Environmental and agricultural applications
Crop modeling and yield forecasting systems are valuable tools for the agricultural sector, administrations and scientific research. Hence, there is an increasing number of yield forecasting platforms emerging from private and public entities. Public platforms in Europe provide free access to accurate nationwide crop yield forecasts and seasonal meteorological insights. However, these results may have limited interest for Spanish agricultural producers and related organisations. Producers operate at lower scales where there is a large and uncaptured agro-ecological variability. Taking advantage of the open data infrastructure in Navarre, we developed a regional forecasting system to simulate two major species in the area (wheat and barley). The forecasting system aimed to capture the inter-annual crop yield variability at the district level using an ensemble of bootstrapped partial least squares regressions. Regressions consider bimonthly and quarterly meteorological indicators as well as seasonal remotely sensed vegetation indices. Forecasts are being published through online annual reports. Experiences so far show that the forecasting system is in its infancy. Accordingly, methodological improvements have been planned to markedly reduce simulation errors in upcoming seasons.
Find out more.