On-farm Evaluation of Optical Sensor Technology for Variable Rate N Application to Corn in Ontario and Quebec Bao-Luo Ma, and Nicolas Tremblay Eastern Cereal and Oilseed Research Centre, Ottawa, ON 613-759-1521, mab@agr.gc.ca Canada

Introduction - Nitrogen Fertilizer N represents the most costly input in grain cereal production, especially corn Producers have to balance crop N needs while minimizing N losses via leaching, emission and runoff NUE is low, about 50% Soil physical, chemical and biological conditions tremendously affect NUE N losses as NO3 to surface and ground water, as N2O and NH3 gases to the atmosphere As we know, fertilizer N represents the most costly input in grain cereal production, especially corn. So my talk will focus on corn N management while also mention other elated projects. Corn growers have to balance crop N needs to maximum NUE and yield, at the same time, to minimize gaseous N loss and NO3 leaching. In general, NUE, meaning N applied to the soil and ended up in the plant, is only about 50%. Soil physical, chemical and biological conditions have an large impact on NUE and yield N losses is the major concern for corn production Agronomic research on corn is thus focused on NUE, what is the mechanism of plant and soil factors associated with NUE? How can we improve NUE by improving soil physical, chemical and biological conditions?

Annual Quantitative N Cycle for Corn/Soil/Atmosphere (kg N/ha) Atmospheric N 10 24 10 Corn Crop N (Grain=123; Stover=61; Roots=22) 83 192 Soil Organic N (6000) 14 252 NO3-N - 120 HN4-N + 14 24 150 23 Fertilizer N Here is a diagram to show N transformation in a corn yield. Let’s starting with soil, on the top 30 cm of soil layer it contains typically 6000 kg of total N, assume 2% of which is released through N mineralization, it first in a form of NH4-N, which quickly converts to NO3-N through nitrification. So that majority of N uptake by plants is in the form of NO3 while only a small portion in NH4. Farmers usually apply 150 kg N /ha, of which 50% is taken up by plants, some of it is fixed into soil organic matter, a considerable amount of N is lost through volatilization, dinitrication, run off or leaching. Nutrient best management practices are aimed to increase NUE while reducing the loss of N to the environment. In the agricultural soils of tropical and temperate regions (except paddy rice fields) nitrate accounts for 70 to 90% of soil mineral N. Nitrate in the soil solution is 10 times more mobile than NH4+. Two reserves of soil NH4+ exist: a small portion of NH4+ is in the soil solution, and a large portion of NH4+ is adsorbed on soil colloids. Only water soluble NH4+ is available for plant uptake.

The basic relationship Chlorophyll Nitrogen Grain Yield As we know, chlorophyll is the factory for producing food, feed, fibre, and biofuels for mankind. It reflects plant healthiness and productivity. Over 70% of plant total N is in the chlorophyll. Under normal production conditions, these three parameters form a quantitative and stable relationship.

Canopy reflectance x Nitrogen Level > N > DW > LA= > reflectance 40 35 High N 30 25 Reflectance (%) 20 Low N 15 < N < chlorophyll= > reflectance 10 5 400 500 600 700 800 900 Wavelength (nm) Numerous studies have shown that canopy reflectance is closely related to leaf greenness and total biomass. At the red and green wavelength regions, leaf greenness is negatively correlated with leaf chlorophyll content while aboveground biomass is positively correlated with reflectance in the near infrared regions. By integrating these two, NDVI is able to reflect the overall plant N status in a field.

Critical levels 58,0 Y = 33,928 + 0,654x - 0,004x2 R2= 0,98 58,0 55,3 55,3 52,1 52,1 Spad 45,4 45,4 V3 V6 V10 Silking Development stages Figure 1 – Chlorophyll meter mesurements in four corn development stages(Argenta, 2001). When determine the critical levels of NDVI or SPAD, it is important to keep in mind that these values change rapidly from one growth stage to the next

Suficiency index (SI): NDVI in the field SI = = 0.95 NDVI in the high N reference strip Strip Corn field In practice, in a corn field, plant some strips with well fertilized corn as reference, by measuring canopy reflectance in different parts of the field in comparison with the reference strip, we will be able to derive a sufficiency index and decide how much N to be applied at sidedress.

