Fjölrit RALA - 15.12.2000, Side 25

23
Yijk=si + tj + uk + aHjjk + bUijk + eijk
Yíj Growth rate (kg ha'1 day'1) year i, location j and week k.
S| Intercept for year i.
tj Intercept for location j.
uk Intercept for week k.
a, b Regression coefficients for temperature and precipitation respectively.
Hjjk Average temperature year i, location j and week k.
Ujjk Average daily precipitation year i, location j and week k, threshold value = 1.1-1.5.
eiJk Residual error.
The coefficients for regression on temperature and precipitation are shown in Table
20. This model explained 26% of the total variation for timothy, 14% for Fylking, 6% for
Lavang but very little for Seida less than corresponding to the degrees of freedom.
Table 20. Coefficients for temperature and precipitation effects on growth.
Temperature Coefficient SE Precipitation Coefficient SE
Lavang 13.8 7.8 96.0 56.8
Fylking 12.2 6.7 77.0 49.2
Seida 3.2 11.3 78.2 82.4
Timothy 8.7 10.5 8.5 46.4
An increase of 1”C in temperature will increase the daily growth rate by 3 -14 kg/ha.
In Icelandic investigations the parameter for temperature has been estimated as 8 kg/ha for
each degree (Thorvaldsson and Bjömsson, 1990; Thorvaldsson, 1998), which falls within the
error limits in this investigation.
Precipitation is not necessarily well suited to express the effect of water supply on
growth rate. Available soil water is continuously needed and is not only dependent on the
precipitation during the current week but also to earlier precipitation, its distribution
throughout the growth period, soil water supply, evaporation, soil type, etc. The coefficients
just confirm that water is an important growth factor up to a certain level. Better fit of the
model was obtained if linear effects of rainfall were assumed up to a threshold level and no
immediate effect of rain excluding this level. Precipitation at 1.1 mm/day gave the best fit as
the threshold value in the model (1.5 mm for timothy).
Height
The height of the plants (crn) was observed at each harvest date. These observations do not
represent the highest plants but the average height. Tables 21-23 show the results on different
harvest dates and different locations. For some reason the plants were generally a little lower
in Kollafjorður than at the other locations.