Monday – 08/30/10

• Preparing committee member research status update.
• Pick up and install additional 8 GB RAM! After crashing the computer several times last night, this is a welcome addition.
• New publication out today in Biotropica titled ‘Distinct Leaf-Trait Syndromes of Evergreen and Deciduous Trees in a Seasonally Dry Tropical Forest’ located in Western Mexico in which we investigate biotic and abiotic selective pressures on leaf traits, including herbivory, phylogenetic signal, and phenologic status.

Sunday – 08/29/10

• Refining direct and diffuse PAR models.
• Learning to write IDL code using multi-threading (using multiple processors simultaneously) techniques for improved computation speed.

Saturday – 08/28/10

First direct path forest structure rays are returned from direct par model. Needs much refinement but nice to have the first chunk functioning. Below see for 45 deg. elevation and 37 azimuth with a max distance of 100 horizontal meters.

Example path through a randomly generated forest. next step is to deal with azimuth and elevation issues beyond 180 degrees, etc…

 

Considering now various approaches to deal with light extinction, including a inverted Beer-Lambert exponential extinction equation where “y = e-k*LAI”

Friday – 08/27/10

• Working on diffuse PAR model. Who knew how simple it was to calculate Euclidean distances within a variable size array, meaning the diffuse model can be adjusted on the fly (using only the code below), and therefore much better parameterized to actual microclimate conditions. Joy!

 

x_2 = (findgen(nx) - xcen) ^ 2     ;X distances (squared)
y_2 = (findgen(ny) - ycen) ^ 2     ;Y distances (squared)  
im = fltarr( nx, ny, /NOZERO)     ;Make uninitialized output array

for i = 0L, ny-1 do begin                      ;Row loop
       im[0,i] = sqrt(x_2 + y_2[i])      ;Euclidian distance
endfor

thanks to NASA GSFC for base code! Here’s a small example from the procedure…

Here is output in cell fractions for a 5x5 array.

2.82843      2.23607      2.00000      2.23607      2.82843
2.23607      1.41421      1.00000      1.41421      2.23607
2.00000      1.00000     0.000000      1.00000      2.00000
2.23607      1.41421      1.00000      1.41421      2.23607
2.82843      2.23607      2.00000      2.23607      2.82843

 

Now how to make this work except within a variable size cylinder?

 

excellent, here it is for 4 cells above our z base in meters, corrected for different x and z dimensions (0.56 and 0.15 m).

1.69375      1.38852      1.27059      1.38852      1.69375
1.38852     0.993579     0.820731     0.993579      1.38852
1.27059     0.820731     0.600000     0.820731      1.27059
1.38852     0.993579     0.820731     0.993579      1.38852
1.69375      1.38852      1.27059      1.38852      1.69375

 

On to the direct PAR model for a change of pace. Here is the preliminary brainstorm, a load of basic trig.

Thursday – 08/26/10

Continuing advances in the brutal world of LiDAR modeling. Memory is such a big problem!

 

Wednesday – 08/25/10

• Ongoing programming work. Now working to read into memory the waveform LiDAR image in small enough chunks to fit into RAM…However, I did upgrade to 12 GB ram after realizing that 6 was completely inadequate. Making progress and hopefully today will begin to code the actual direct and diffuse PAR models based on the waveform structure data.

Random screenshot of the IDL programming environment.

Tuesday – 08/24/10

• Refine the data read in code. Need to code it able to both run iteratively through the entire image for the full on model, with adjustable temporal resolution and output parameters, and also to return specific location data.
• Start the microclimate model.
• Start to consider the approach for linking the leaf area density profiles and branch vs... leaf offsets with the hyperspectral and waveform data.

Example screen shot of one binned waveform, resolution 0.15 m vertical, 0.56 horizontal.

 

• Also prep for field work tomorrow, unless I’m on a programming roll, then will stay here…

Monday – 08/23/10

• Working on IDL code to read in a 3D  image subset based on specified UTM coordinates. Necessary as the waveform LiDAR image is 6 GB in size. Not as easy as it seems…  (1:45 pm; currently have a working script to read in a portion of an image, now improving it to read in correct area for microclimate modeling work)
• Prepping gear for field work starting tomorrow through Thursday collecting LiCOR curves.

