Estimating soybean ground cover from satellite images using neural-networks models

Abstract

The ground cover is a necessary parameter for agronomic and environmental applications. In Argentina, soybean (Glycine max (L.) Merill) is the most important crop; therefore it is necessary to determine its amount and configuration. In this work, neural-network (NN) models were developed to calculate soybean percentage ground cover (fractional vegetation cover, fCover) and to compare the accuracy of the estimate from Moderate-Resolution Imaging Spectroradiometer (MODIS) and Landsat satellites data. The NN design included spectral values of the red and near-infrared (NIR) bands as input variables and one neuron output, which expressed the estimated coverage. Data of fCover were acquired throughout the growing season in the central plains of Córdoba (Argentina); they were used for training and validating the networks. The results show that the NNs are an appropriate methodology for estimating the temporal evolution of soybean coverage fraction from MODIS and Landsat images, with coefficients of determination (R 2 ) equal to 0.90 and 0.91, respectively.

Machine learning methods for sub-pixel land-cover classification in the spatially heterogeneous region of Flanders (Belgium): a multi-criteria comparison

“…State-of-the-art machine learning algorithms like Support Vector Machines (SVMs), Random Forests (RFs), and Boosted Regression Trees are getting widely accepted in many remote-sensing-related studies and have shown robust and reliable regression and classification results. From the recent literature, we can conclude that spectral unmixing and machine learning are becoming equally popular for performing land-cover classification at the sub-pixel level (Bocco et al 2012;Cortés, Girotto, and Margulis 2014;Fan and Deng 2014;Farook, Sivaraman, and Kesavaraj 2013;Reschke and Hüttich 2014;Schwieder et al 2014;Wang, Shao, and Kennedy 2014;Zhang, Zhang, and Lin 2014;Benhadj et al 2012).…”

Determination of agricultural land use: incidence of atmospheric corrections and the implementation in multi-sensor and multi-temporal images

“…In general classification methods can be grouped into parametric and non-parametric; Maximum likelihood (ML) is in the first group and has been widely used in the study area with very good statistic results, for classifying individual scenes and for time series (TS) (Nolasco et al, 2014). Neural networks (NN), a very powerful and easy to implement tool belongs to the second group, can compute, process and classify information, and allow adjusting a response to non linear, complex and multi-data problems (Bocco et al, 2012).…”