ETAA Project Objectives To develop a crop-based indicator for corn N management that accounts for spatial variability; To determine if the new vehicle based optical sensing technology is sufficiently robust and practical for on-farm use; To determine if variable rate N application based on spectral reflectance can improve NUE and corn performance on a farm scale. In 2005, a new project was support by the environment technology assessment for agriculture program through the Ontario Soil and Crop Improvement Association to demonstrate on farm evaluation of optical sensor technology to minimize environmental impacts and maximize production efficiency associated with N application in corn. So, the overall objectives were to determine

ETAA Variable Rate Nitrogen Study - 2005 960ft = 292.6 m 30ft * * * * 7 7 9 7 * * * * 9 * 17 * * 27 35 * * * * * 5 3 2 1 6 9 4 * 3 4 5 9 6 * 1 2 6 * 5 1 7 3 4 2 2 9 * 5 4 1 6 3 8 * 16 * * 26 34 * o +o o o * * * * 10 * 8 * * 10 10 * * * * * 30+60 7 15 25 33 o +o + +o + * o +o +o * 30+60 * 30+120 30+120 * 8 * o + * +o o + * +o o o o * + +o + 1 2 3 4 5 6 10 * 11 12 13 14 18 * 19 20 21 22 23 24 28 29 30 31 32 * 36 37 38 39 40 Rep 1 Rep 2 Rep 3 Rep 4 North Nitrogen Rates Hybrid Dimensions 1) 0 kg N/ha preplant Pioneer 39D 80 (2550 CHU) 12 rows per plot 2) 30 kg N/ha preplant (26.7 lb/acre) - (Roundup Ready + Poncho 250) 30" rowsplot width = 30 ft 3) 60 kg N/ha preplant (53.4 lb/acre) 4) 90 kg N/ha preplant (80.1 lb/acre) plot length = 80 m (for treatments 1-6), Nitrogen Type 5) 120 kg N/ha preplant (106.8 lb/acre) 6) 150 kg N/ha preplant (133.5 lb/acre) 1) Urea (46-0-0) broadcast preplant 20 m (for treatments 7-10) 7) 30 kg N/ha preplant + 30 kg N/ha sidedress at V6 2) Urea (46-0-0) sidedressed at V6 8) 30 kg N/ha preplant + 60 kg N/ha sidedress at V6 Total Area 9) 30 kg N/ha preplant + 90 kg N/ha sidedress at V6 10) 30 kg N/ha preplant + 120 kg N/ha sidedress at V6 2.34 ha = 5.8 acres * 30 kg N/ha preplant + variable rate N at V6 Revised June 29, 2005 Here is the diagram of the field layout. We had 6 rates of preplant N and 7 rates of preplant plus sidedressed N conducted in a 6 acre field.

Canopy reflectance of the field was mapped both with handheld GreenSeeker and vehicle-mounted N Sensor at the V6 stage.

Responses of NDVI_GS and NDVI_NS to fertilizer N rates   Applied N dose (kg N ha-1) Contrast analysis 30 60 90 120 150 Linear Quadratic Residual NDVI_NS a 30DAS 0.35 0.40 0.43 0.41 * (41%) b ** (57%) NS NDVI_GS a 26DAS 0.37 0.38 (18%) (58%) 33DAS 0.42 0.46 0.48 0.50 0.47 (47%) (44%)   This table shows the responses of canopy reflectance measured by both N Sensor and GreenSeeker shortly before sidedress in 2005 growing season. In general, there was a linear relationship between NDVI and preplant N rates, NDVI saturated between 90 and 120 kg N/ha, and both instruments had very similar results.

GreenSeeker measures canopy differences at pre-sidedress (2005) This slide shows the quantitative relationships between canopy reflectance and preplant applied fertilizer N rates.

Saturation index shows seasonal N needs (Qc data) Yield difference due to N (t/ha) In Quebec, Nicolas Tremblay found an relationship between yield difference due to applied N and leaf chlorophyll saturation index.

Grain yield * * * In 2005 growing season, we manually applied variable N. This slide shows the overall yield of each treatment. Last year, corn yield was exceptionally high, up to 14 t/ha was obtained with 30 plus 120 kg N treatment. when we compare these 3 treatments, of which all received 120 kg N/ha, but applied at different time, it clearly shows the advantage of N sidedress.