Sunday – 08/22/2010

• Revising solar positioning software, (finished)
• Working on IDL code to convert from UTM to Lat long coordinates. This is necessary to link my waveform LiDAR voxels ('”volumetric pixels”) 3D locations, which are in UTM coordinates, to my solar positioning model which uses Latitude Longitude Altitude. (finished)

 

Yes! Just finished and tested to sub-meter conversion accuracy versus online sources. View of part of the code below…

Friday & Saturday – 08/20-21/10

• Finished R stats and figure generation for Peru and Bolivia datasets for Angelica’s presentation at IUFRO S. Korea.
• Completed preliminary version of solar positioning software. The code uses JPL ephemeris data for maximum accuracy, reads rectangular universal coordinates of earth and sun locations and converts them to celestial equatorial  (right ascension and declination) compensating for light-time displacement and gravitational effects of other planetary bodies, then converts from celestial to topocentric using lat, long and altitude for location adjusting from true to apparent location caused by refraction, then adjusts from Gregorian to Julian date and compensates for precision issues with double floating point precision data in IDL, maintaining < second temporal accuracy via normalizing all Julian dates to the epoch 01.01.2009 at 00:00:00 Universal Time.
• Accuracy/Precision of the code has been tested extensively versus the JPL Horizons online ephemeris generator and has < 0.2 degrees of Elevation and Azimuth from 1950-2050, more than adequate!

* Image from http://www.bsu.edu.

Wed-Thur: 08/18-19/10

Mired in the depths of IDL astronomy programming trying to create a high precision solar path model, an important module within my microclimate / physiology model.

Tuesday – 08/17/10

• Finished new LiCOR programs,
• Installed new internal hard drive (1.5 TB) for upcoming modeling work,
• Started coding sun tracking IDL module,

Mon – 08/16/10

• Finish entering foliar data (done),
• Calculate leaf areas and input (done),
• Check LiCOR data quality for each sample (9; done) – have 5 valid sets from traverse 1, leaving one day of understory LiCOR work there,
• Update LiCOR programs,
• Clean and organize low & high tower climate data,
• Clean and organize first interior climate dataset,

 

Work with Angelica to develop R stat codes for Bolivia dataset.

 

Calculating scanned leaf areas using magnetic lasso tool in Photoshop at 600 dpi…

Sat & Sun – 08/14-15/10

• Finished entering leaf/trunk offset data,
• Entering foliar data and organizing,

Friday – 08/12/2010

Finally came up with a working method to rain proof the LiCOR head while taking measurements, thanks to some great new clamps found at Home Depot.

Next steps are:

1. to get the interior forest station running. (Done)
2. finish entering leaf vs. wood data. (in progress)
3. Finish calculating leaf areas for collected leaves, (later i guess)
4. update LiCOR programs. (tomorrow i guess)

Sensors up and running:

One of 4. this one is setup to show the lowest strata micro-climate.

Final weatherproof setup, charged desiccant and ready for the next 10 days of data collection.

Wed – 08/11/2010

• Heading to town, truck seems to have a broken strut following field work yesterday. Also getting a much needed oil change.
• Following this I will:
• finish the LiCOR programming,
• finish the weather proof cover for the LiCOR,
• continue organizing quality checking the tower climate data,

Tuesday – 08/10/2010

• Spent morning getting tower climate data raw files which need to be cleaned and organized.
• Differentially processed first two of ten plot final GPS stake coordinates.
• Tree down in access road to field plots so went town instead to get parts for LiCOR rain cover.

Saturday – 08/07/10

Installed interior forest micro climate stations. Need to finish wiring them tomorrow morning, then spend the rest of the day programming R stats for Angelica’s data analysis.

Friday – Aug 6th 2010

• Back from a family visit.
• Download data form all towers, check desiccant levels, battery charge,
• Pack for install of microclimate stations tomorrow,
• Enter data collected last week,
• Help Angelica with R coding for her stats,
• Finish review for FORECO LiDAR article,

Saturday – 10/24/10

Leaf oven works perfect. Min and max temperatures over the last 24 hours were 55 and 62 C.

