The equation developed by retaining only significant weekly rainfall variables is:
Y = 2.22316** + 0.00199** Rn_wk25 + 0.00163** Rn_wk26 – 0.00418** Rn_wk30 – 0.00574** Rn_wk32 – 0.00130* Rn_wk33 – 0.00248** Rn_wk34 + 0.00578** Rn_wk42
The coefficient of determination obtained for the regression model was 0.8142 which was highly significant.
Resulting stepwise model with weekly relative humidity revealed that, out of the considered 27 standard weeks, the relative humidity of 8 weeks namely, of the week 25, 27, 28, 30, 32, 39, 40 and 42 significantly influenced the T.Aman rice yield, accounting for 82% of the total variation in yiel. The equation developed by retaining only significant weekly relative humidity variables is:
Y = 1.09020** + 0.04090* Rh_wk25 - 0.03753* Rh_wk27 + 0.03499* Rh_wk28 – 0.07695** Rh_wk30 – 0.05267** Rh_wk32 + 0.04605** Rh_wk39 - 0.03738* Rh_wk40 + 0.09227** Rh_wk42
The coefficient of determination obtained for this regression model was 0.8174 which was highly significant.
Resulting stepwise model with weekly average temperature revealed that, out of the considered 27 standard weeks, the temperature of only 3 weeks namely, of the week 28, 30 and 34 significantly influenced the T.Aman rice yield, accounting for only 38% of the total variation in yield (Table 14). The equation developed by retaining only significant weekly temperature variables is:
Y = -6.49252ns - 0.20793* Tm_wk28 + 0.25840** Tm_wk30 + 0.23169* Tm_wk34
The coefficient of determination obtained for this regression model was 0.3754 which was also highly significant.
Finally, the above 7 significant weekly rainfall variables, 8 significant weekly relative humidity variables and 3 weekly average temperature variables were passed together for stepwise regression on eliminating the multicolinearty. The resulting stepwise regression model revealed that out of these 18 variables tested (Table 11), 11 variables significantly influenced the yield of rainfed T.Aman rice, accounting for 90% of the total variation in yield
The equation developed by retaining only significant weekly variables is:
Y = -4.73501ns + 0.00245 Rn_wk25 + 0.00137 Rn_wk26 – 0.00360 Rn_wk30 – 0.00279 Rn_wk34 + 0.00334 Rn_wk42 + 0.05141 Rh_wk28 – 0.08401 Rh_wk32 + 0.04641 Rh_wk39 – 0.04321 Rh_wk40 + 0.04306 Rh_wk42 + 0.18452 Tm_28
The coefficient of determination obtained for the final regression model was 0.8986 which was highly significant.