Table 3. Time series models analysis on forage production of sorghum-sudangrass hybrid with weather variables or some extreme weather events (heavy rainfall and typhoon) as independent variables

Models	Variables		Estimate	R2	RMSE	MAPE	Ljung-Box Q
1	Forage production (kg/ha)	Intercept	−61.11	0.7705	2,941.91	12.69	22.74 (p = 0.158)
		AR 1	−0.45*
		D	1
2	Forage production (kg/ha)	Intercept	1,568.90	0.6057	3,784.55	14.80	28.06 (p = 0.045)
		AR 1	−0.46*
		D	1
	GDD (°C)	AR 0	6.16*
		AR 1	6.65*
3	Forage production (kg/ha)	Intercept	1,390.25	0.6101	3,773.65	14.71	26.25 (p = 0.057)
		AR 1	−0.37
		D	1
	GDD (°C)	AR 0	6.88*
		AR 1	7.32*
	Accumulated rainfall (mm)	AR 0	−3.55*
		AR 1	−3.56*
4	Forage production (kg/ha)	Intercept	1,568.47	0.6058	3,789.21	14.79	27.92 (p = 0.046)
		AR 1	−0.37
		D	1
	GDD (°C)	AR 0	6.18
		AR 1	6.66
	Heavy rainfall 2 = 1 (over 1,000 mm)		−71.09
5	Forage production (kg/ha)	Intercept	1,416.63	0.6061	3,787.84	14.72	28.20 (p=0.043)
		AR 1	−0.37
		D	1
	GDD (°C)	AR 0	6.15
		AR 1	6.69
	Typhoon = 1 (frequency ≥ 1)		−347.74

p < 0.05.
RMSE, root mean square error; MAPE, mean absolute percentage error; AR, autoregressive- to forecast the dry matter yield using a linear combination of past value; D, differencing time lag- to make a non-stationary time series stationary for dry matter yield; GDD, growing degree days.