FlowerPower predicts the Zadoks flowering stages of cereal crops based on DPIRD trial data. The predicted flowering stages are Z65 for wheat, Z49 for barley, and Z71 for oats.
|Crop||Zadoks stage||Default cultivar|
Use the left panel to select crop, portzone, sites, and cultivars by typing in the names or using the drop-down menu. More cultivars can be added by clicking the 'Add Cultivar button.
Once the fields are filled in, select the tabs at the top of the page to see the predictions.
The predictions are in two formats:
Comparison: a table of predicted flowering dates given the time of sowing and the estimated flowering response curves for each cultivar. The prediction is based on a 'median' season;
10 Years Variations: predictions of the first selected cultivar based on weather data for the past 10 years.
Available cultivars differ by sites as the flowering data are not available for all cultivars at all sites. Available sites also differ by crops.
If the selections are not available, the fields will revert to the default cultivar of the respective crop.
Last updated: v7.0.10, 23 Mar 2022 by Kenyon Ng
The estimated dates in the table and line graph are the dates with the highest probability of flowering (based on the statistical model and under the 'median' weather condition). They should not be interpreted as the definite date on which the cereal will flower. There is a 50% chance that the crop will flower before the estimated dates.
The 'Upper' and 'Lower' columns show the uncertainty in the estimated flowering dates. There is a 80% chance that the cereal will flower within the upper and lower range, given the 'median' weather condition.
The table can be sorted by clicking on the column names.
The lines for each variety can be toggled on/off by clicking on the cultivar name in the legend.
A Scepter wheat crop sown on 9th May in Borden has a 10% chance of flowering by 31st Aug, 50% chance of flowering by 8th Sept, and 90% chance of flowering by 16th Sept.
Predictions based on weather data of the past 10 years
This graph shows the predicted flowering dates of the first cultivar on the selection list on the right. Each of the coloured curves shows the predicted flowering dates based on historial weather data (last 10 years from 2012 to 2021). The black curve shows predictions based on the 'median weather' of the past 10 years.
The underlying statistical model relies on, among others, heat accumulation rates during the growing season (1 Apr to 31 Oct) at a particular site to predict flowering dates. The 'heatsum index' is a value bounded by 0 and 1 that shows the relative differences in the heat accumulation rates for the past 10 years. In particular, the 'warmest' year in the past 10 years has a heatsum index of 1 and the 'coldest' year a heatsum index of 0.
The flowering response curves were modelled with semiparametric models and
fitted with a penalised splines estimator provided in the
mgcv R package (Wood
The model uses 'days to flowering' as the response variable with the sowing day of the year as the time-covariate, as opposed to the two-steps process proposed in Sharma and D'Antuono (2011) which makes predictions on the heat-sum required for flowering.
A key assumption in the model is that the environment is only characterised by 'heatsum slope', which measures the rate of heat accumulation between 1 April and 31 October. The 'sowing day' interacts with 'cultivar' and heatsum slope, and this implies that the shapes of flowering response curves differ among cultivars and environments. The latitude and longitude of the site are also among the model predictors, but they will only affect the vertical placement of the flowering response curves.
This model aims to give an indication of how different varieties compare for flowering dates based upon historical data. However, as with all statistical models, this app is not 100% accurate but is expected to give a reasonable approximation.
The Chief Executive Officer of the Department of Primary Industries and Regional Development and the State of Western Australia and their employees (collectively and individually referred to as DPIRD) accepts no liability whatsoever, by reason of negligence or otherwise, arising from any use or release of information in, or referred to in or linked to, this website, or any error, inaccuracy or omission in the information.
Sharma, D.L. and D'Antuono, M.F., 2011. Predicting flowering dates in wheat with a new statistical phenology model. Agronomy Journal, 103(1), pp.221-229.
Wood, S.N., 2011. Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 73(1), pp.3-36.
FlowerPower has been developed from a Microsoft Excel program into a robust web-based system by Mario D’Antuono (until 2018) with guidance from Dr Darshan Sharma. Since 2019, Kenyon Ng has been maintaining and upgrading FlowerPower with support from Dr Kawsar Salam and Dr Dean Diepeveen. Many thanks to DPIRD agronomy researchers who provided trial data (some co-funded by GRDC) and feedback on the user interface.
The weather data were obtained from the SILO database hosted by the Queensland Department of Environment and Science (https://www.longpaddock.qld.gov.au/silo/point-data/) under the Creative Commons Attribution 4.0 International (CC BY 4.0) licence.