Geospatial Data Science for Soil and Paddy Productivity Monitoring in Sri Lanka: A Multi-Source Public Dataset Analysis

Abstract

Planetary-health-oriented soil and crop monitoring depends on integrating multiple, independently collected public datasets across a shared geographic frame. This paper presents a geospatial data science analysis of paddy productivity and soil organic carbon (SOC) in Sri Lanka, built entirely on four publicly available real-world datasets: the national open 3D digital soil database of Sri Lanka; a digital soil organic carbon

mapping study for paddy soils in Northern Sri Lanka; district- level paddy extent, yield and production statistics published by

the Department of Census and Statistics (DCS) of Sri Lanka; and standard Sri Lankan administrative district geography used to georeference the statistical data. We integrate these sources to map district-level paddy production contribution, track a three-year national yield trend, and compare SOC digital-mapping accuracy across depth and geography. Five geospatial visualisations and three summary tables are presented. The analysis identifies a small number of dry-zone districts (Ampara, Polonnaruwa, Kurunegala) that jointly account for a disproportionate share of national paddy production, a declining national average yield trend across the three most recent reported Yala seasons, and a consistent pattern of higher SOC-mapping

accuracy in subsurface relative to topsoil layers across two inde- pendent digital soil mapping studies. We discuss implications for

prioritising monitoring investment geographically and conclude with a lightweight spatial hotspot classification that requires no specialised geostatistical training to reproduce.