Abstract
To explore precision farming profits, the variability within a plot can be evaluated using digital technology by different remote means. The objectives of this study were to determine crop coverage (CC) of soybean (Glycine max (L.) Merril) with Normalized Difference Vegetation Index (NDVI) data obtained by digital photographs on the field and from the satellites LANDSAT (7 and 8), with an overpass each 16 days and a pixel of 30 m, and PROBA-V, which has daily frequency and 100 m of spatial resolution, in order to evaluate productivity differences between sectors of a 45 ha rainfed plot located at south of Córdoba city, Argentina. In the plot, sowed on 22/11/2014 and harvested on 10/04/2015, 16 sampling areas were established to record periodically photographs with a modified camera and, in 8 of them, supplementary crop information. A non-linear model was developed from NDVI data of digital camera (NDVIC) to estimate the soybean CC that showed an appropriate predictive performance. Furthermore, NDVI data of LANDSAT (7 and 8) (NDVIL) and PROBA-V (NDVIP-V) were also applied to estimate CC, resulting in models whose structure and accuracy was similar to that obtained with the digital camera (R2 = 0.956 and 0.939, respectively). According to the radiometric information the two instruments provide, the digital images classification procedure to determine CC requires increasing the threshold from 0.0 to 0.05 when soybean progresses towards the maturation and senescence stages and green material is mixed with the senescent one. Growing conditions were very favorable for soybean in 2014–2015, since precipitation (PP) not only showed a marked continuity with 60 rainy days during the cycle, but also 642 mm accumulated in this period far exceeded maximum evapotranspiration (ETmax) of 389 mm. The CC had a major development in all sectors, maintaining a complete coverage condition for more than 50 days during most of the reproductive stage. However, prevalent overcast sky restricted significantly solar radiation (SR) and reduced potential yield (PY) to an average value close to 6000 kg ha−1 which, according to the plot yield map, produced a reduced yield gap (YG) between 10.6 and 19.8%. From the proposed model and with the NDVI data of LANDSAT 7 (NDVI7), soybean CC was estimated in the same plot for 2010–2011. Water availability were less favorable in this case, with accumulated values of 584 mm and 460 mm, for PP and ETmax, respectively, while a higher availability of SR during the crop season increased notably PY that reached a range between 7347 and 8224 kg ha−1. Moreover, lower water availability was evidenced increasing YG in the plot (40–53%). From the spatial evaluation carried out, only one-third of the plot located at the south reached the highest productivity in both crop seasons, leaving open the question about the weather influence in each productive cycle with respect to the effectiveness of the site-specific management.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have