Sensors 2015, 15, 5504-5517; doi:10.3390/s150305504 sensors
ISSN 1424-8220 www.mdpi.com/journal/sensors
An Approach to Precise Nitrogen Management Using Hand-Held
Crop Sensor Measurements and Winter Wheat Yield Mapping in a Mediterranean Environment
Lucía Quebrajo 1, Manuel Pérez-Ruiz 1,*, Antonio Rodriguez-Lizana 1 and Juan Agüera 2 1 Aerospace Engineering and Fluids Mechanics Department, University of Seville,
Ctra. Sevilla-Utrera km. 1, Seville 41013, Spain; E-Mails: firstname.lastname@example.org (L.Q.); email@example.com (A.R.-L.) 2 Rural Engineering Department, University of Cordoba, Córdoba 14071, Spain;
E-Mail: firstname.lastname@example.org * Author to whom correspondence should be addressed; E-Mail: email@example.com;
Academic Editor: Gonzalo Pajares Martinsanz
Received: 29 December 2014 / Accepted: 27 February 2015 / Published: 6 March 2015
Abstract: Regardless of the crop production system, nutrients inputs must be controlled at or below a certain economic threshold to achieve an acceptable level of profitability. The use of management zones and variable-rate fertilizer applications is gaining popularity in precision agriculture. Many researchers have evaluated the application of final yield maps and geo-referenced geophysical measurements (e.g., apparent soil electrical conductivity-ECa) as a method of establishing relatively homogeneous management zones within the same plot. Yield estimation models based on crop conditions at certain growth stages, soil nutrient statuses, agronomic factors, moisture statuses, and weed/pest pressures are a primary goal in precision agriculture. This study attempted to achieve the following objectives: (1) to investigate the potential for predicting winter wheat yields using vegetation measurements (the Normalized Difference Vegetation Index—NDVI) at the beginning of the season, thereby allowing for a yield response to nitrogen (N) fertilizer; and (2) evaluate the feasibility of using inexpensive optical sensor measurements in a
Mediterranean environment. A field experiment was conducted in two commercial wheat fields near Seville, in southwestern Spain. Yield data were collected at harvest using a yield monitoring system (RDS Ceres II-volumetric meter) installed on a combine. Wheat yield and NDVI values of 3498 ± 481 kg ha−1 and 0.67 ± 0.04 nm nm−1 (field 1) and
Sensors 2015, 15 5505 3221 ± 531 kg ha−1 and 0.68 ± 0.05 nm nm−1 (field 2) were obtained. In both fields, the yield and NDVI exhibited a strong Pearson correlation, with rxy = 0.64 and p < 10−4 in field 1 and rxy = 0.78 and p < 10−4 in field 2. The preliminary results indicate that hand-held crop sensor-based N management can be applied to wheat production in Spain and has the potential to increase agronomic N-use efficiency on a long-term basis.
Keywords: NDVI; yield estimation; winter wheat 1. Introduction
The goal of site-specific management practices is to enable more efficient use of fertilizers, pesticides, fuel, management and labor inputs. Most farming systems use spatial variability information related to crop status and soil characteristics to implement innovative management strategies to achieve a site-specific scenario. This new method of implementing agriculture is being bolstered by emerging cost-effective remote sensing techniques. Field crops must receive appropriate rates of nitrogen (N) fertilizer to achieve optimal yields; both underfertilization and overfertilization can negatively affect the desired growth pattern of plants and reduce yields. Furthermore, repeated machinery passes for N applications increase driving distances, require more time, increase soil compaction, consume more farming inputs and increase the environmental load .
Andalucia, in southern Spain, serves as an example of high agricultural value and represents 62% of the area (197,826.00 ha) used for and more than 80% of the national production of winter wheat, with an average yield of 3.11 t ha−1 (MAGRAMA , advancing production and area, July 2014). Using the average N fertilization application rate of 120 kg ha−1 per year and a price of 8–9 € ha−1 for application by a contractor company (two passes per fertilization) at 110 € t−1 of urea with 46% N amounts to a cost of 46.5 € ha−1 cost per year (fertilizer plus application). As much as 20% of inputs can be saved with the use of precision farming techniques and variable rate fertilizer application using proper machinery and precision application in areas with good yields and reduced inputs in areas with low yields (in which the lower yield may be due to soil limitations rather than insufficient N fertilization).
In this case, the cost would be 37.2 € ha−1 per year for the same yield at the end of the season.
Andalucia could save approximately 1.8 M € using precision agricultural techniques for N application.
Raun and Johnson  reported that conventional N management strategies in world cereal production systems have resulted in a lower percentage of applied fertilizer N being recovered in the aboveground crop biomass during the growing season; they estimate that an average of only 33% of fertilizer N is recovered. Although it is impossible to achieve 100% efficiency of N fertilizer use in any production system worldwide, the use of large amounts of N fertilizer suggests that there is a significant opportunity for reducing N losses associated with conventional practices.
Detailed knowledge of the relationship between applied fertilizer and crop yield in a zone under given soil conditions may be obtained through numerical approximations. Crop production models can be characterized as empirical and mechanistic (process-oriented) models. Empirical models directly employ a relationship between model variables and model outputs without requiring a description of fundamental (physical) processes. These models are usually site specific. Mechanistic models are often
Sensors 2015, 15 5506 more complex because they describe known physical and biological processes in crops and soils.