colinsheppard / instream Goto Github PK

Here's a cleanup you could please do to inSALMO to keep it from driving people nuts in Windows. I globally changed "ReddSuperImp_" to "ReddSuperimp_" (small "i" instead of "I", because when one fish builds a redd on top of another redd, it is "superimposition", although nontechnically it is also a super imposition.)

But somehow there are still two files in there called ReddSuperImpXXX.h. When you un-tar the package in Windows, it overwrites the good file ReddSuperimpxxx with the old file ReddSuperImpXXX. THen it won't compile.

So if you could just find and delete the two files with capital I in their name...

Change the column label from "InitialNumberOfEggs" to ""InitialViableEggs"

Output file Redds.out: (a) get rid of header for “Transect Number” (transects no longer exist in this version. they were eliminated from the output lines but not the header). (b) make the output .csv format.

Will be fixed first in inSALMO. Code to open the file and write headers is in TroutModelSwarm and code to write each line is in Redd.m.

Is there a place in the repository for the standard input files? Could be a separate branch or something for each model. That would be nice.

This was fixed in inSALMO but not instream:

So, SteveJ had cleverly hidden a limit on the number of flows you can include in the hydraulic file. Look for this line in HabitatSpace.m:

flow = (double *) [habitatZone alloc: 24*sizeof(double)];

He's allocating an array of flows with length 24.

I fixed my copy; can you fix the official version? Let's replace the 24 with a local variable declared and set at the top of the method, called something like "maxFlowsInFile" and set it to 100.

HabitatSpace drop: It looks like we’re trying to drop a lot of stuff that is never used (cornerCoords etc.) or drop stuff twice. And in PolyCell -drop we’re trying to drop something we’re not using, which causes warning statements at end of run.

Clean out junk we do't need, etc.

(Submitted for inSALMO too)

Clean up getNeighborsWithin: to fix movement among reaches. Copy from inSALMO.

HabitatSpace.hm: add methods addUpstream(downstream)CellsWithinRange: toList
Overhaul getNeighborsWithin
Fishcell.hm, TroutObserverSwarm.m: add tag method to tag neighborsWithin:
Put the cell and neighbor tagging methods from inSALMO into the habitatSpace and cell probe displays.

Let's go with my old favorite: PeachPuff

Both FishCell and PolyCell have methods getPolyCellDepth. Put in FishCell. Velocity update is in FishCell but depth update is in PolyCell. Put both in FishCell. PolyCell should only have geometry and color stuff. Look for other redundancies/confusion. (FishCell is a subclass of PolyCell.)

Fixed in inSALMO -- its issue #21. Should be able to copy PolyCell (not FishCell) from inSALMO.

Incorporate changes from inSALMO into instream

See which of the optional output files listed in Sect. 24 of the inSTREAM manual need fixing-- e.g., they no longer work, or need their format updated to remove "transect number" etc.
Also submitted to inSALMO.

DO NOT round floats off; leave in scientific notation. (We need the precision for code testing.)

This may have already been fixed: HydraulicFileReader crashes if there is white space (tabs, at least) at the end of a line.

Also: the documentation says that the code makes these checks, not all of which are implemented yet:
The software makes several checks when reading in the cell hydraulics
file:
• The cell numbers must correspond with those in the cell geometry
file: each cell defined in the geometry file receives hydraulic input
and there are no cells in the hydraulic file but not in the geometry file.
• The flows on line 3 must be in order of increasing magnitude.
• The number of values on each row of hydraulic input (file lines 5 and
higher) must equal one (the cell number) plus two times the number
of flows defined line 3.
• There must be no negative velocities.
• Any negative depth values are set to zero (some hydraulic models as-
sign negative depths to locations above the water level).

(submitted for inSALMO too.)

TimeSeriesInputManager.m has an optional method printDataToFileNamed. This method opens an output file and forgets to close it (just found this problem in a completely different code).

I think we just need:

fclose(filePtr);

right before the return statement.

(This class is part of EcoSwarm, which I need to update as we finish up this version of instream.)

In PolyCell.m, delete line 634 (which multiplies cell area by 10,000). It’s wrong because we already converted polygon coordinates from meters to cm.

Add new optional output file that reports the state of each fish. It should be named "IndividualFishReport.csv"

The file reports one line per live fish, on each output date. The line reports: Scenario, Replicate, date, fish ID, reach name, cell number, species name, age, length, weight, condition.

