"
         ReferenceLength = "20"
         ColorRange      = "3.0e-4, 2.2e-3"
         VisibleRange    = "3.0e-4, 2.6e-3"
         AlphaRange      = "0.02, 0.30"
      />
 
      <!-- The demersal cod layer. This is the one that shows what a seafloor-referenced
           field buys: the layer is 5 m above the BOTTOM, so it rides the rolling
           bathymetry, and the box is a flat 110..154 m slab that the layer wanders
           through rather than a band the layer fills. 1 m of vertical spacing, because
           the band's sigma is 2.5 m. -->
      <Lib
         LibName         = "environment"
         SimObject       = "VisualScalarVolume"
         Name            = "CodVolume"
         FieldName       = "Cod"
         BoxCentre       = "150, 50, 132"
         BoxSize         = "420, 260, 44"
         Resolution      = "84, 52, 44"
         ReferenceLength = "10"
         ColorRange      = "1.0e-3, 1.2e-2"
         VisibleRange    = "1.0e-3, 1.6e-2"
         AlphaRange      = "0.03, 0.45"
      />
 
      <!-- A vertical temperature section along the track, from the surface to the
           seabed. It is 1900 m long and the survey box is 300 m of that on purpose: the
           water mass front sits 1000 m out and inside the box it is very nearly a
           constant -0.53 degC offset, so a slice cropped to the box would show the
           thermocline and nothing else. Run north far enough and the front, and the
           warm lens on top of it, are both in the picture, and it is then visible that
           this mission samples one corner of a much larger structure.
 
           It sits at the EASTERN edge of the survey box, flush with the far face of
           the two volumes, and not down the middle of it. That is not cosmetic: the
           slice is nearly opaque where the field is strong, so anywhere it stands
           between the camera and the vehicle it is a wall rather than a section. At the
           far face it is a backdrop, and from the default camera - which is south of
           the box at East 0, looking north - it is off to the right and seen at a
           glancing angle. 4 m by 1 m sampling is 66.5 k cells, a tenth of either
           volume. -->
      <Lib
         LibName      = "environment"
         SimObject    = "VisualScalarPlane"
         Name         = "TemperatureSection"
         FieldName    = "SeaTemperature"
         Position     = "600, 180, 70"
         LocalX       = "1, 0, 0"
         LocalY       = "0, 0, 1"
         Size         = "1900, 140"
         Resolution   = "475, 140"
         ColorRange   = "6.5, 9.8"
         VisibleRange = "6.2, 10.4"
         AlphaRange   = "0.45, 0.80"
      />
 
      <Lib
         LibName                   = "fishery"
         SimObject                 = "Auv/Vehicle"
         Name                      = "Auv"
         Length                    = "2.0"
         Diameter                  = "0.20"
         AverageDensity            = "1025"
         Rho                       = "1025"
         COG                       = "0, 0, 0.020"
         AddedMass                 = "3.2, 58, 58, 0.1, 19, 19"
         ThrustPerPower            = "0.30"
         MaxPropulsionPower        = "120"
         HotelPower                = "15"
         ThrusterPos               = "-1.0, 0, 0"
         AxialDragCoeff            = "0.25"
         CrossFlowDragCoeff        = "1.1"
         RollDampingLinear         = "4.6"
         RollDampingQuadratic      = "1.0"
         RudderArea                = "0.016"
         RudderAspectRatio         = "1.25"
         RudderPos                 = "-0.85, 0, 0"
         RudderClMax               = "0.8"
         RudderAoaMax              = "0.436"
         RudderCd0                 = "0.02"
         RudderOswald              = "0.9"
         MaxRudderAngle            = "0.35"
         InitialBattery            = "1.8e6"
      />
 
      <!-- A short three leg survey box: 300 m north, 100 m east, 300 m south. About
           700 m of track, that is about 470 s at cruise, so the mission finishes well
           inside the 600 s run and EndMission has time to latch. RecoveryPosition is
           the last waypoint, so no return leg is added. -->
      <Lib
         LibName               = "fishery"
         SimObject             = "Auv/UndulateMissionControl"
         Name                  = "Mission"
         WaypointsX            = "300, 300,   0"
         WaypointsY            = "  0, 100, 100"
         RecoveryPosition      = "0, 100"
         WaypointRadius        = "15"
         CruiseSpeed           = "1.5"
         BatteryAtEndOfMission = "2.7e5"
         LoopWaypoints         = "0"
         UpdatePeriod          = "1.0"
      />
 
