Spatializing the Coffee Yield Model SAFERNAC with Soil Fertility Data Across Tanzania

The aim of this work was to explore the behavior and usability of the new model SAFERNAC over coffee growing areas throughout Tanzania. Soil fertility data from 1,131 georeferenced points in three zones were fed into the model under four distinct approaches – baseline (no input), organic (manure), inorganic (NPK) and combination of manure and mineral fertilizer. The simulated yields were descriptively compared per zone. They were loaded into QGIS 3.2, interpolated using the Inverse Distance Weighting (IDW) algorithm and the resultant raster maps clipped on basis of digitized boundary shapefiles. Baseline yields were effectively computed from 99.2% of the surveyed sites. The model showed high sensitivity to pH, which has a greater influence on P than N or K. Calculated yields decreased in the order Zone 2 > Zone 1 > Zone 3. The difference in yield between NPK 160:80:80 alone and a combination of NPK 80:40:40 (half dose) plus 5 tons manure was neither quantitatively nor spatially significant. SAFERNAC has proved its usability across the Tanzanian coffee soils, in simulating yield of parchment coffee. The combination approach (organic materials and mineral fertilizers) is most appropriate, as it can reduce the fertilizer cost by about 50% without seriously compromising the expected yields.