 

TO DO:

• - a way to attach them to the cuvette (big snaps?, bungee cord? ),
• - make new dynamic par that extends for 3 minutes instead of 2 (every 2 seconds taking a measurements),
• - change par curve program to include one 800 par reading at the end after I have clicked abort etc, using the same protocol. this will be aborted if I go to the very high light levels, but will be useful for a full light curve if I stop too early. Co2 curve seems to be fine as is.
• - measure areas of all leaves in Photoshop,
• - write on envelopes the measured area, and the date when put into the drier and final weight at IPIF,
• - make revised schedule of dates and LiCOR data collection,
• - enter offset data into excel (~350),
• - enter foliar data into excel,
• - enter new azimuth data for lad profiles into excel
• - order 1.5-2 TB internal hard drive for computer (confirm that it will fit) – 7200 rpm, etc…

How to attach this umbrella to the tripod….harder than it first appeared.

Friday – 07/23/10

Finished leaf drying oven, making refinements to get a steady temperature of 55-65 C. Seems to be working well with a couple of 60 watt bulbs.

 

Trying to make a final weatherproof system so I can LiCOR in the rain, one of the normal conditions here. How to attach this umbrella to a tripod in a more permanent way?????? This is the question I am heading to Home Depot with at this moment.

 

Entering data from our 250 leaf and trunk offsets. Looking forward to combining these with the differential GPS points and seeing if it is possible to distinguish them in the waveform data.

Tuesday – Thursday 07/23/10

Working in the field collecting points, verifying the data using SibII, revising data collection procedures, all seems to be almost good.

 

Heading to field with LiCOR in weatherproof backpack and puppy, Kai Turbo.

 

My two field assistants hard at work.

Tuesday – 07/20/10

Spent the last 2 days collecting LiCOR curves in the understory. All seems well, making some improvements to the protocol and now checking the data quality to make sure all is well. Need to finish the leaf oven today and get the collected leaves starting to dry.

Heading to the field, LiCOR strapped to my back with our fancy weatherproof backpack, the tripod on my front. Second trip necessary to carry the battery supply and additional field gear (including the laser georeferencing system).

System working collecting a curve sequence from a invasive ginger in a gap area in the ‘valley of the giants’ forest. Two huge trees fell in the last few days in this area close to where we are working, strange…. (and a bit scary!).

Saturday – 07/17/10

Heading to field to get another GPS point of traverse 8 and finish as many traverse locations of branch/leaf offsets as possible.

 

Still need T5, T8, T9, T10, T11 for 20-40 offsets each location.

Friday – 07/16/10

Testing data downloaded yesterday. As expected for wind data the two sensor stations in the higher more open area are highly correlated (1 and 3) while those in different understory environments are not (2 and 4).

PAR data shows as expected readings, with the sensor in the deepest understory having the lowest readings throughout the day, those higher up having more similar readings. Sensor 2 is also lower down in the canopy and in a more obstructed area.

Sunfleck data is shown below, shows 3 minutes of 2 second PAR readings at the start of every hour. As expected the greatest variability is in the sensor located in the forest understory area of greatest variability (middle height) – robot 2.

 

Heading out to the forest now to:

1. collect final GPS points (T6 and T8)
2. Collect azimuth offsets for T4
3. Get the position of the interior forest microclimate stations (henceforth termed ‘robots’)
4. Collect 3-5 more traverse sites of leaf and trunk offsets

Thursday 07/15/2010

• Test Trupulse mounting bracket and calibration procedure in the field – tested and calibration procedure is successful and seems to deliver precise readings
• Test previous azimuth readings – all tested and a new procedure for locating them on the traverse was developed
• Final GPS points (2 today, 1 tomorrow) – 3 hours each – finished 2 GPS points
• Collect locations of transect start and end trees - upcoming
• Collect 30 leaf and 30 branch/trunk locations using Trupulse per traverse site – finished T1,2,3
• Bring pole clippers, reflectors, 3 tripods, trupulse, GPS, extra batteries, measuring tape, plumb bob, rite-in-the-rain and pen,
• Verify data downloaded by CR-10x is good, bring field laptop and download test section. – verified, all data looks good.
• Download yesterday GPS point data to computer and format memory card – completed, final point measured combined 7 hours of GPS data is good accuracy in 3D.

