Identifying with interested AMS the main gaps in the current data available and building on this a strategy for generating an enhanced database, targeting soil health across priority land use/cropping systems

Agriculture is one of the important sectors of the Southeast Asian (SEA) region, contributing 11.4% to the regional GDP. Diverse conditions in climate, vegetation, physiography, geology, and lithology have resulted in 19 complex soil geographical areas, creating 11 farming systems, of which four are considered significant. Soil health has been recognized as key to a successful and productive cropping system, giving rise to sustainable agriculture. Consequently, the development of soil health data is strategically essential to identify local, national, and regional trends for prioritizing appropriate policy and investment decisions and extension programs. For this purpose, the emphasis on the development of satellite and other aerial sensor technologies could be highlighted to capture relevant data at a macro scale. The Alliance Bioversity-CIAT duly noted that the soil health of the SEA region had not been systematically studied using contemporary concepts and technologies. This short-term consultancy project has been proposed as an open-ended literature review to address this gap as much as practically possible, strategically prioritizing Vietnam, Cambodia, and Thailand (Fig-1). The report is divided into three chapters. Chapter 1 reviews available literature encompassing three priority countries of the SEA region mentioned above. The first part of this chapter includes a comprehensive review of the Soil map of the world, 1:5000,000 series of FAO-UNESCO, relevant articles, and their critiques. It was revealed that 19 soil-geographic types of the SEA region resulted in 20 associations distributed in the selected priority countries (Fig-2). These associations were further elaborated with their locations, parent materials, possible cropping systems described by FAO, together with potential land use. The second part includes a further review focusing on the concepts of soil health and the relevant situation in the SEA region. The Intergovernmental Technical Panels on Soil and FAO’s voluntary guidelines for healthy soils were underpinned in this section. The University of Florida’s soil data model project, which might serve as the guide for future works of The Alliance Bioversity-CIAT, was briefly described in this section. Finally, a link to the snapshot of The World Soil Information Service (on page 19) and indicators of soil health developed by the US Department of Agriculture was also included here. Within time constraints, Chapter 2 delved into published literature, analysing soil with remote sensor (RS) technology, primarily from satellite platforms. The chapter began with a comprehensive literature review article from Ohio State University that assessed 3047 papers published within the last 10 years. The authors of this article commented that the agriculture sector is yet to fully implement RS technologies and revealed that only 33% of papers originated from outside Europe, the USA, and China. Furthermore, Chapter 2 includes brief reviews of published papers related to soil health, topography, soil moisture, land use, soil organic carbon, and soil microorganisms (Chart-1). Various authors employed different technologies in their research, yet a clear trend emerged in utilizing spectral reflectance from multi to hyperspectral RS images to analyse soil attributes obtained from indirect sensors mounted on satellite platforms. Subsequently, vegetation indices and AI algorithms were employed to unveil the results. The third and final chapter explores and discusses the scopes and limitations of using RS technology in agricultural soil health analysis. Sensors capturing near-infrared and thermal data, hyperspectral imagery, free low-resolution imagery, LiDAR data, etc., together with advanced machine learning algorithms and various tools and instruments, reveal various levels of potentiality in soil attribute analysis. On the flip side, common limitations of using RS technology include the large size of hyperspectral data, high economic costs, unsuitability of large spatial data for small projects, deterrent weather factors like clouds and snow, and the unavailability of appropriate training data for machine learning algorithms. However, recent developments in space-edge computers and satellite-based LiDAR technology are expected to bring further positive changes to the use of RS in soil analysis.