Corn Grain Yield Response to Nitrogen Rates

Name: Adrian Correndo

Semester: Spring 2019

Project area: Agronomy

Last update: Apr-15-2019

Objective

Automating the calculation of grain yield (GY) response to different rates of nitrogen (N) fertilizer, and related fertilizer use efficiencies (NAE) in a database of more than one thousand experiments with different designs. A secondary goal is to explore descriptive statistics of variables of interest grouping experiments by soil texture classes (STx).

Input

*.csv file with 4 columns: TRIAL, STx, N rate, and GY, where:

-TRIAL: Experiment ID number;

-STx: Soil texture class of typical pedon (Soil Survey Staff, 2018);

-Nrate: Nitrogen rate (kg N ha^-1);

-GY: Grain Yield when Nrate=0 (Mg ha^-1, 15.5% moisture);

Outputs

Two "*.csv" files

1. N0plots.csv, with the following columns: TRIAL, STx, Y0, Ymax, NRmax, NRmaxr, NAEmax.
2. Nf_plots.csv, with the following columns: TRIAL, STx, Nrate, GY, Y0, Ymax, NR, NRr and NAE, where:

-Y0: GY when Nrate=0 (Mg ha^-1);

-Ymax: maximum observed GY (Mg ha^-1);

-NR: absolute nitrogen response corresponding to each fertilizer rate different from 0 (Mg ha^-1).

-NRr: relative nitrogen response corresponding to each fertilizer rate different from 0 (%).

-NRmax: maximum absolute nitrogen response (Mg ha^-1).

-NRmax_r: maximum relative nitrogen response (%).

-NAE: nitrogen agronomic efficiency as NR divided by its corresponding Nrate (kg NR kg^-1 of applied N).

Main challenges were related to:

i) the # of Nrate levels and the ammount of applied N (kg) vary across trials;

ii) Y0 and Ymax values take place at Trial level, while the NR and NAE values, at a sub-level by a given Trial-Nrate combination.

Rationale

I'm working on a review paper for which I collected more than 1200 experiments. Automating these calculations will save me a significant amount of time and avoid potential misscalculations when processing the data. This code might be potentially useful for other similar databases studying crop responses to fertilizer.

Data Example

Fig. 1. Scatter plot of Corn Grain Yield (kg ha^-1) and nitrogen rate (kg N ha^-1) from two typical corn nitogren experiments from the database (#1: circles, #2: triangles). Y0, Ymax and NR variables are shown for Trial 1. Fitted functions correspond to quadratic-plateau models (y = a + bx +cx^2, when Nrate < Xc -critical value or threshold-; otherwise y = plateau).

Sketch

Fig. 2. Main steps of the project from data input to expecta data outpouts.

References

1. Soil Survey Staff, Natural Resources Conservation Service, United States Department of Agriculture. Web Soil Survey. Available online at the following link: https://websoilsurvey.sc.egov.usda.gov/. Accessed [05-06-2019].
2. os - Miscellaneous operating system interfaces. https://docs.python.org/3/library/os.html. Accessed [05-06-2019].
3. glob — Unix style pathname pattern expansion. https://docs.python.org/3/library/glob.html. Accessed [05-06-2019].
4. The pandas project. https://pandas.pydata.org/. Accessed [05-06-2019].
5. Folium 0.1.5. https://pypi.org/project/folium/0.1.5/. Accessed [05-06-2019].