Angelica and our new friend who found us in the woods yesterday collecting leaf and trunk offset locations.

Wednesday – 07/14/2010

Heading up to reinstall all dataloggers and collect another 2 3 hour GPS points. The Trupulse mounting system should be arriving today via Express mail meaning tomorrow we can georeference the microclimate stations and start LiCOR work!

A quick overview of the lower top of canopy climate station shows an interesting diurnal pattern in the wind direction. Likely related to the fact we are situated on the side of Mauna Kea and sea breezes related to temperature differences (as shown in your temperature data)…

 

Now going to check the high elevation TOC sensor data, then pack bags for field work and be off. Beautiful day to spend in the woods!

 

The High TOC PAR sensor seems to be working well also, yes!

Tuesday – 07/13/2010

Downloading data from the loggers collected last night, too much rain to download in the field. Morning objectives are to:

1. return the loggers to the towers,
2. double check some Trupulse azimuth measurements,
3. get another 3 hours of GPS points (T3),
4. Verify BF3 data once it is downloaded (seems to be taking 2-3 hours to download, a ton of data). BF3 data has passed the test, see Figure below.
5. Following this I will finish the leaf oven and dry the desiccant. Update: completed.

Update: Well, the data from the CR10x was gibberish. Apparently something went wrong with the memory. Talked with Campbell and we reformatted the memory module and replaced the lithium battery (which I had to drive to Hilo to find), now it seems to be working fine. Will reinstall it tomorrow morning. The Trupulse mounting system finally shipped today and should be arriving on Wednesday. Also changed the desiccant packs which were surprisingly completely spent.

 

 

How beautiful is a CR10x mother board!

Monday 07/12/2010 – 4:40 am

More issues with the BF3, thought we had these resolved back in December 09. Oh well. Talked with our contact at the manufacturer and they are very willing to help. Following confirmation with the tower climber’s this morning, if needed, they will overnight a new unit for installation, which hopefully we could install later this week. The Trupulse issues, just ordered the mounting arm and bracket to avoid any of the compass magnetic interference issues we had from the tripod. Need to do a quick calibration in the field to verify previous data points. Beauty of the Hawaii time zone difference.

 

Today's objectives are to:

• Resolve the BF3 issue, – update: the BF3 sensor seems to be fine, I am downloading data now to verify, it was actually the CR upwelling radiation sensor that had condensation issues, what a relief!
• Order the Trupulse components, update: ordered, should arrive by Wednesday.
• Finish the leaf oven,
• 2-3 more hours of GPS points for traverse 8 and 3
• Change datalogger desiccants,
• Download middle elevation tower data, – have key from Hilo and now heading to the forest to download. am dropping off a new BF3 desiccant package if needed at later date in zip lock container.
• Recharge desiccant here in the house.

Saturday 07/10/2010

• Completing review of RSE manuscript.
• Figuring out sync between Trupulse and laptop ArcPAD via Bluetooth connection for rapid collection of wood and leaf areas within the forest for waveform LiDAR validation.

Friday 07/09/2010

An unexpected day spent on the Trupulse. This morning realized the azimuth readings were not as reliable as I required. 2 degrees of error at 28.4 meters is roughly a meter off! This is way beyond the 20 cm I am shooting for all leaf and climate readings, which matches the airborne waveform LiDAR spatial error of 15 cm.

 

So my first step was to develop an alternate method for minimizing spatial error during calculation of offset angles. This took me to Hilo Steel where, instead of 200 as Lasertech would have asked, I built one for 5 dollars. See below.

 

 

This was pretty great until I realized on the drive home it was made of solid steel. The aluminum I was using originally cracked when we tried to make the 90 degree angle turn in it. Man….

 

Then I checked the camera tripods we are using to mount the Trupulse and realized they have metal screws, resulting in potential errors of 2 degree if the Laser is placed within 5 cm, unfortunately the location where they are if positioned like a normal camera.

