Earth Observation for Accurate Mapping of Soil Functional Properties, Land Health and Soil Infiltrability in Rwanda

Earth observation (EO) has a large potential for mapping of soil functional properties such as soil organic carbon, soil pH or acidity, soil fertility parameters and soil texture. Recent advances in the application of EO data in combination with systematic field data sampling, standardized soil data reference analysis and the use of soil spectroscopy have shown these approaches to be both robust and scalable. We present a case study from Rwanda where we apply EO data in combination with field and laboratory data collected using the Land Degradation Surveillance Framework (LDSF) to map functional soil properties, soil erosion prevalence and land cover at fine spatial resolution. Digital soil maps were produced at a spatial resolution of 30m with an accuracy of 85 to 90%, while soil erosion prevalence was mapped with an accuracy of 86% using Landsat satellite imagery and machine learning models. 

We also assess interactions between spatial assessments of soil organic carbon, soil erosion prevalence and land cover at a spatial resolution of 30m in order to identify land degradation hotspots and better target interventions to restore degraded land across four districts in Rwanda. We further explore the effects of soil erosion, root-depth restrictions and soil organic carbon content on saturated hydraulic conductivity in three LDSF sites in Nyagatare, Kayonza and Bugesera districts, respectively. Saturated hydraulic conductivity was modeled based on single-ring measurements of infiltration capacity using a modified Reynolds & Elrick steady-state single ring model for 48 LDSF plots per site. The results show significant spatial variation in infiltrability within sites.

The results of the study show the importance of rigorous protocols for sampling and analyses of soil properties and indicators of land health across landscapes. By simultaneously assessing soil properties, indicators of land degradation and soil infiltrability we demonstrate the utility of these approaches in understanding drivers of land degradation across multiple spatial scales for targeting of options for land restoration and monitoring of the effectiveness of these interventions over time across multiple dimensions of land health.