The code to write the file lines is in a temporary version on Steve's laptop Linux at:
SwarmApps2.2.5/inSTREAM_4.2.2_AllFishOut/TroutModelSwarm.hm. (This will require slight modification because it's an old version with transects and cell numbers; now we do not have trnasect numbers. This temporary version called the file "AllFishReport", but let's use "IndividualFishReport".)
It requires:
a) Adding a unique integer ID to each fish, (done; new IVar in Trout.h; new fish counter and assignment in TroutModelSwarm.m createNewFishWith...)
b) Optional (IFDEF) statements to open and write to the file: see model swarm method printAllFishReport (and how outputting the report was added to the schedule).

Still to do: create separate file open method that writes header information to file and decides whether it should be overwritten and appended; can use openReddSummaryFilePtr as a template.

Added to software documentation 11/30/2011

Incorporate the checks mentioned in the inSTREAM documentation. (or suggest other ones). See Sect. 3.4 of attached documentation (in email from Steve 4/12/11).

Double check if second for loop from the bottom of the method is actually doing anything, remove if not.

I am pretty sure we're no longer using the class StationObject ..... so remove it from code archive.

The reddMortalityFile is huge and not easy to use... Let's make it an optional file, controlled like the other optional files from TroutModelSwarm.h. I'll update the documentation. Hardwire its name to Redd_Mortality_Out.csv and remove from Model.Setup.

Make sure we're using the same versions of TimeSeriesInputManager, and BreakoutReporter, from inSALMO.
TimeSeriesInputManager should figure out if files are in .csv format
BreakoutReporter should have two create methods, one for .csv; and produce breakout results in one instead of multiple columns.

PolyCell.m updatePolyCellDepthWith: Negative depths from interpolator – Can happen at flow < lowest flow in interpolator. Fix so they are set to zero.

(This was already done in inSALMO)

Make it allowable for the cell data file and the hydraulic data file (not cell geometry file) to be in .csv format

These files are identical to those in inSALMO, so can copy insalmo code for reading them.

In Experiment.Setup, allow "InstanceName NONE" to also accept "None" and "none" etc.
(Should have been done for insalmo.)

The fish parameter fishEMForUnknownCells is used only for barriers, which we removed from this version. So remove the param from Trout.h and the trout parameters input file.

FishCell has unused method calcDepthAndVelocityWithIndex. Why don’t we use it? (To understand this, see documentation for EcoSwarm InterpolationTable)

Done in inSALMO -- see #20. Seemed to speed it up quite a bit.

Do a little Valgrinding etc. when code is about done to look for new leaks.

When inSALMO is stable, copy its habitat input files as example files.

Modify the optional move report as done for insalmo:
a) Give each fish a unique integer ID, in [troutmodelswarm createNewFishWith...]
b) Fish update variables for prevCell, prevReach whenever they move (look for places where moveToBestDest is called)
c) MoveReport includes fish ID, current and previous cell and reaches.

Some optional output files (e.g., spawn_cell) include a FishID variable. Currently these are the fish's address, created by printing "self" with format "%p". I found out that these addresses are not unique: when one fish dies, another often gets its address. Instead, use the integer fishID number implemented to address issue 22. Remember to change the format from %p to %d.

I don't think we really need the model setup parameter for number of species. Instead, can we set it from the Species.Setup file? It could be that we're already doing that. Let me know if it turns out not to be so simple.

Observer.Setup: raster zoom factor and rasterResolution seem to do nothing and need to be cleaned out.

We want to get rid of rasterResolution because it is never used for anything. rasterResolution is used to set a variable polyRasterSize but that variable is never used. rasterResolution is also passed from TroutObserverSwarm to TroutModelSwarm (in two places) and frmo there to HabitatManager to HabitatSpace to PolyCell. But none of these seem to use it.

Raster zoom factor: So, our display rasters are Swarm class ZoomRaster, which supposedly lets you zoom in and out but we never implemented the code (which would be pretty complicated, apparently) to draw stuff correctly. I think the best thing would be to change its class to Raster and get rid of ZoomFactor. DONE in INSALMO- no problem.

You could also trim out some of the print statements that aren't really necessary any more. It's nice to see the major steps as the model gets built, and a few lines at the end of a run to know whether the experiment is all done, but other stuff we could get rid of.

