Change detection for road condition mapping based on dry and rainy season VHR dual-pol radar imagery over tropical areas. Application to Chad

New Very High Resolution dual-polarization radar sensors in X-band (TerraSAR-X and Cosmo-Skymed) are offering technical opportunities for the detection of road condition changes between dry and rainy period in tropical areas. We applied classification methods (fuzzy logic and learning algorithms), multi-incidence and optical-radar combination in order to estimate surface elements (relief, flooded areas), soil properties (wetness, surface roughness) and road conditions from multi-date acquisitions over SW Chad. These estimations are calibrated with data measurements on the field. They are included in a road condition service designed for military, humanitarian and industrial logistics.

I. Introduction

In difficult areas, road conditions mapping relies on surface properties analysis extracted from pedological, climatological, geological and hydrological dynamically correlated parameters. Every year in South-Western Chad, monsoon rises towards the North, and heavy rains cause Logone - Chari rivers to overflow their alluvial plains. This has a positive impact as floods dwell on wide areas as long as vegetation growth. But saturated soils reduce mobility over the whole region. Road condition modeling over tropical areas requires a continuous monitoring. As ground truth measurement equipments are not available (no weather radar, no reliable hydrologic measurement), high resolution satellite sensors provide added capabilities. During monsoon, VHR radar sensors should give a good opportunity to observe landcover evolution in order to provide road condition mapping during the most stressful season.