LAUSR

Abstract Over fertilization has resulted in serious soil compaction and acidification, as well as nonpoint pollution. A strategy for variable fertilization according to Site-Specific Management Zones (SSMZ) is urgently needed. We developed an object-oriented method using image segmentation for SSMZ delineation for a farming cooperative founded in 2013 in northeastern China. The method is based on Normalized Difference Vegetation Index (NDVI) during the crop growth period from 2011 to 2016. We analyzed and validated SSMZ results with plant sampling data, NDVI, soil temperature and moisture. The results indicated that (i) SSMZs derived from NDVI and image segmentation enhanced the difference between SSMZs and the homogeneity within SSMZs; (ii) crop dry biomass was consistent with NDVI, and NDVI showed a clear change in the regular SSMZ pattern with terrain variation after 2013; and (iii) SSMZ patterns were mainly affected by different farming practices or by topographic factors before and after the cooperative was founded. This new approach for SSMZ delineation will accelerate the application of precision agriculture in relief areas. If the study results are applied to other areas, the main factors of SSMZ need to be determined because of climate differences as well as specific soil and terrain conditions.