DISCUSSION
1
Observation of the general temperature trends through time between PRIMET150-5 (150cm sensor, probe 5, enclosed thermister) and CS2MET show a decoupling of daily TMAXs from the 1970s though the 1990s.
Starting with the daily temperature trends between the two sites in the 1970s, it can be seen that TMAXs correspond more closely during the late spring and summer months, a similar coupling in the winter, with significant decoupling in the spring and autumn. We believe the reasons behind these seasonal trends to be: 1. Variation in direct solar radiation throughout the year due to sun’s position in the sky, 2. Variations in cloud cover (i.e., cloudy conditions in winter, clear in summer.) In fact, comparison between these charts and daily cloud cover reports from Eugene show that on cloudy days, the difference between PRIMET and CS2MET TMAXs are minimal, while sunny days produce the greatest differences. The fact that these differences are relatively minimal in summer, when the sun is the highest, suggest that forest canopy cover plays a significant role.
It is unclear what is causing the general decoupling through time, however. Lack of correspondence between the temperatures during the summer months in the 90’s when compared to those in the 70s shows this, as well as the chart showing the differences in daily TMAX for the entire period (slowly upward trend). Changes in forest canopy as CS2MET (i.e., trees growing taller) is conjectured to be the reason. The fact that TMAXs are affected and TMINs are not seems to verify these hypotheses.
2
The second issue involves discrepancies between the two 150cm sensors at PRIMET. Probe 5 is the long-term sensor, a thermister enclosed in a Stevenson shield providing temperature data since 1972. Probe 4 is and open thermister, operational since 9/94. Comparisons between them reveal that probe 5 developed an apparent fault during the winter of 96/97. Observation of the TMAX and TMIN daily differences between the probes show that the probe 5 TMIN reading rose by almost a degree during a period of 4 or 5 months. This discrepancy appears to have persisted through the data period which ends in 1998. Charts showing the relationships between other sensor heights and the 2 probes at 150cm verify that probe 5 is the one causing the problems, not probe 4. The fact that the increase is gradual over a few months and not a sudden step will make it difficult to stastically deal with this problem. Probe 5’s temperature data was used to generate all historical plots unless otherwise noted.
Other Issues:
Observation of daily TMIN differences between PRIMET150-5 and CS2MET130 during winter 93/94 show a curious jog in the data.