9th International Conference on Precision Agriculture

July 20-23, 2008 Denver, Colorado

So Where Is Precision Ag?

a brief history, current expression and future directions

Joseph K. Berry
W. M. Keck Visiting Scholar in Geosciences, Geography, University of Denver Principal, Berry & Associates // Spatial Information Systems Email jberry@innovativegis.com Web www.innovativegis.com/basis/

What Is Precision Agriculture?

Things to keep in mind
PA is about doing the right thing at the right place and at the right time it identifies and responds to the variability within a field it augments indigenous knowledge
(not a replacement)

(PA has been around awhile, Circa 1992)

it is a radically different technology with extremely high expectations

(Berry)

Historical Setting and Evolution

Computer Mapping
automates the cartographic process (70s) links
GIS
8,000 years of mapping

Spatial Database Management

computer mapping techniques with traditional database capabilities (80s)

CM + SDBM of the first two decades is often referred to as Desktop Mapping and Inventory What is Where Map Analysis Why and So What

Map Analysis

representation of
focus of this presentation

relationships within and among mapped data (90s)

Toolbox supporting Precision Ag

Multimedia Mapping

full integration of GIS, GPS, RS, Internet and visualization technologies (00s)
Note: U.S. Dept. of Labor identifies Geotechnology (GPS, GIS, RS) as one of three "mega technologies" for the 21st century and promises to change how we conceptualize, utilize and visualize spatial relationships in scientific research and commercial applications
(the other two are Biotechnology and Nanotechnology) (Berry)

Yield Limiting Factors (the basis of PA)

Water Weather Topography Nutrients Weeds Pests Genetics Seeding Rate Other

On-Farm Studies (Research?)

Candidate for Precision Agriculture and Site-specific Management if and only if the factor is a significant driving variable it has measurable spatial variability its variability can be explained and spatial relationships established it exhibits a spatial response to practical management actions
and results in production gains, increased profitability and/or improved stewardship
(Berry)

Whole Field vs. Site Specific Management

Whole-field assumes the average conditions are the
same everywhere within the field (uniform/homogenous)
Management action is the same throughout the field

Z2

Discrete Management Zones

break the field into areas of similar conditions (zones)

The bulk of agricultural research has been non-spatial but PA is all about spatial relationships/patterns Research Opportunity

Z1

Z3 Z2

Z1

Management action is the same within each zone

Is Smart Sampling really dumb?

Continuous Surfaces break the field into small

consistent pieces (cells) that track specific conditions at each location
Management action varies throughout the field
(Berry)

MAP Analysis Framework (Keystone Concept)

Click on

Shading Manager

Zoom Pan Rotate

Display

Analysis Frame (Grid )

each map layer is organized as a geo-registered matrix of numbers

Map Stack

Grid Table

Continuous regular grid cells (objects)

: --, --, --, --, --, --, --, --, --, --, --, --, --, 149.0, --, --, --, --, --, :
(Berry)

Surface Modeling Techniques

Grid-Based Map Analysis (workshop topics)
Surface Modeling maps the spatial distribution and pattern of point data
Map Generalization characterizes spatial trends (tilted plane) Spatial Interpolation deriving spatial distributions (e.g. IDW, Krig) Other roving windows and facets (e.g., density surface; tessellation)

Spatial Data Mining investigates the numerical relationships in mapped data

Descriptive aggregate statistics (e.g. average, stdev, similarity; clustering) Predictive relationships among maps (e.g., regression) Prescription appropriate actions (e.g., decision rules; optimization)

Spatial Analysis investigates the contextual relationships in mapped data

Reclassify reassigns map values (position, value, size, shape, contiguity) Overlay map coincidence (point-by-point; region-wide; map-wide) Distance proximity and connection (movement; optimal paths; visibility) Neighbors roving windows (slope; aspect; diversity; anomaly)
(Berry)

Geographic Distribution (Mapping the Variance)

The iterative smoothing process is similar to slapping a big chunk of modelers clay over the data spikes, then taking a knife and cutting away the excess to leave a continuous surface that encapsulates the peaks and valleys implied in the original field samples

Numeric Distribution Average, Standard Deviation

Continuous Surface Geographic Distribution

(Berry)

Spatial Interpolation (soil nutrient levels)

Spatial Interpolation maps the geographic distribution inherent in the data
Corn Field Phosphorous (P)

Data Spikes

IDW Surface

(Berry)

Comparing Spatial Interpolation Results

Comparison of the IDW interpolated surface to the whole field average shows large differences in localized estimates
(-16.6 to 80.4 ppm)

Comparison of the IDW interpolated surface to the Krig interpolated surface shows small differences in localized estimates
(-13.3 to 11.7 ppm)

(Berry)

Spatial Data Mining Techniques

Grid-Based Map Analysis (workshop topics)
Surface Modeling maps the spatial distribution and pattern of point data
Map Generalization characterizes spatial trends (tilted plane) Spatial Interpolation deriving spatial distributions (e.g. IDW, Krig) Other roving windows and facets (e.g., density surface; tessellation)

Spatial Data Mining investigates the numerical relationships in mapped data

Descriptive aggregate statistics (e.g. average, stdev, similarity; clustering) Predictive relationships among maps (e.g., regression) Prescription appropriate actions (e.g., decision rules; optimization)

Spatial Analysis investigates the contextual relationships in mapped data

Reclassify reassigns map values (position, value, size, shape, contiguity) Overlay map coincidence (point-by-point; region-wide; map-wide) Distance proximity and connection (movement; optimal paths; visibility) Neighbors roving windows (slope; aspect; diversity; anomaly)
(Berry)

