N Management in Potato Production

David Mulla, Carl Rosen, Tyler Nigon and Brian

Bohman

Dept. Soil, Water & Climate

University of Minnesota

Topics

Background and conventional nitrogen management

Evaluate the use of remote sensing to predict N needs

using a nitrogen sufficiency index

Examine the ability of hyperspectral imagery to detect N

stress in potato

Identify the best indices associated with leaf N status

Background

Potatoes have a high N requirement and

shallow root system

N is the most limiting nutrient for potato growth

Fertilizer N is essential to optimize yield, but

management can be challenging in the

Midwest with unpredictable rainfall

N rate is important but timing also plays a

critical role – especially on sandy soils

Conventional N management

Depends on variety and market type

Long season varieties like Russet Burbank

respond to split applications

Planting (10-20% of N)

Emergence/hilling (50-60 % of N)

Fertigation (30-40% of N)

Fertigation timing is often based on petiole

nitrate analysis

Conventional N management

Petioles collected on a 7 to 10 day schedule from tuber

initiation through bulking

If petiole nitrate falls below a certain level, additional N is

applied

Approach is simple, but does not account for spatial

variability

Remote sensing better suited for precision agriculture

and variable rate N applications

Objectives

1. To utilize remote sensing to determine the need for in-season variable rate N-fertilizer applications

2. To assess agronomic outcomes from management using adaptive-N rates

SPAD Meter Cropscan Meter

Methods

Sand Plains Research Farm

Becker, MN

Hubbard Loamy sand

Russet Burbank variety

Split-Plot with 4 replicates in

RCBD

Nitrogen is split plot factor

Nitrogen Treatments

2016 22 Apr 1 June 23 Jun 14 Jul 21 Jul 27 Jul

2017 29 Apr 30 May 28 Jun 10 Jul 20 Jul 27 Jul

Plant. Emerge. --------- Post-Emergence -------- Total

----------------------------- lb N ac-1 -------------------------------

1 Control 40 DAP - - - - - 40

2 160 Split 40 DAP 60 Urea 15 UAN 15 UAN 15 UAN 15UAN 160

3 160 CR 40 DAP 120 ESN - - - - 160

4 241 Split 40 DAP 120 Urea 20 UAN 20 UAN 20 UAN 20 UAN 241

5 241 CR 40 DAP 241 ESN - - - - 241

6 VR Split 40 DAP 120 Urea ? ? ? ? ?

Spectral Indices

• Canopy reflectance is affected by water,

chlorophyll, canopy density and age, soil, etc

• Leaf water potential and leaf temperature

– 1981 Crop Water Stress Index (Jackson)

• Leaf Area Index

– 1974 NDVI (Rouse, 670 & 800 nm)

NDVI = (NIR-R)/(NIR+R)

Newer Spectral Indices

Remote Sensing of N Stress

Remote Sensing + Var. Rate N

Nitrogen Sufficiency Index [NSI]

NSI =Variable N treatment

Well Fertilized Reference

CROPSCAN

Multispectral

Radiometer

(16 Narrow Bands)

MTCI =R 751 nm– R 713 nm R 713 nm–R 676 nm

MERIS Terrestrial Chlorophyll Index [MTCI]

751 nm (Near-IR), 713 nm (Red-Edge), 676 nm (Red)

If NSI < 95%, then 20 lb N/ac applied as UAN

Measurements collected every 1-2 weeks

Results

1. Remote sensing and variable rate

nitrogen

2. Agronomic outcomes

23 Jun 14 Jul 21 Jul 27 Jul Total

------------------- lb N ac-1 -------------------

Control - - - - 40

241 Split 20 UAN 20 UAN 20 UAN 20 UAN 241

241 CR - - - - 241

VR Split - 20 UAN 20 UAN 20 UAN 220

2016

± 5% NSI

28 Jun 10 Jul 20 Jul 27 Jul Total

------------------- lb N ac-1 -------------------

Control - - - - 40

241 Split 20 UAN 20 UAN 20 UAN 20 UAN 241

241 CR - - - - 241

VR Split - 20 UAN - 20 UAN 200

2017

± 5% NSI

Results

1. Remote sensing and variable rate

nitrogen

2. Agronomic outcomes

Marketable YieldContrasts

Control ***

Rate **

Source –

Var. Rate –(Note: 70 Mg ha-1 = 625 cwt ac-1)

Impact on Quality

ESN VRN

Hyperspectral Remote Sensing for N

Management in Potato

Tyler Nigon, Carl Rosen and David Mulla

Department of Soil, Water, and Climate

University of Minnesota

Remote Sensing Platforms

Improving Spatial Resolution

Types of Remote Sensing

• Panchromatic reflectance– An average over all wavelengths

• Broad band or multispectral reflectance– Reflectance at a few specific discrete wavelengths

– B, G, R NIR portions of spectrum

• Hyperspectral reflectance– Reflectance at specific narrow band discrete wavelengths across

a large continuous spectral range

• Thermal emission at NIR and MIR wavelengths

Panchromatic Image

Thermal Infrared Imagery

Variable Irrigation via Thermal Imaging

Hyperspectral Imagery Collection

Hyperspectral Data Cube (RGB)

Hyperspectral Remote Sensing• Reflectance at specific narrow band discrete

wavelengths across a large continuous spectral

range

Derivative Spectra• The derivative of hyperspectral reflectance

data indicates portions of the spectrum

where the slope of the reflectance curve

changes rapidly

Lambda-Lambda Plots• Calculate the r2

coeff. for leaf N content at all hyperspectral reflectance bands

• Graph r2 coefficient for all possible combinations of band 1 on the x-axis and band 2 on the y-axis

• Look for band combinations with low redundancy

Best Reflectance Wavelengths?

• The greatest information about plant characteristics

with multiple narrow bands includes the longer red

wavelengths (650-700 nm), shorter green

wavelengths (500-550 nm), red-edge (720 nm), and

NIR (900-940 nm and 982 nm) spectral bands

• The information in these bands is only available in

narrow increments of 10-20 nm, and is easily

obscured in broad multispectral bands that are

available with older satellites

Correlation Between Reflectance and

Total N Concentration in Potato Leaf

Hyperspectral Data Cube (SR8)

SR8 = (R860/(R550*R780)

Potato Hyperspectral Imagery (SR8 = (R860/(R550*R780))

vs NDVI (NIR-R)/(NIR+R)

Russet Burbank

Alpine Russet

NDVI

SR8

Conclusions

• Spatial resolution of aerial and satellite remote sensing

imagery has improved from 100’s of m to sub-meter accuracy

• Spectral bandwidth has decreased with the advent of

hyperspectral remote sensing

• Return frequency of satellite remote sensing imagery has

improved dramatically

• A variety of useful spectral indices now exist for various

precision agriculture applications in potatoes

Conclusions

• Variable-rate nitrogen application reduced total N-

application by 20-40 lb N ac-1 relative to the

recommended rate of 241 lb N ac-1, with a significant

improvement in yield and no effect on quality

• Urea produced the highest total yield, while ESN had the

lowest ratio of misshapen tubers

• Remote sensing of NSI based on MTCI or SR8 spectral

indices is an effective strategy to determine variable N-

rate without impacting tuber quality

Thanks! Funding from:

David Mulla

mulla003@umn.edu

(612) 625-6721