An Operational Remote Sensing Based Service for Rice Production Estimation at National Scale
Holecz, Francesco1; Barbieri, Massimo1; Collivignarelli, Francesco1; Gatti, Luca1; Nelson, Andrew2; Setiyono, Tri Deri2; Boschetti, Mirco3; Manfron, Giacinto3; Brivio, Pietro Alessandro3

Reliable and seasonally updated information on rice area, phenology, crop status, and yield are key information for policies related to reduce food insecurity and to guide decisions for better management of agriculture and resources. Spaceborne remote sensing iV particularly through the use of Synthetic Aperture Radar (SAR) iV combined with crop growth simulation modeling offers an effective alternative to conventional terrestrial methods, that are time consuming, expensive, and in some cases providing questionable rice production figures

One goal of the Remote Sensing based Information and Insurance for Crops in emerging Economies (RIICE, iV financed by the Swiss Development Cooperation) is to estimate, on an operational basis, rice production at national scale in primis targeted to food security purposes. Two are the technical novelties of this operational service: 1. Multi-temporal SAR data acquired from all existing operational spaceborne SAR Stripmap and ScanSAR systems are considered and complemented by quasi-daily medium resolution optical data. This solution enables therefore:

  • to overcome the spatial-temporal problem, hence assuring even over large areas an appropriate temporal repetition at an adequate scale (i.e. spatial resolution);
  • to provide operational monitoring being sensor independent and based on data redundancy.
    2. In addition to the usual meteorological, soil, and plant parameters, products obtained by remote sensing products are ingested into the yield crop model enabling:
  • to reduce the amount of non-remote sensing parameters;
  • to include relevant information on the rice phenology;
  • to take into account to the spatial distribution of rice fields;
  • to overall improve yield estimation figures.

    In collaboration with national partners, the RIICE service is today national-wide employed in six countries and two states in South-East Asia. From a remote sensing perspective ENVISAT ASAR (Wide Swath and Image Mode complemented by Alternating Polarization), ALOS PALSAR-1 (Fine Beam Single and Dual Beam), Cosmo-SkyMed (Stripmap and Huge), Radarsat-2 (Standard Beam), RISAT-1 (Medium and Fine Resolution) and 8-days MODIS composites data are supported. Due to the large amount of data iV tens of thousands of images iV an automated processing chain (MAPscape - RICE) based on a low cost local cluster solution has been developed. Rice products (area, start of season, peak of season, phenological stages, and flood/drought damages, Leaf Area Index) generated from multi-temporal multi-sensor remote sensing data are subsequently ingested into an upgraded version of Oryza2000, hence enabling to estimate yield, and finally, production figures at a given administrative level. Intermediate and final products are validated by the national partners by means of a standard procedure.

    It is well known that today SAR data availability is problematic, in primis due to the failure of the ENVISAT ASAR and ALOS PALSAR-1 systems. Hence, the current use of RIICE products is partially hampered, but the potential to expand the service is clear. Nevertheless, today, at the ground segments a large (tens of thousands unexploited) scenes archive are existent. These data have been exploited, in order to generate, in a first round, rice baseline maps at 1 hectare and 15 meter resolution representing, in most of the cases country-wide, the most updated rice information. On one hand, these maps serve to drive (in terms of space and time) the new acquisitions, on the other hand, due to the procurement of very high resolution SAR data, to enhance the level of detail of these maps. The launch of Sentinel-1A (and subsequently 1B) and other scheduled SAR systems will doubtless assure the sustainability of the service and its extension at continental scale.