 

So after all the work at the steel works I ended up ordering the 220$ setup from Trupulse and hopefully it will arrive by Tuesday. This keeps the laser compass at least 10 cm from the tripod and therefore no magnetic errors, so all is well. See a picture below of the mounting equipment just ordered:

 

 

Sweet. hopefully this will reduce spatial errors associated with the laser from 40 cm to < 20 cm as hoped.

 

The beauty of the new system is we won’t have to deal with offset errors when calculated locations at inclinations, with the old system the laser actually offsets itself up to 10 cm, the new one pivots at the laser center point removing this error. :)

Thursday 07/08/2010

Finished first round of GPS points. Going to take some additional hours of points at locations having 3D errors greater than 30 cm.

Prepping for changing the location of the interior forest microclimate stations tomorrow.  Will likely require setting up a traverse within one of the upper elevation plots.

Reflection on car window with double back packs heading to the forest.

07/07/2010 - Wednesday

GPS points of final upper elevation traverse sites completed. Tomorrow morning will know if the 3D quality of the points is accurate (< 25 cm).

GPS setup. We’re now taking points for 3 hours per site in order to achieve < 20 cm accuracy (SD).

Lunch break from taking points and relocating the traverse sites.

Monday 06.05.2010

• Accuracy testing of the Trupulse. So far we are getting between 20-50 cm accuracy using one bounce off a intermediate reflector tripod at up to 30 m distance. Tomorrow we will finalize the methods and hopefully reduce the error a bit more.

Here is the testing setup around the house. Tomorrow we’ll test this out in the woods while collecting our near final GPS points. You can see the laser on a tripod, an intermediate reflector and the destination point, which will in the field be either a LiCOR cuvet or the interior forest microclimate stations.

Saturday 07/03/10

Two goals today. One is to verify the location of the GPS points being collected in the waveform LiDAR imagery, the second is work more on the laser methods.

 

The GPS points seem excellent, as their standard deviation shows. Their elevation (Z coordinates) match those of the modeled elevation from the LiDAR image. See below. The flows of lava seen in this topography below the points is pretty cool. now thousands of years old.

 

 

As does the general locations within the gaps and tree canopies, as seen below.

 

Friday 06/02/10

 

Discussing project with group of students. At the site of the lower climate tower.

Collected 2 more GPS points, down to 10 cm SD! Now have points at T1,2,3,4,5,6,8. Only have 7,9,10 and 11 left to do early next week.

 

Received the laser repaired and tuned up. Preliminary test shows it to be working perfect. Yes! Now working to develop the final methods and test the precision of the device for multiple bounces off reflectors.

Thursday 07/01/2010

• GPS points are XYZ accurate (SD in meters) of around: 0.1,0.1,0.4. Great! These points are some located in fairly dense understory so this is really good news. New tripods arriving today and the repaired Trupulse 360 laser, so this weekend will finalize the methods/components for georeferencing the microclimate and LiCOR collection data.

Traverse 5 stake location.

Wednesday - 06/30/10

• Georeferencing traverse sites in the rain today. Finished two points. Each point takes about 2-3 hours and gives a final spatial accuracy of <10 cm standard deviation in X and Y dimensions, and < 25 cm in the Z dimension. Post-processing takes place 24 hours following data collection.

Traverse 4 GPS setup.

Traverse 3 GPS setup.

Tuesday 06/29/2010

• First day conducting final differential georeferencing in the field with new methods and tools. A bit rainy and cold, but otherwise a good day.

Step 1. Set up tripod over stake. Use plumb bob and bubble levels on tripod to make sure it is directly vertically over the stake.

  View from above. See bolt and nuts used to modify a standard camera tripod for static GPS measurements.

Step 2. Mount GPS antenna on tripod.

Step 3. Measure vertical height from base of antenna (phase center model adjusts for actual vertical distance) to the ground. Input this value into the GPS configuration file.

 

Step 4. Run GPS for 1-3 hours per location collecting differential data every 5 seconds. Back at home run post-processing using base station data collected from Mauna Kea.