Neural network and crop residue index multiband models for estimating crop residue cover from Landsat TM and ETM+ images

Crop yield estimation using satellite images: comparison of linear and non-linear models

“…As Bocco et al (2014) state, its productive characteristics are very representative of the central region of Argentina (Figure 1). It presents a slightly undulating relief of hills developed on loessic material of silt loam texture with a slight slope to the east; soils in this area are classified as Entic and Typic Haplustoll characterized by Rollán and Bachmeier (2014).…”

“…On the other hand, corn is sown by direct seeding also, with a distance between rows of 0.53 m and fertilized with urea (Bocco et al, 2014;Ferreyra, 2016) (Figure 1). Both crops were sown between October 15 and November 15 and harvested between April 20 and May 30.…”

“…The general steps that describe the training algorithm were executed according to Bocco et al (2014). This process was repeated 3,000 times or until the mean square error was lower than a desired value (in these models equal to 0.01).…”

“…Therefore, the use of remote sensing techniques is an effective tool for assessing and monitoring crop yield because satellite data provide a spatial and periodic, comprehensive view of the real crop state (Geipel et al, 2014). The information provided by remote sensing allows obtaining field data at a lower cost compared to other methods, the coverage of large areas and the possibility of periodic repetitions (Bocco, Sayago and Willington, 2014). Several types of images can be used to monitor the agricultural surface.…”

“…Neural network (NN) methodology is an alternative modeling and simulation tool, which is specially designed for nonlinear systems and they do not require prior assumption about the statistical behavior or any specific relationship between variables (Bocco et al, 2014). Kaul, Hill and Walthall (2005) investigated the NN models performances in predicting corn (Zea mays L.) and soybean (Glycine max (L.) Merr.)…”