1.Faster method for identifying neighbor cells within some radius, for habitatSpace getNeighborsWithin:

Perhaps: each cell has sorted list of cells by distance from self. Use Swarm arrays: one for cells, one for distance to. Use “OFfsets”: file:///C:/Swarm/Documents/Swarmdocs-2.2-objc-set/swarm.collections.offsets.protocol.html

Low priority

In TroutObserverSwarm.m, need to clean up the probe maps. They contain some variables that don't exist any more. One is:

probeMap = [CustomProbeMap create: aZone forClass: [TroutRedd class]
withIdentifiers: "Species",
"numberOfEggs",
"emergeDays",
"fracDeveloped",
"spawnerLength",
"transect",
"cellNo",
":",
We don't have transects now, and "cellNo" should be "cellNumber".

Even after fixing that, i still get warnings about null probes when run in graphics mode, so there's more wrong. (I can do this if you want.)

I need to produce a standard input set...
Update all the setup files, especially headers

I emailed these to Colin ~Dec 5 for him to commit

We're no longer using the model variable populationInitializationDate so delete from code and Model.Setup. (I'm pretty sure we don't use it.)

///////////////////////
Errrr, actually this true only for salmon model, not inSTREAM. The pop init date is still how we determine which input in the initial population file to use for a particular model run.

I re-opened this just to make sure the change was not made. Steve

It wouldn't be too much work to update the way that they optional output files are specified so we don't have to require the user to recompile in order to opt for them.

For example, there could be additional parameters in Observer.Setup which control each of the optional files.

The code for barriers will not work in this version. We need to:
a) If barriers appear in Reach.Setup per sect. 20.2.3 of inSTREAM documentation, then raise an error that says barriers don't work in this version.
b) Remove Barrier.hm from code directory and Makefile

ObserverSwarm –writeFrame: un-comment the block “if([habitatRasterList getCount] > 0)”. Uncomment “id raster = nil;”.
ObserverSwarm –buildObjects: after this statement:
[habitatRasterMap at: habitatSpace insert: polyWorldRaster];
add this one: [habitatRasterList addLast: polyWorldRaster];

Submitted also for inSALMO - same method should work for both models.

in inSALMO we made some format changes in output files--floats are not always in scientific notation.

a) Are these changes made in standard output files for inSTREAM ? If not, talk to Steve. They seem less important with the new breakout reporter format.
b) but not in the optional output files, where we are likely to need full precision for code testing.

Get rid of warning:

EcoAvenger.m:85: warning: assignment from incompatible pointer type

I'll do this.

When inSALMO is stable, copy its drawSelf methods etc. so fish and cells and redds are drawn the same way. Implement the same shading code as inSALMO.

Colin: did we decide that fish are drawn as pixels while they are age 0, and rectangles when age > 0? That's what I'm putting in the documentation.

Check ExperSwarm.m and ExperBatchSwarm.m for bug that was in inSALMO, commit cdf9526d3d Simple logic error in check whether InstanceName = None or none. SHould be "==" instead of "!=".

The method FishCell setHabShelterSpeedFrac is not called, so the cell variable habShelterSpeedFrac is never updated from its initial value of zero. Best solution seems to be to get rid of this cell variable and have FishCell just get this reach variable from its space.

See insalmo issue 45

Add output files for the limiting factors tool. Per documentation on trout LFT done 11/23/11, replace the salmon output with the following method:
To make results easy to interpret, the LFT analyzes only two outputs from inSTREAM: abundance and biomass of adult trout. The specific measures of abundance and biomass are set by two LFT parameters. The parameter resultsAgeThreshold defines the age (years, with age 1 attained on the individual’s first January 1) of simulated trout included in the LFT results: trout with age equal or exceeding resultsAgeThreshold are included in results. The value of resultsAgeThreshold can be set to reflect the age at which trout at a study site typically spawn or (if angling is an important objective) become catchable.
The second parameter defining results is resultsCensusDay (day of year, in mm/dd format): LFT results are “censused” from the model on this day of each simulated year, and then averaged over the model run.
As an example, if resultsAgeThreshold is set to 1 and resultsCensusDay is set to 10/01, then the LFT will measure the number and total biomass (g) of all age 1 and older trout on each October 1st in the simulations. The results analyzed for each scenario will be the mean, over all simulated years, of these October 1 values.

Comment out those damn depth- and velocity-use histograms that have been annoying me for decades. (I will do this.- Steve)

The fix in this commit in insalmo:
colinsheppard/insalmo@e4ff8eb
needs to also be made in instream. It's a significant bug. Be careful because there are 4 places with nearly identical code.