      <Lib
         LibName             = "fishery"
         SimObject           = "Auv/UndulateDiveControl"
         Name                = "Dive"
         MaxDepth            = "45"
         MinDepth            = "10"
         MinSeafloorDistance = "15"
         DiveAngle           = "0.349"
         DepthHysteresis     = "4.0"
         UpdatePeriod        = "1.0"
      />
 
      <!-- Heading. POSITIVE gains; Ki deliberately zero, because a symmetric vehicle
           needs no steady rudder and an integrator winds up through the saturated part
           of a large step. -->
      <Lib
         LibName           = "fishery"
         SimObject         = "Trawl/PIDController"
         Name              = "HeadingPid"
         InitOutput        = "0"
         ProportionalGain  = "0.4"
         IntegratorGain    = "0"
         DerivativeGain    = "0.6"
         MinOut            = "-0.35"
         MaxOut            = "0.35"
         MaxDerivativePart = "0.2"
         MaxIntegratorPart = "0.1"
      />
 
      <!-- Pitch. NEGATIVE gains, because a positive RudderDive lowers the pitch.
           IntegratorGain -0.08 as in tests/in/Auv/Auv_Undulate_in.xml (R112, FISH-0095):
           with the Munk moment and the hull lift the standing deflection at DiveAngle is
           0.093 rad rather than 0.20, and -0.15 overshot each DiveAngle step by 27 %. -->
      <Lib
         LibName           = "fishery"
         SimObject         = "Trawl/PIDController"
         Name              = "PitchPid"
         InitOutput        = "0"
         ProportionalGain  = "-0.6"
         IntegratorGain    = "-0.08"
         DerivativeGain    = "-0.8"
         MinOut            = "-0.35"
         MaxOut            = "0.35"
         MaxDerivativePart = "0.2"
         MaxIntegratorPart = "0.3"
      />
 
      <!-- Speed. POSITIVE gains. InitOutput 30 W is the cruise feed forward for the
           commanded 1.5 m/s; change it with CruiseSpeed. -->
      <Lib
         LibName           = "fishery"
         SimObject         = "Trawl/PIDController"
         Name              = "SpeedPid"
         InitOutput        = "30"
         ProportionalGain  = "200"
         IntegratorGain    = "10"
         DerivativeGain    = "0"
         MinOut            = "0"
         MaxOut            = "120"
         MaxIntegratorPart = "120"
      />
 
      <!-- The CTD head is on the nose, ahead of the hull wake. The time constants and
           the noise are an SBE49 FastCAT. NoiseSeed is non zero, so the instrument is
           realistic rather than perfect, and the hand rolled generators make it
           reproducible across compilers all the same. -->
      <Lib
         LibName                  = "fishery"
         SimObject                = "Auv/CTD"
         Name                     = "Ctd"
         TemperatureFieldName     = "SeaTemperature"
         SalinityFieldName        = "SeaSalinity"
         MountOffset              = "0.8, 0, 0"
         SamplePeriod             = "0.0625"
         TemperatureTimeConstant  = "0.07"
         ConductivityTimeConstant = "0.06"
         DepthTimeConstant        = "0.02"
         NoiseSeed                = "7"
      />
 
      <!-- The silhouette camera images 1.43e-4 cubic metres per frame, so a single
           frame is almost pure noise: at the design Calanus peak the rate is 0.22
           particles per frame after the capture and detection efficiencies. Forty
           frames, that is 10 s and about 15 m of track, brings it to about 9 counts
           and a 33 percent counting error. That is why the aggregation exists. -->
      <Lib
         LibName              = "fishery"
         SimObject            = "Auv/SilCam"
         Name                 = "SilCam"
         SpeciesFieldName     = "Calanus"
         MountOffset          = "0.6, 0, 0"
         FrameRate            = "4.0"
         FramesPerOutput      = "40"
         SampleVolume         = "0.040, 0.055, 0.065"
         CaptureEfficiency    = "0.85"
         DetectionProbability = "0.92"
         FalsePositiveRate    = "0.05"
         SizeErrorStdRelative = "0.08"
         NumSizeBins          = "8"
         SizeBinMin           = "0.0005"
         SizeBinMax           = "0.0060"
         NoiseSeed            = "11"
      />
 
      <!-- MountDirection must be given. SensorFrame defaults a beam axis to body x,
           which is 90 degrees wrong for a nadir sounder.
 
