Note: all readings are in lux units, taken with an Optikon Hagner model EC1 luxmeter.
Note: We had intended to take readings on or about April20, June20, August20, in each of 1999, 2001 and 2003; but the first April readings were taken in 2000 instead of 1999; and we took a complete set of readings in 2002, but not 2001 or 2003. We attempted to take readings on a sunny low-wind day. Each day consists of readings (a) at 3 hours before high-noon, (h) at high-noon, and (p) at 3 hours after high-noon. (High-noon occurs at approximately 1:20pm central-daylight-time at this location.)
Weather notes on the readings days:
1999 June 19 (9906) -- Bud reading, Doris recording. Clear and sunny.
1999 Aug. 15 (9908) -- Bud reading, Doris recording. Intermittent cloud.
2000 April 24 (9904) -- Doris reading, Bud recording. Sunny, some cloud around 1600hrs. A large tree has fallen from the west across plot#17.
2002 April 22 (0204) -- Bud reading, Doris recording. Sunny, very windy. Luxmeter not working so we did the 1600hr readings on Apr26/02.
2002 June 20 (0206) -- Bud reading, Eugene recording. No notes on weather.
2002 Aug. 19 (0208) -- Bud reading, Doris recording. Clear and sunny. Bog wet, standing water in some plots, most overgrown with grass.
Reliability of this data:
Many of these readings are inconsistent with being outdoors on a sunny day. So either there was cloud, or the meter was malfunctioning, or it was misread. The easiest mistakes to make when reading the type of meter being used, are (1) to be out by a factor of 100, and (2) to be out by a factor of 10. On one occasion we had a borrowed meter which behaved more like a random-number generator, than a light-measuring device -- several readings at the same location would differ by more than a factor of 10.
One thing that most human observers agree on, is that Plot#20 is the least shaded of all these plots; and yet one observes at best a weak tendency for Plot#20 to have the brightest readings; it is among the top-3 only 6 times out of 18 (one-third of the time it falls into the top one-eighth). Plot#18 has the most first-places; plots 11, 13, 15, 20 also have multiple firsts; and plots 6, 10, 12, 16 have a first; in all 18 cases the plot with the brightest reading is among those that human observers had placed into the top half.
One expects June readings to be brighter than either April or August readings; one also expects high-noon readings to be brighter than either the morning or afternoon readings. On 4 of the 6 days, the brightest reading was among the high-noon ones. In both years, the brightest of the morning readings came in June; in 1 of 2 years, the brightest noon reading came in June; in neither year did the brightest afternoon reading come in June; the all-time brightest reading came in June, but in the morning.
The low readings are the dubious ones, for 2 reasons: (1) the most likely reading-errors involve the omission of the "times-10" and "times-100" portions of the lux-meter, and (2) on a partly-cloudy day, the light-conditions are highly variable over the time it takes to visit all 24 plots. Useful readings could be taken on clear days and on thoroughly overcast days, though we have no readings from an overcast day. For readings on a partly-cloudy day, the data is likely made more useful by discarding suspiciously low readings.
The original design called for light-level readings in years 1, 3 and 5 of the 5-year study-period, in order to show how canopy-closure changed over that period; however the frequency of dubious readings has caused us to abandon such use of the data. However once we think of canopy-closure as being constant over the 5-year period, then we think our data becomes a reasonably good estimator of that canopy-closure.
In order to use this data as an estimator of canopy-closure, a reasonable approach seems to be one of:
(A) for each plot, use the brightest reading ever recorded there, or
(B) for each plot, take the average of brightest morning, brightest noon, brightest afternoon readings, or
(C) for each plot, take the average of brightest April, brightest June, brightest August readings, or
(D) for each plot, take average of 9 values, the brightest from each time-of-day from each month.
Method-A uses one-eighteenth of the readings, both B and C use one-sixth, and D uses one-half. Method-D would seem best if most of the readings were reliable; however the high number of dubious readings, casts doubt on method-D.
The following table summarizes the results for both methods A and B.
The rankings are very similar for the 2 methods; we will use the method-B rankings as an estimator of canopy-closure.