Delineation of management zones with spatial data fusion and belief theory

Abstract

Precision agriculture, as part of modern agriculture, thrives on an enormously growing amount of information and data for processing and application. The spatial data used for yield forecasting or the delimitation of management zones are very diverse, often of different quality and in different units to each other. For various reasons, approaches to combining geodata are complex, but necessary if all relevant information is to be taken into account. Data fusion with belief structures offers the possibility to link geodata with expert knowledge, to include experiences and beliefs in the process and to maintain the comprehensibility of the framework in contrast to other "black box" models. This study shows the possibility of dividing agricultural land into management zones by combining soil information, relief structures and multi-temporal satellite data using the transferable belief model. It is able to bring in the knowledge and experience of farmers with their fields and can thus offer practical assistance in management measures without taking decisions out of hand. At the same time, the method provides a solution to combine all the valuable spatial data that correlate with crop vitality and yield. For the development of the method, eleven data sets in each possible combination and different model parameters were fused. The most relevant results for the practice and the comprehensibility of the model are presented in this study. The aim of the method is a zoned field map with three classes: "low yield", "medium yield" and "high yield". It is shown that not all data are equally relevant for the modelling of yield classes and that the phenology of the plant is of particular importance for the selection of satellite images. The results were validated with yield data and show promising potential for use in precision agriculture.

Methodology for automated classification of farmland based on Earth remote sensing data

“…For green vegetation, the reflection in the red region is always less than in the near infrared, due to the absorption of light by chlorophyll, so the NDVI values for vegetation cannot be less than 0. To obtain the NDVI map, an index was calculated for each pixel of the Landsat 5 space image with a spatial resolution of 30 meters in the red and near infrared spectral zones [3,4].…”

Suitability of satellite remote sensing data for yield estimation in northeast Germany

“…For the correlation analysis, 15 indices that are frequently used both in research and in precision farming practice, were selected (Lilienthal, 2014;Siegmann et al, 2012). In accordance with the studies of Georgi et al (2017) and Vallentin et al (2019), which are based on the same satellite and yield data sets, the choice of indices was narrowed down to yield-relevant spectral indices (Table 3). The indices were calculated from appropriate spectral bands of the different sensors, which are varying in number and wavelength range.…”

“…The correlation result is given as Spearman Correlation Coefficient r (Daniel, 1990), also called "Spearman's Rho", and always refers to the correlation between the yield data and the respective data source to be analyzed.. Earlier studies (Georgi et al, 2017;Vallentin et al, 2019) showed a monotonous but non-linear correlation between satellite data and yield data, which would rule out correlation methods based on linear correlation assumptions. One reason for the non-linearity is the saturation of vegetation indices (Esau et al, 2016;Haboudane, 2004), which in high ranges do not correlate as strongly with the yield as in the lower and middle ranges.…”

Data Fusion For Intelligent Decision Making in Agriculture Resource Management

“…Multi sensor Data Fusion Environment: Sensors with Different agriculture variants have been used to provide information that are capable in monitoring and optimizing growth level of plants, land water stage, humidity stage, soil category and temperature point of the agriculture ground as per the varying environmental factors [19] .These sensors inspect by penetrating the soil and records resistive forces before sending them to the central node for further data integration and data fusion [12].…”