Towards Data-intensive, More Sustainable Farming: Advances in Predicting Crop Growth and Use of Variable Rate Technology in Arable Crops in the Netherlands
1C. Kempenaar, 2C. Kocks, 1T. Been, 1F. van Evert, 3S. Nysten, 2K. Westerdijk
1. WUR-Plant Research International
2. UAS-CAH Vilentum
3. CAH Vilentum
Precision farming (PF) will contribute to more sustainable agriculture and the global challenge of producing ‘More with less’. It is based on the farm management concept of observing, measuring and responding to inter- and intra-field variability in crops. Computers enabled the use of Farm Management Information Systems (FMIS) and farm and field specific Decision Support Systems (DSS) since mid-1980s. GIS and GNSS allowed since ca. 2000 geo-referencing of data and controlled traffic farming. Several types of soil and plant sensors provided site specific data on spatial variation in crops. Today we see the development of several cloud based data platforms, and apps for soil and crop monitoring and site-specific crop care. This R&D is likely to continue in the coming years, yielding more apps for tactical decisions and operational interventions in crops, and strategic decisions on more-complex crop rotation issues. PF requires these developments, needing ‘big-data’ to produce more with less.
In this paper, we show results of three research topics in which we evaluate 1) correlations between remote and near biomass sensing data, 2) correlations between biomass and yield sensing data and 3) the use of task maps based on biomass sensing. The studied crops are common in The Netherlands: winter wheat, potato and onion.
The studies showed acceptable correlation between remote and nearby measured biomass data. It is essential to remove irrelevant variation in order to get better biomass maps that can be used for yield prediction and task maps. In general we showed poor correlations and irregular trends in the correlation between biomass indices and final yield (winter wheat and onion). The correlation improved when seasonal mean biomass index was used.
Finally, we showed two examples in which biomass maps were successfully used in task maps for chemical haulm killing and N topdress fertilizer use. The task maps were made within the web-based Akkerweb GIS-platform (http://www.akkerweb.nl/). Inputs were reduced by 15 – 30 % when the task maps were applied.