) Grain yield (Mg ha Fertilizer (kg N ha ) 15 y = -0.0002x -1 14 R 2 = 0.979 13 12 11 2 10 y = -0.0002x + 0.066x + 7.3 Grain yield (Mg ha 9 R 2 = 0.995 8 7 6 30 60 90 120 150 Fertilizer (kg N ha -1 ) Preplant 30+sidedress Response of grain yield to fertilizer N followed a quadratic function. Clearly, grain yield responded more to sidedressed N than preplant N: 77 kg of yield was produced for every kg of sidedressed N compared to 66 kg of yield.

This year the experiment was repeated in an adjacent farmer’s field with about 30 acres

Adjustable Sensor Mounting It was the first attempt to carry out the variable N rate application in Ontario and Quebec based on NDVI mapping using an RT200 system

Ottawa Equipment Ford 7610 tractor with modified hydraulic returns Yetter 2995 Bubble Fertilizer Coulters with rear Knife 250 liter tank with filter John Blue positive displacement piston pump with flow divider Rawson Accurate Controller Dickey John Radar Raven GPS GreenSeeker RT200

GreenSeeker Mapping June 26, 2006 REP AVG MAX MIN 1 .57 .78 .32 2 .51 .76 .25 3 .58 .83 .26 4 .66 .85 This is our GreenSeeker data taken 3 days before application. Scanned at 15’ wide, Mapped at 10’ wide NDVI Range below .3 to above .8 when mapped June12,14,26&29

Corn 32” to 36” high on Application Day, Ottawa

Ottawa Application Non-irrigated spring wheat with modified inputs gave a very reasonable application curve On application day the curve could not be manipulated to suite our rates because of high NDVI values. Only choice we had was to use the 16 point stepped (not interpolated) graph In StJean, the non-irrigated spring wheat was manipulated to give a very reasonable curve

Ottawa Rate Accurate controller settings: NDVI US gal/ac .2 .25 .3 .35 .4 .45 .5 .55 9.2 .6 35.9 .65 35.0 .7 33.1 .75 32.1 .8 28.5 .85 21.2 .9 11.0 .95 Accurate controller settings: 9.2 to 36.8 gal /ac was the preset range (30 kg N/ha to 120 kg N/ha) Controller works on percentage change from a mid point. 23 gal/ac as the mid point with 4 % increment giving .92 gal There are 32 increments. Based on the maximum NDVI value in the reference strip, saturation index was calculated; the value was then converted back to NDVI; and N rates were calculated based on N response curve derived from an earlier study with Nmin = 30 and Nmax = 120 kg /ha. The 16 points NDVI-N rates were used for variable rate application as we were unable to receive an updated version of the software. In our Quebec site, the rainfed spring wheat algorithm was tested to produce a reasonable response curve, and thus used.

Summary (2005) Very high grain yield in 2005; yield did not plateau; Yield was more responsive to sidedress (77 kg/kg N) than preplant (66 kg/kg N) fertilizer N; Canopy reflectance (NDVI) differentiate low from high soil N NDVI ranges are narrow, sensitive with crop growth stage Better economic return with sidedressed fertilizer N. But, further research is needed for variable N rate application. Year 2005 was Very high in grain yield; yield did not reach plateau. Yield was more responsive to sidedress (77 kg/kg N) than preplant (66 kg/kg N) fertilizer N. Canopy reflectance (NDVI) differentiated low from high soil N. Better economic return with sidedressed fertilizer N. But, further research is needed for variable N rate application

Summary (2006) First year of variable rate N application based on NDVI saturation index (SI) Canopy NDVI values change rapidly from V6 to V8 (plants 2.5-3 feet tall) and can be quickly saturated (NDVI >0.82). The rainfed spring wheat algorithm provided by the RT-200 GreenSeeker was used in Quebec where NDVIref <0.7 In Ottawa site, differences in canopy reflectance among preplant broadcast fertilizer N rates was confounded with poor seedling growth under high N rates due to low soil pH coupled with cool spring. Although such confounding effect was minimal around or after sidedress, it made difficult interpret the NDVI data. The maximum NDVI value was used to derive SI, and Nmin and Nmax were set at 9.2 and 36.8 gal/ac; variable N rate application was made using the 16 points algorithm. It is expected that a better NDVI-N requirement algorithm will be available or developed at site for use next year.

Acknowledgements Ontario Soil and Crop Improvement Association Lynne Evenson, Doug Balchin, Vivianne Deslauriers