Visualizing Spatial Relationships

Interpolated Spatial Distribution Phosphorous (P)

What spatial relationships do you see?

do relatively high levels of P often occur with high levels of K and N? how often? where?
HUMANS can see broad generalized patterns in a single map variable
(Berry)

Clustering Maps for Data Zones

COMPUTERS can see detailed patterns in multiple map variables

groups of floating balls in data space identify locations in the field with similar data patterns data zones
(Berry)

The Precision Ag Process (Fertility example)

Steps 13)

As a combine moves through a field 1) it uses GPS to check its location then 2) checks the yield at that location to 3) create a continuous map of the yield variation every few feet (dependent map variable).
Prescription Map
As-applied maps

On-the-Fly Yield Map

Zone 3

Intelligent Implements Step 5)

Step 4)
Derived Nutrient Maps
Zone 2

Zone 1

Variable Rate Application

The yield map 4) is analyzed in combination with soil, terrain and other maps (independent map variables) to derive a Prescription Map 5) that is used to adjust fertilization levels every few feet in the field (action).
more generally termed the

Spatial Data Mining Process (e.g., Geo-Business application)

(Berry)

Data Analysis Perspectives

(Data vs. Geographic Space)

Traditional Analysis
(Data Space Non-spatial Statistics)
Field Data

Map Analysis
(Geographic Space Spatial Statistics)

Standard Normal Curve fit to the data

Spatially Interpolated data

Central Tendency
Typical How Typical Average = 22.0 StDev = 18.7

22.0

28.2

Discrete Spatial Object

(Generalized)

Continuous Spatial Distribution

(Detailed)

Identifies the Central Tendency

Maps the Variance

(Berry)

So Where Are We in Precision Ag?

Yield Mapping
done deal for many crops procedures need validation

Soil Nutrient Mapping

Mgt Zone Mapping

alternative approaches need study & validation

The Full Precision Farming Process

IF <condition> THEN <action>

a fair piece to go

based on spatial relationships

Description (Where is What) coming on line Prediction (Why and So What) needs lots of work PA Nugget Prescription (Do What Where) barely on the research radar Action (Precisely Here) done deal for many farm inputs
(Berry)

Spatial Analysis Techniques

Grid-Based Map Analysis (workshop topics)
Surface Modeling maps the spatial distribution and pattern of point data
Map Generalization characterizes spatial trends (tilted plane) Spatial Interpolation deriving spatial distributions (e.g. IDW, Krig) Other roving windows and facets (e.g., density surface; tessellation)

Spatial Data Mining investigates the numerical relationships in mapped data

Descriptive aggregate statistics (e.g. average, stdev, similarity; clustering) Predictive relationships among maps (e.g., regression) Prescription appropriate actions (e.g., decision rules; optimization)

Spatial Analysis investigates the contextual relationships in mapped data

Reclassify reassigns map values (position, value, size, shape, contiguity) Overlay map coincidence (point-by-point; region-wide; map-wide) Distance proximity and connection (movement; optimal paths; visibility) Neighbors roving windows (slope; aspect; diversity; anomaly)
(Berry)

Micro Terrain Analysis

(a simple erosion model)

Determining Erosion Potential: slope and flow classes are

combined into a single map identifying erosion potential

times 10 plus

renumber

Field Elevation is
formed by assigning an elevation value to each cell in an analysis grid
(1cm Lidar)

(Berry)

Precision Conservation (compared to Precision Ag)

Precision Conservation
(Farm, Watershed, Focus) Wind Erosion

Precision Ag
(Individual Field Focus)

Chemicals

Soil Erosion Runoff

Terrain

Leaching
Leaching Leaching
Soils Yield Potassium

3-dimensional

Interconnected Perspective
(Stewardship Focus)

2-dimensional

CIR Image

Isolated Perspective
(Production Focus)
(Berry)

Deriving Erosion Potential (regional scale)

Maps of surface flow confluence and slope are calculated by considering relative elevation differences throughout a project area

(Berry)

Calculating Effective Distance (variable-width buffers)

Effective erosion buffers around a stream expand and contract depending on the erosion potential of the intervening terrain

(Berry)

Cyclical Development (future directions)

Revisit Analytics

Future Directions

(2020s)

2D Planar
(X,Y Data)
(2000s)

3D Solid
(X,Y,Z Data)

Multimedia Mapping

Revisit Geo-reference
(2010s)
Square (4 sides) Hexahedron (6 squares)

Contemporary GIS Spatial dB Mgt (1980s) Map Analysis

(1990s)

The Early Years

Mapping focus Data/Structure focus Analysis focus

Hexagon (6 sides)

Dodecahedron (12 pentagons)

Computer Mapping
(1970s)

Is GIS Technology Ahead of Science?

Five critical questions underlying Precision Agriculture

Is the scientific method relevant in the datarich age of knowledge engineering?

1)

Is the random thing pertinent in deriving mapped data?

2)

Are geographic distributions a natural extension of numerical distributions?

3)

Can spatial dependencies within a map variable (spatial autocorrelation) and among map variables (spatial correlation) be modeled?
4)

How can site-specific analysis and on-farm studies contribute to the scientific body of knowledge?
5)
(Berry)

Where To Go From Here

www.innovativegis.com/basis/

Online References Analyzing Precision Ag Data

workbook with hands-on exercises

Textbook Presentation handout Workbook