It was found that the panicle initiation of the rainfed rice crop occurred between 46 DAP (days after planting) (24 August) and 48 DAP (26 August). Heading (Anthesis) occurred between 30 September (83 DAP) in some cases and 6 October (88 DAP) in some other cases. The crop reached physiological maturity happened between 9 November (122 DAP) and 16 November (129 DAP).
4.1. ERF performance on yield and HI:
The relation between stage-wise yield, HI of the crop was analysed by estimating the correlation between them. These growth stages of the rice crop were categorised under five (5) phases, viz.
- Emergence -End of Juvenile phase (stage 1),
- End of Juvenile -Panicle Initiation phase (stage 2),
- Panicle Initiation -End of Leaf Growth phase (stage 3),
- End of Leaf Growth –Beginning of Grain Filling phase (stage 4) and
- Grain Filling Phase (stage 5).
While simulating, it was observed that duration of stage 1 range between 31 and 35 days, stage 2 between 30 and 44 days, stage 3 between 33 and 40 days, stage 4 between 6 and 9 days and stage 5 between 27 and 36 days.
4.1.1: Vegetative phase:
Throughout the vegetative phase, prior to panicle initiation, 7 initial dates of forecast were selected and investigated (table.2a).
In the season 2019: 10th, 17th, 24th and 31st July, 7th, 14th and 21st August initial dates of ERF splicing were considered. It was noticed that yield obtained by splicing up of observation data with ERF forecast at these initial dates and climatological normals (category 2) were significantly closer to yield at and after 31st July splicing. However, correlations were still not as better as category 3 datasets (without ERF). Regarding HI, 4 out of 7 dates of ERF splicing in category 2, were better correlated significantly than category 3 datasets. These dates were: 10th July (r=0.777**), 24th July (r=0.881**), 31st July (r=0.832**) and 7th August (r=0.672*).
In season 2020: 8th, 15th, 22nd, 29th July, 5th, 12th and 19th August initial dates of ERF forecast spliced datasets (category 2) were analysed and was found that none of their yield and HI had any significant correlation with yield and HI obtained with category 1 dataset (only observed weather).
In season 2021: 7th, 14th, 21st, 28th July, 4th, 11th and 18th August spliced datasets of category 2 were found to be having significantly positive correlations with observed yield. However, except 14th July (r=0.805**) and 28th July (r=0.872**) splicing, all the other data were less correlated when compared to those data of category 3 where splicing of only climatological normals are made without ERF. Harvest Index (HI) of any dates weren’t correlated with HI obtained with category 1 data.
Summarising, the correlations carried between Yield, HI obtained using category 1 dataset (only observed weather) and category 2 (observed + ERF +Normal) and category 3 (observed + Normal) shows that replacing climatological normal with ERF at vegetative phase of the rainfed rice crop doesn’t show much improvement in its performance, except for season 2019 HI.
4.1.2: Reproductive phase:
Reproductive phase covers stage 3 and 4 (Panicle Initiation -End of Leaf Growth phase and End of Leaf Growth –Beginning of Grain Filling phase, respectively). It begins from panicle initiation and lasts up to heading (or Anthesis).
From table.2b., it can be inferred that in season 2019, 28th August, 4th, 11th, 18th September initial dates of ERF splicing were considered, which covered the reproductive phase of the crop. It was observed that yield obtained with both category 2 hybrid and category 3 datasets significantly correlated positively with yield obtained with category 1 dataset. Despite of this, the category 2 coefficients were found to be still lower than category 3, indicating that splicing with ERF did not provide improvement. Regarding HI, none of category 2 dataset derived HI had significant correlation, whereas splicing only normals excluding ERF (category 3), at all the dates, were found significantly related to category 1 dataset.
In season 2020, 26th August, 2nd, 9th, 16th September initial dates of ERF splicing were considered. It was found that the yields and HI obtained by using category 2 and category 3 datasets did not correlate significantly, except 16th September splicing wit ERF (category 2) which showed a significant positive correlation (r=0.709*) with yield obtained from category 1.
In season 2021, 25th August, 1st, 8th, 15th September initial dates of ERF splicing were considered, covering the reproductive phase of the crop. It was remarkable to notice that by splicing ERF along with normal (category 2 hybrid) datasets, instead of only climatological normals (category 3), fruitful impact was obtained. It was inferred from correlation of yield obtained by splicing with ERF at all dates had positive significant relationship with category 1, and it was found to be better than yield obtained with category 3. Even in HI, the hybrid data (category 2) showed significant positive correlation with category 1 dataset HI, at two dates: 8th and 15th September, 2021.
Summarising, in reproductive phase, only in 2021, yield and HI obtained with category 2 dataset with ERF splicing, were found to have better correlation coefficients with those obtained with category 1 dataset (observed weather). This improvement in this phase of the crop, opens up a new arena into usability of ERF forecast for gap-filling the weather data in weather file of DSSAT, along with climatological normals, for crop yield forecasting.
