iPot: industrial potato monitoring for the Belgian potato sector using remote sensing and crop growth modelling

Joost Wellens, chercheur post-doc
Courriel: Joost.Wellens@ulg.ac.be

Financement: BELSPO

The Belgian potato processing industry has known a spectacular growth during the last decades. The companies active in this sector (most of them family owned enterprises) have succeeded – together with primary producers and traders – to grow from 500.000 tons of processed potatoes in 1990 till 3.650.000 tons of raw material in 2012. When considering the production (in tons) per habitant, Belgium is ranked on the 4th place with a production of 395 tons per 1.000 inhabitants. Nowhere in the world are higher yields obtained than in Belgium. To keep up its position on the forefront of innovation, the potato processing industries are continuously looking at promising novel technologies that could improve business processes at any stage of growth. The close follow up of parcels on the land as form above is becoming an important tool to improve the quantity and the quality of the potato crop, and reduce the risks in order to plan the storage, packaging or processing and as such to strengthen the competiveness of the Belgian potato in a global market.

The iPot project, financed by the Belgian Science Policy Office (BELSPO), aims to provide the Belgian potato sector, represented by Belgapom, with near real time information on field condition (weather-soil) and crop development, and with early yield estimates, derived from a combination of satellite images and crop growth models.

Satellite images are used since long for crop monitoring at regional level by the public sector. Traditionally, low to medium resolution images, with pixel sizes from 250m to 1km are used for providing information on crop growth and development, and on yield at regional scale. However, the recent launch of constellations of high resolution optical sensors (DMC, Sentinel-2) offering spatially detailed information (20m pixel size) at high frequency (every 2 to 3 days) and at low cost, is creating a (r)evolution in the world of crop monitoring. Automatic processing chains have been developed to derive vegetation indices (NDVI) and biophysical parameters (fAPAR, LAI, fCover). Maps are being derived on a regular basis with for each pixel the actual development stage of the potato crop.

This on-line scanning of the growth (phenology) is expected to result in an increased performance of crop growth models and better yield predictions. Once the data for crop phenology are incorporated into the model (B-CGMS), the yield data collected at the parcel level will be assimilated into the model using optimization algorithms to minimize the discrepancies between observed and simulated yield time series in order to improve further yield estimations.

The combination of these new earth observation and modelling techniques arose the interest of a new type of users, mainly from the private sector (agro-industry, agricultural insurers, etc.). The iPot project aims to bridge the gap between the latest research efforts regarding crop growth monitoring and the industry. An intuitive web based geo-information platform is being developed to allow both the Belgian potato industry and research centers to access, analyze and combine the data with their own field observations in closed collaboration with the farmers, for improved decision-making.