Precision Agriculture at Scale in Senegal

Harch Agriculture designed and deployed an integrated precision farming platform — HarchOS Agri — that transforms CSAD's 120,000-hectare operation from uniform management to site-specific, data-driven agriculture.

The foundation layer is a comprehensive sensing infrastructure. Harch deployed 3,600 soil moisture sensors across the estate — one sensor per 33 hectares on average, with higher density in areas of known soil variability. Each sensor measures volumetric water content at three depths (15cm, 30cm, 60cm), soil temperature, and electrical conductivity at 15-minute intervals. Complementing the ground sensors, Harch equipped 24 agricultural drones with multispectral cameras (Red, Green, Blue, Near-Infrared, Red-Edge) that survey the entire estate on a 5-day cycle, generating vegetation health indices (NDVI, NDRE, SAVI) at 10-centimeter resolution. Two fixed-wing survey drones provide monthly high-resolution orthomosaic maps at 3-centimeter resolution for crop stand establishment assessment and weed mapping. Additionally, 84 automated weather stations provide real-time microclimate data — temperature, humidity, wind speed, solar radiation, and rainfall — at 2-kilometer spatial resolution, enabling hyperlocal weather forecasting for spray scheduling and irrigation timing.

The second layer is AI-driven irrigation optimization. HarchOS Agri ingests soil moisture data, weather forecasts, crop growth stage models, and evapotranspiration calculations to generate field-specific irrigation prescriptions. Instead of flooding entire 500-hectare sectors uniformly, the system divides each sector into management zones of 5-10 hectares based on soil type and topography, and prescribes different irrigation durations for each zone. The prescriptions are delivered directly to CSAD's irrigation control system, which activates and deactivates pumps and valves according to the HarchOS schedule. The optimization algorithm simultaneously minimizes water application (reducing pumping costs) while ensuring that soil moisture never drops below the crop-specific threshold that triggers water stress. Where CSAD has invested in drip irrigation infrastructure — approximately 18,000 hectares for high-value tomato and onion production — HarchOS provides daily scheduling for each drip zone, adjusting flow rates and durations based on real-time soil moisture feedback. The system has reduced total water application by 28% while improving crop water satisfaction from 64% to 97% of optimal.

The third layer is predictive pest and disease management. The drone multispectral surveys detect vegetation stress up to 14 days before visible symptoms appear to the human eye, by identifying changes in chlorophyll fluorescence and leaf water content that are imperceptible in visible light. When the system detects an anomaly, it automatically triggers a targeted drone survey at higher resolution and alerts CSAD's crop protection team with a GPS-precise location and a differential diagnosis of likely causes based on the spectral signature, crop growth stage, and local weather conditions. The early warning capability has been particularly transformative for fall armyworm management: HarchOS detects initial infestations an average of 11 days before visible crop damage occurs, enabling targeted biological control applications that are 92% effective at preventing yield loss, compared to the 40% efficacy of reactive chemical treatments applied after visible damage appears.

The fourth layer is variable-rate fertilizer application. HarchOS generates field-specific fertilizer prescriptions based on soil nutrient maps derived from 12,000 soil samples (one per 10 hectares), drone-derived vegetation health indices, and crop removal rates from previous harvests. The prescriptions are loaded into CSAD's variable-rate applicators, which adjust fertilizer blend and application rate on-the-go as the machine traverses the field, applying more nutrients where soil is deficient and less where it is sufficient. The variable-rate approach has reduced total fertilizer application by 22% while increasing the proportion of applied nutrients that are actually taken up by the crop from 47% to 78%, eliminating both the economic waste and the environmental damage of blanket application.