4.1.3: Ripening phase:
Reproductive phase covers stage 5 (Grain filling) or ripening phase also. Ripening phase is from flowering to grain maturity. Ripening phase includes Milky stage, Dough stage, and yellowing & ripening stage.
In the season 2019, 25th September, 2nd, 9th and 16th October initial dates of ERF splicing were considered, which covered the ripening phase of the crop (table. 2c). It was observed that yield obtained with both category 2 hybrid and category 3 datasets significantly correlated positively with yield obtained with category 1 dataset. Also, the correlation coefficient of yield obtained with category 2 hybrid and category 3 datasets was at par (r=0.999**) when spliced at 9th October, whereas splicing with ERF (category 2 dataset) at 16th October revealed a far better correlation (r=1.00**) as compared to without ERF (observed data spliced with only normal data- category 3) (r=0.703*). Coefficients of HI proved that all the data of category 2 and 3 were positively correlated significantly with HI obtained with category 1. However, only HI using 16th October ERF spliced up data had better correlation (r=0.974**) to category 1 HI, compared to category 3 (r=0.659*).
In the season 2020, 23th and 30th September, 7th and 14th October initial dates of ERF splicing were considered. It was observed that 23rd September and 14th October (r=0.906** and r=0.999**, respectively) splicing with ERF (category 2) had better correlated with yield, than category 3 (without splicing) (r=0.885** and r=0.935**, respectively, for those two dates). The coefficients of yield achieved with other dates of splicing ERF were at par with those obtained without ERF. Only ERF spliced hybrid data at 16th October had HI which correlated better (r=0.992**) with category 1 HI than without ERF (r=0.965**).
In the season 2021, 22nd and 29th September, 6th and 13th October initial dates of ERF splicing were considered. It was noticed that yield obtained with October splicing with ERF were better correlated to observation dataset than the yield obtained using dataset without ERF. In HI, 22nd September and 13th October splicing with ERF were better correlated.
Summarising, yield obtained by splicing ERF data during October were better correlated to the yield obtained using observation weather. However, only second week splicing with ERF proved to have been better correlated in HI.
4.2. ERF performance on rainfall:
The rainfall accumulated in each of the stage of the crop was analysed and compared. The spliced data with ERF and without ERF were investigated on how far they are closer to the observed weather data.
4.2.1: Vegetative phase:
Rainfall accumulated in vegetative phase (Emergence -End of Juvenile phase (stage 1) and End of Juvenile -Panicle Initiation phase (stage 2)) (table.3a) was analysed for splicing of ERF data in this phase. It was found that rainfall accumulated in stage 1 by splicing ERF data, were at par with rainfall without ERF. Both category 2 and 3 data rainfall at stage 1 were positively and significantly correlated with observed rainfall and were at par, when dates of 2nd fortnight as initial condition were selected in all the three seasons.
However, in only 2021, stage 2 rainfall accumulation with ERF splicing was found to be better correlated with observed rainfall than category 3.
4.2.2: Reproductive phase:
Rainfall accumulated in reproductive phase (Panicle Initiation -End of Leaf Growth phase (stage 3) and End of Leaf Growth –Beginning of Grain Filling phase (stage 4)) (table.3b) was analysed for splicing of ERF data in this phase. It was found that rainfall accumulated in stage 3 by splicing ERF data, was significantly correlated to observed rainfall than category 3 weather dataset rainfall without ERF, in season 2019. In season 2020, only two dates have better coefficient between category 2 weather and observed rainfall at stage 3. However, it further diminished in season 2021. In stage 4 of season 2019, 11th and 18th September (r=0.778** and r=0.894**, respectively) was observed as correlation between ERF spliced data and observed rainfall, which was better than correlation between non-ERF spliced data with observed rainfall (r=0.520 and r=0.351, respectively). In seasons 2020 and 2021, however, only one date out of four of stage 4 was having significant correlation for category 2 rainfall (r=0.702* and r=0.643*, respectively for 2020 and 2021) to observed rainfall (category 1).
4.2.3: Ripening phase:
Rainfall accumulated in ripening phase (Grain Filling phase (stage 5)) (table.3c) was analysed for splicing of ERF data. It was found that rainfall accumulated in stage 5 by splicing ERF data, was significantly correlated to observed rainfall only in season 2020 at date 14th October as initial (r= 0.922**), and was better than splicing without ERF (category 3 dataset) whose correlation coefficient was 0.663* to observed rainfall.
While summarising the phasic rainfall accumulation and comparison of ERF spliced dataset with dataset having no ERF, it was clearly found that, most of the stage 1 (Emergence -End of Juvenile phase) rainfall accumulation of all three seasons 2019, 2020 and 2021 were better correlated with ERF splicing confirming its nearness to the observed phasic rainfall accumulation. With regard to reproductive stage, only in stage 3 (Panicle initiation to end of leaf growth stage) ERF spliced data was better correlated with observed rainfall. Grain filling phase did not have enough dates proving efficiency of ERF spliced dataset, as most of the cases of both with and without ERF were non-significantly correlated with observed rainfall.