Modern technology provides tools and devices that allow to make a qualitative leap also in practices considered more rooted in tradition such as agriculture. In particular, precision farming exploits the use of GPS-based applications and machineries able to use geo-spatial information for commanding site-specific treatments to increase agricultural production. In this context, the detection and analysis of within-field variability and anomalies is a crucial point for the application of precision farming practices in general, and for the application of variable rate technologies for fertilization in particular. Delineation of zones to be differently managed is usually performed by analyzing past season(s) yield maps in order to identify areas of low/high production. This however requires the availability of high-end harvesting machinery. An alternative to this approach is to derive information related to the current season from high resolution remote sensing data. The typical approach is based on the analysis of spectral indices maps to highlight the intra-field variability at specific crop stages (e.g., crop emergence, tillering, heading), with the objective of providing a baseline for adopting differentiated management for fertilization and other treatments. In this context, a two year test study took place in the framework of ERMES project (http://www.ermes-fp7space.eu/) for developing (2014) and then operationally testing (2015) the exploitation of information derived from Very High Resolution (VHR) satellite images (Worldview-2 - Rapid-eye) during the crop season as a support for variable rate technology (VRT) nitrogen fertilization in paddy rice fields. During the first year, two Worldview-2 acquisitions (July - August) were scheduled for testing the performance of several vegetation indices (VIs) and the influence of the period of acquisition in explaining the variability of rice production from VHR data. This was done by analyzing the correlation between the VIs maps with the yield maps produced at the end of the season by a smart harvester. This allowed to identify the more suitable VIs to be used in the different phenological stages in order to explain the spatial yield variability. A variable fertilization was also done during the last application of nitrogen of the season in a single field for testing the sensibility of the before mentioned VIs to these kind of practices. In the second year, 2014 experimentation outcomes were used for i) scheduling images acquisition, ii) automatic processing of images and iii) defining the information flow to allow farmers to exploit maps produced. This last aspect was fundamental to provide information in time for supporting the nitrogen fertilization and test the effect of site specific applications. Two Worldview-2 images were acquired in June and July before rice fertilization periods and two Rapid-eye images were acquired after the fertilizations. The first two images (Worldview) provided the base maps to target variable fertilizations while the Rapid-eye were used to verify the effect of the applications. A final evaluation was also done by comparing the maps delivered during the season with the rice yield maps produced with the harvester at the end of the season. This operational test demonstrated how the use of information related to within field variability generated in near-real time from satellite imagery, is a suitable alternative to the traditional use of previous year yield maps for the management of variable rate dosage of nitrogen.
Remote sensing for supporting precision farming: an operational test case in Italy for nitrogen fertilization in rice crops
Crema A;Nutini F;Busetto L;Boschetti M
2016
Abstract
Modern technology provides tools and devices that allow to make a qualitative leap also in practices considered more rooted in tradition such as agriculture. In particular, precision farming exploits the use of GPS-based applications and machineries able to use geo-spatial information for commanding site-specific treatments to increase agricultural production. In this context, the detection and analysis of within-field variability and anomalies is a crucial point for the application of precision farming practices in general, and for the application of variable rate technologies for fertilization in particular. Delineation of zones to be differently managed is usually performed by analyzing past season(s) yield maps in order to identify areas of low/high production. This however requires the availability of high-end harvesting machinery. An alternative to this approach is to derive information related to the current season from high resolution remote sensing data. The typical approach is based on the analysis of spectral indices maps to highlight the intra-field variability at specific crop stages (e.g., crop emergence, tillering, heading), with the objective of providing a baseline for adopting differentiated management for fertilization and other treatments. In this context, a two year test study took place in the framework of ERMES project (http://www.ermes-fp7space.eu/) for developing (2014) and then operationally testing (2015) the exploitation of information derived from Very High Resolution (VHR) satellite images (Worldview-2 - Rapid-eye) during the crop season as a support for variable rate technology (VRT) nitrogen fertilization in paddy rice fields. During the first year, two Worldview-2 acquisitions (July - August) were scheduled for testing the performance of several vegetation indices (VIs) and the influence of the period of acquisition in explaining the variability of rice production from VHR data. This was done by analyzing the correlation between the VIs maps with the yield maps produced at the end of the season by a smart harvester. This allowed to identify the more suitable VIs to be used in the different phenological stages in order to explain the spatial yield variability. A variable fertilization was also done during the last application of nitrogen of the season in a single field for testing the sensibility of the before mentioned VIs to these kind of practices. In the second year, 2014 experimentation outcomes were used for i) scheduling images acquisition, ii) automatic processing of images and iii) defining the information flow to allow farmers to exploit maps produced. This last aspect was fundamental to provide information in time for supporting the nitrogen fertilization and test the effect of site specific applications. Two Worldview-2 images were acquired in June and July before rice fertilization periods and two Rapid-eye images were acquired after the fertilizations. The first two images (Worldview) provided the base maps to target variable fertilizations while the Rapid-eye were used to verify the effect of the applications. A final evaluation was also done by comparing the maps delivered during the season with the rice yield maps produced with the harvester at the end of the season. This operational test demonstrated how the use of information related to within field variability generated in near-real time from satellite imagery, is a suitable alternative to the traditional use of previous year yield maps for the management of variable rate dosage of nitrogen.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
