UC Davis

Improving nitrogen (N) management is critical to maximizing the productivity and sustainability of our agroecosystems. Improving N management requires an understanding of crop N status and yield potential early enough in the growing season when changes to N management can influence yields. However, given the lack of tools currently available to accurately assess crop N status, farmers continuously face the challenge of determining whether their crops require additional N fertilizer. The recent emergence of remote sensing technology has provided a promising alternative that can provide farmers the information they need in an accurate and timely manner. Several studies have demonstrated the potential of remote sensing technology to accurately assess the health and vigor of vegetation at the landscape scale, however few have explored how this technology can be utilized to inform sustainable crop management at the farm scale. This knowledge gap is what inspired this research and led us to investigate how remote sensing technology can be utilized to improve in-season N management in California rice systems.In California, where more than 200,000 ha of flooded rice (Oryza sativa) is cultivated annually, the recommended N management strategy is for farmers to apply the average seasonal N fertilizer requirement prior to flooding and planting as aqua-ammonia injected into the soil. On-farm studies have reported that N fertilizer applied in this manner is efficiently utilized by the crop as it remains protected from denitrification and ammonia volatilization losses until the crop needs it. At panicle initiation (PI), it is recommended to assess crop N status to determine if additional N fertilizer inputs are required as top-dress. The current tools available to assess rice N status include the SPAD chlorophyll meter and Leaf Color Chart, but these tools are not often used as they are time consuming and subjective. Thus, most top-dress N applications take place without evaluating crop N status; possibly resulting in inefficiencies due to over application. Our goal was to improve N management in California rice systems by developing a sensor-based decision support tool that could guide California rice farmers in their mid-season top-dress N management. This was pursued through N response trials that were established over a 4-yr. period across fourteen on-farm locations throughout the Sacramento Valley rice growing region of California. At PI, Normalized Difference Vegetation Index (NDVI) was measured using both a proximal crop sensor and a multispectral aerial sensor, and Normalized Difference Red-Edge Index (NDRE) was measured only using an aerial sensor. After NDVI and NDRE measurements, biomass was sampled destructively and then top-dress N fertilizer was applied. At maturity, rice plants were harvested to quantify grain yield. In the first chapter, our objective was to determine which N status parameter is best assessed by NDVI at PI and how accurately NDVI at PI can predict grain yield. The N status parameters quantified in this study were aboveground biomass, plant N concentration, and total N uptake. Quadratic linear regression models were developed to describe the relationship between each N status parameter and NDVI, and a simple linear regression model was developed to describe the relationship between grain yield and NDVI. Our results showed that PI N status was best assessed by NDVI when quantified as total N uptake and that NDVI at PI was positively correlated with grain yield. However, our results also showed that NDVI saturated once crop N uptake exceeded a certain threshold, suggesting alternative indices that do not saturate may provide a basis for a better assessment. In the second chapter, our objective was to compare the sensitivity of aerially sensed NDVI and NDRE to proximally sensed NDVI for assessing rice crop status when quantified as PI N uptake and grain yield. In order to make direct comparisons across the three indices, the raw values from each index were normalized by calculating the Sufficiency-Index (SI). Quadratic-plateau linear regression models were developed to describe the relationship between each SI and PI N uptake and linear mixed effects models were developed to describe the relationship between each SI and grain yield. Our results showed that aerial NDRE SI and proximal NDVI SI were similarly sensitive at assessing PI N uptake and grain yield, whereas aerial NDVI SI was poorly sensitive. The difference in sensitivity among the three indices was attributed to the relative amount of saturation of each index. Our finding that both the aerial NDRE SI and proximal NDVI SI measured PI rice crop status effectively provides a unique advantage for end-users as it allows them the flexibility to choose the sensor most suitable for their goals. In the final chapter, our objective was to develop a NDVI Response-Index capable of predicting the grain yield response to top-dress N fertilizer applied at PI. The NDVI Response-Index was developed by comparing the NDVI of each field treatment to the NDVI of a N non-limiting plot. At PI, top-dress N fertilizer was applied to every plot, and at maturity grain yield was quantified. A linear mixed effects model was developed to describe the relationship between NDVI Response-Index and grain yield with and without top-dress N. An economic analysis was performed to determine the magnitude of grain yield response required for top-dress N applications to be economically feasible. Based on our results, we found that top-dress N applications become profitable once NDVI Response-Index exceeds 1.07 by PI. The NDVI Response-Index presented here provides a useful tool for farmers to make precise mid-season top-dress N decisions which can result in positive outcomes for both crop productivity and environmental sustainability.