           NoiseSeed is non zero on purpose, and it is not only about realism: with the
           seed at zero the instrument adds no range noise, no speckle AND no noise
           floor, and the noise floor is precisely what makes the Calanus layer
           invisible at 38 kHz and visible at 120 and 200 kHz.
 
           PingPeriod is 1 s rather than the 0.5 s default. One hertz is a realistic
           survey rate, it halves a file that would otherwise reach about 7.5 MB over
           this mission, and StepMax of 0.0125 s still puts 80 integrator steps in a
           ping interval. -->
      <Lib
         LibName           = "fishery"
         SimObject         = "Auv/Echosounder"
         Name              = "Echo"
         MountOffset       = "0, 0, 0.10"
         MountDirection    = "0, 0, 1"
         PingPeriod        = "1.0"
         MaxRangeDown      = "150"
         RayStepSize       = "1.0"
         RayTolerance      = "0.01"
         SpeciesFieldNames = "Calanus, Mackerel, Cod"
         BinSize           = "1.0"
         SvPrecision       = "2"
         FlushEveryNPings  = "20"
         WriteFile         = "1"
         FileName          = "AuvSurvey_echogram.csv"
         NoiseSeed         = "13"
      />
   </OBJECTS>
 
   <INTERCONNECTIONS>
      <Connection Mission.Pos = "Auv.Pos" Mission.Battery = "Auv.Battery" Mission.Heading = "Auv.Heading"/>
      <Connection Dive.Depth  = "Auv.Depth" Dive.Altitude = "Auv.Altitude"/>
 
      <Connection HeadingPid.Reference = "Mission.Heading" HeadingPid.Measurement = "Auv.Heading"/>
      <Connection PitchPid.Reference   = "Dive.PitchRef"   PitchPid.Measurement   = "Auv.Pitch"/>
      <Connection SpeedPid.Reference   = "Mission.Speed"   SpeedPid.Measurement   = "Auv.Speed"/>
 
      <!-- Two cycles, neither of them algebraic: Mission, Dive and PidController all
           return output members written only in AcceptedStep, so nothing in either
           cycle has direct feedthrough. -->
      <Connection
         Auv.RudderYaw     = "HeadingPid.Out"
         Auv.RudderDive    = "PitchPid.Out"
         Auv.ThrustPower   = "SpeedPid.Out"
         Auv.ExternalForce = "0,0,0"
      />
 
      <!-- The instruments ride the vehicle. Each one rotates its own mount offset and
           beam axis out of the body frame, so the pose is all they need. -->
      <Connection Ctd.Pos    = "Auv.Pos" Ctd.Quater    = "Auv.Quater"/>
      <Connection SilCam.Pos = "Auv.Pos" SilCam.Quater = "Auv.Quater"/>
      <Connection Echo.Pos   = "Auv.Pos" Echo.Quater   = "Auv.Quater"/>
   </INTERCONNECTIONS>
 
   <INITIALIZATION>
      <InitialCondition
         Auv.Pos      = "0,0,20"
         Auv.Quater   = "1,0,0,0"
         Auv.LocalVel = "1.5,0,0"
         Auv.Omega    = "0,0,0"
      />
   </INITIALIZATION>
 
   <SIMULATION>
      <Timing TStart="0" TEnd="600"/>
      <Integrator Method="RK45_i" NumCores="1">
        <StepControl StepMax="0.0125" StepMin="1e-8" AbsTol="1e-4" RelTol="1e-4"/>
      </Integrator>
   </SIMULATION>
 
   <OBSERVERS>
      <!-- A range, not a list: TOutput is start:delta:end. -->
      <FileOutput TOutput="0:0.5:600" Select="objects:all"/>
 
      <!-- Without this the scenario never draws. An OBSERVERS block is honoured
           exactly as written, so once one is present the legacy default set - which is
           where the live view used to come from - is gone, and FhVis runs the model
           headless. In a headless binary the element is a silent no-op: FhSim never
           attaches a live view, makeVisualizationObserverFromModel returns nothing and
           nothing is registered, which is why this file can carry the observer and
           still be the regression input byte for byte. -->
      <Visualization frameRate="25"/>
   </OBSERVERS>
</Contents>