Integrating Statistical and Earth AgriData in Small Farming Systems for Food Security

The present work unveils the role of small farming plots (less than 5 ha) in the context of food security. It determines their contribution by estimating the spatial distribution (location), the crop types (diversity), the crop area extent (acreage), and the yield (production), factors that remain unclear, mainly because the official statistical offices rarely include them in surveys. The development introduces a novel RS-based approach that fulfills this gap. It provides stakeholders with the appropriate tools to accurately and timely acquire crop type map information and objectively quantify their crop production capabilities. Approaches based on the Land Parcel Identification System (LPIS) of the Integrated Administration and Control System (IACS) applied by many countries in Europe are proved useful in providing information on location, diversity, and acreage but not crop production per farm owner applying and eligible for receiving subsidies. The developed RS approach is implemented in twenty European NUTS-3 regions and one in Africa. Nevertheless, in this research, we focus on its development, testing, and evaluation in three pilot prefectures of Greece, producing the corresponding land cover maps. Notably, the unbiased crop area computation and the crop production estimates are performed only for the highly accurate key crop products (per crop type classification, FScore > 75%), considering that the key crop production estimations are obtained by combining the key crop areas with the field-level yields provided by the key informant surveys. The above choice ensures that the estimation of crop production will be derived only for the best-classified crops per reference region. The RS approach reduces the error propagation when estimating the area and production of the crop types that are classified with low or very low accuracy levels. These levels could reduce the strength of the overall conclusions about the main contributions of small farming plots. Potential changes occurring in the key crop cultivations of small farming plots are also estimated and mapped using the LPIS geodatabase. Under various environmental and territorial conditions, the results of the RS approach show good classification accuracies for several key crops per reference region. Their integration with the existing official statistical data and those derived from the LPIS geodatabase shows the consistency and significant contribution in estimating all the factors needed to determine the small farming plots. Finally, the applied innovative integrated approach can be expanded beyond the Greek case to cover other regions with various agricultural practices.