Erratum to “Modelling fertiliser significance in three major crops” [Eur. J. Agron. 90 (2017) 1–11] (S1161030117300941) (10.1016/j.eja.2017.06.012))

Ben Parkes, Benjamin Sultan, Philippe Ciais, Xuhui Wang

    Research output: Contribution to journalComment/debatepeer-review

    Abstract

    The Publishers would like to inform the esteemed readers that in the original version of the article, some values had to be updated in few of the article Sections and Table 2 and the updated sections of the article is as follows: Page number: 9 First paragraph of the Section 2.2: Model Validation; there is a value change which is highlighted: In the cases of models which do not arrive at a solution after a set number of iterations were discarded. This is the case for much of Russia, Ukraine, Central Asia, Uruguay and Mauritania for all crops in addition to Angola for maize and wheat. This exclusion of models without a solution accounts for less than 10% of maize and rice results, however for wheat 16.9% of the grid cells are removed in the Princeton dataset and 74.0% for WFDEI. The large number of wheat models that cannot find a solution are in the former USSR. Page number: 9 Fourth paragraph of the Section 2.2: Model Validation; there is a text change which is highlighted: The addition of variables to a statistical model also allows greater freedom in the parameter space at the cost of a less informative fit. The Akaike information criterion (AIC) measures the amount of information lost by a model, in a comparison between models the model with the minimum AIC is considered superior (Akaike, 1974). For models with a finite sample size the AIC is corrected and the AICc is calculated, this is shown in Figure 4. For almost all results there is a decrease in AICc, however for maize in Ecuador for the Princeton dataset there is an increase in AICc. In the WFDEI dataset models there is an increase in AICc for Turkey. Page number: 9 Fifth paragraph of the Section 3: Results; there is a text change which is highlighted: The wheat results, primarily the RMSE and SD, are significantly improved over much of Europe, (Poland, Czech Republic, Slovakia, Bulgaria, Turkey) and in addition the Egyptian Nile delta. The WFDEI results mirror those from the Princeton dataset however they show smaller improvements and the Turkish results should be treated with caution. Page number: 9 Sixth paragraph of the Section 3: Results; there is a text change which is highlighted: The improvements in the yield simulations in for grid cells for each model is shown in Tables 1 and 2. For the Princeton dataset the number of cells with an improvement in RMSE overwhelms the number where there is a reduction in quality. The extra yield explained by the models with fertiliser is close to 18%. The WFDEI dataset results are stronger with improvements of yield forecast by approximately 24%. Table 2 values have been completely updated with new values; please find attached the new table below: Table 2 Improvements in yield estimate for climate and fertiliser models when compared with climate only models for the WFDEI dataset. [Table presented] The Publishers would like to apologise for any inconvenience caused.

    Original languageEnglish
    Number of pages1
    JournalEuropean Journal of Agronomy
    Volume93
    DOIs
    Publication statusPublished - 1 Feb 2018

    Fingerprint

    Dive into the research topics of 'Erratum to “Modelling fertiliser significance in three major crops” [Eur. J. Agron. 90 (2017) 1–11] (S1161030117300941) (10.1016/j.eja.2017.06.012))'. Together they form a unique fingerprint.

    Cite this