Topdressing nitrogen recommendation for early rice with an active sensor in south China

Fertilizing based on soil test and crop nitrogen (N) demand is the key to optimize yields and minimize fertilizer cost. In 2008, a field experiment with different N rates was conducted with early rice near Yingtan City, Jiangxi Province, in southern China. Canopy normalized difference vegetation index (NDVI) with an active sensor and plant N uptake (PNU) were collected at key fertilization stages; and the sufficiency index (SI) was calculated as the ratio of under-fertilized and well-fertilized NDVI. Rice PNU and yield were positively correlated with NDVI and SI at the tillering and panicle initiation stages. Canopy SI improved the PNU and yield estimations when the relationship was validated with a different dataset. A spectrally-determined N topdressing model (SDNT) was established and used in combination with a target yield strategy and split-fertilization scheme. An allocation coefficient for plant N requirement to accommodate the potential for high yield and soil N supply was introduced. Optimum nitrogen use efficiency (NUE) at different growth stages was incorporated into the model. The model was validated with data from a 2009 plot experiment and three production fields in 2010. The difference of recommended N rate and yield between SDNT and the current yield curve recommendation method was 2.1 and -0.7 % at high planting density and -2.4 and -4.8 % at low planting density, respectively. Compared with farmers' N management, the SDNT strategy resulted in similar or higher yield with reduced N rates, higher NUE and higher net profit in both 2009 and 2010. Because canopy NDVI can be obtained while sidedressing N fertilizer in a single field pass, the potential of SDNT to accommodate within-field spatial and temporal variability in N availability should improve N management in rice.