Sensors and Systems
Breaking News
California police adopt WingtraOne drone for wildfire damage assessment
Rating12345LAFAYETTE, California – The Lafayette City Police Department and an...
Missouri Technology Corporation and National Geospatial-Intelligence Agency Announce a Geospatial Focused Corporate Accelerator in St. Louis
Rating12345Missouri Technology Corporation (MTC) and the National Geospatial-Intelligence Agency...
Kratos Acquires Jet Powered Unmanned Aerial System Engineering Leader 5-D Systems
Rating12345SAN DIEGO – Kratos Defense & Security Solutions, Inc....
  • Rating12345

precag_thumb.gifThe growing field of Precision Agriculture has effectively harnessed geotechnology to understand and manage the dynamic flows and cycles within a landscape perspective. In less than 15 years the application has moved from inception to an operational reality on millions of acres. Farmers adopting site-specific technologies discover ways to cut their costs, to use inputs appropriate to the productive capacity of the site, and to optimize their outputs for a safe and stable supply of food and fiber.

The growing field of Precision Agriculture has effectively harnessed geotechnology to understand and manage the dynamic flows and cycles within a landscape perspective. In less than 15 years the application has moved from inception to an operational reality on millions of acres. Farmers adopting site-specific technologies discover ways to cut their costs, to use inputs appropriate to the productive capacity of the site, and to optimize their outputs for a safe and stable supply of food and fiber.

So where is Precision Agriculture headed? The short answer is that it is being extended from a focus on crop production on individual fields to understanding and managing the complex flows and cycles defining stewardship at a landscape scale. The mix of spatial technologies that help increase yields per hectare will be applied to understand and manage agricultural systems and to connect the flows from these systems to natural areas in an effort to manage entire regions for maximum yield and agro-environmental sustainability.

Defining the Movement

Precision Conservation can be defined “as a set of spatial technologies and procedures linked to mapped variables, which is used to implement conservation management practices that take into account spatial and temporal variability across natural and agricultural systems.” Contrary to Precision Farming that is oriented to maximize yields in agricultural fields, Precision Conservation connects farm fields, grasslands, rangelands and managed forests with their natural surrounding areas such as buffers, riparian zones, natural forest, and water bodies (Figure 3). The goal of Precision Conservation is to use information about surface and underground flows and cycles to analyze the systems to make the best decisions about management practices that contribute to conservation of agricultural, rangeland, and natural areas.

The approaches, as well as the objectives, of the two disciples are fundamentally different. Precision Farming solutions characterize a field as a series of 2D data layers that usually summarize the statistical inference of coincidence in a stack of map layers with minimal attention to cause and effect explanation. Precision Conservation, on the other hand, is poised to take the analysis to another level by investigating the interconnections of map features and processes within three-dimensional space, such as leaching and wind erosion. This ecosystem perspective enlarges both the geographic and intellectual scales of the problems that can be addressed. Precision Farming is transitioning to an operational technology; however, Precision Conservation is still very much in its formative stages.

(Figure reproduced with permission from Journal of Soil and Water Conservation)

Figure 3. The site-specific approach in Precision Farming can be expanded to a three-dimensional approach that assesses infows and outflows from individual fields at watershed and regional scales.

While there can be different degrees of Precision Conservation, such as the use of non-digital, non-GIS maps and the use of survey methods, it is generally thought of as technologically-based, requiring the integration of spatial technologies (Global Positioning System, Geographic Information Systems, Remote Sensing) and focused on the ability to analyze spatial relationships within and among mapped data to better understand cause and effect interactions—more of a scientific than a technological endeavor.

The new map analysis capabilities enabled by geotechnology can be grouped into two broad categories: 1) Spatial Statistics involving numerical relationships of surface modeling and spatial data mining, and 2) Spatial Analysis involving geographical relationships, such as proximity and terrain configuration.

Spatial Statistics of Continuous Geographic Space

The new Spatial Statistics techniques contribute to an integrated evaluation of topography, hydrology, weather, management, and other physical and chemical parameters, to generate new insights into site-specific effects for management of flow and cycles that interconnect agricultural and natural systems.

Figure 3 outlines the fundamental differences between the traditional mapping approach and the map analysis approach used in Precision Conservation. Most desktop mapping applications that take a set of spatially collected data (e.g., parts per million, kg/ha, etc.), then reduces the sample set to a single value (total, average, median, etc.) and “paints” a fixed set of polygons with colors reflecting the “typical” condition occurring within each polygon.

(Figure reproduced with permission from Journal of Soil and Water Conservation)

Figure 4. Desktop Mapping uses aggregated, non-spatial statistics to summarize spatial objects (points, lines and polygons), whereas Map Analysis uses continuous spatial statistics to characterize gradients in geographic space (surfaces).

The left side of the figure depicts the position and relative values for a set of field-collected data; the right side shows a derived spatial distribution of the data for an individual reporting parcel. The average of the mapped data is shown as a superimposed plane “floating at average height of 22.0” and assumed the same everywhere within the polygon. But the data values themselves, as well as the derived spatial distribution, suggest that higher values occur in the northeast and lower values in the western portion.

The first thing to notice in the figure is that the average is hardly anywhere, forming just a thin band cutting across the parcel. Most of the mapped data is well above or below the average. That is what the standard deviation attempts to reveal—just how typical the computed typical value really is. If the dispersion statistic is relatively large, then the computed typical isn’t typical at all. However, all non-spatial statistics and most desktop mapping applications ignore data dispersion and simply “paint” a color corresponding to the computed average value regardless of numerical and spatial data distributions within a parcel.

However, the central tendency assumption can be misleading. Assume the data is characterizing a toxic chemical in the soil that, at high levels, poses a serious health risk. The mean values for both the parcel on the left (22.0) and the right (28.2) are well under the “critical limit” of 50.0. Desktop mapping would paint both parcels a comfortable green tone, as their typical values are well below concern. Even considering the upper-tails of the standard deviations, the limit is not exceeded (22.0 + 18.7= 40.7 and 28.2 + 19.8= 48.0). So from a non-spatial statistics perspective, the aggregated results indicate acceptable levels of the chemical in both parcels.

However, the lower-right portion of the figure portrays a radically different set of conditions. The left and right parcels are displayed as an increasing gradient from low levels (green) through areas that are above the critical limit (red). The high regions, when combined, represent a contiguous sub-area of nearly 15% of the combined area that likely extends into other adjacent parcels. The bottom line is that the aggregated, non-spatial treatment of the spatial data fails to uncover the spatial patterns of the data needed for effective management actions. Similar surface modeling investigations can be used to interpolate point data into continuous surface representations of data across the landscape to better explain the variance inherent in field-collected data.

Point density mapping, spatial interpolation, and map generalization are examples of surface modeling techniques in the map analysis toolbox. Point density mapping can be used to evaluate the number of aggregate points within a specified distance (e.g., number of occurrences per hectare). Conservation practitioners and scientist will collect point-sampled data to derive maps of nutrient concentrations such as soil carbon. For example, Kriging for spatial interpolation of weight-average measurements within a localized area could be used to assess carbon sequestration potential. An example of map generalization is the use of polynomial surface fitting to the entire data set to identify dominant directional trends or flows.

There are new techniques for spatial data mining that can be used to try to uncover relationships within and among mapped data layers such as water tables, erosion potential, topography, soil texture, yields, vegetative cover, soil depths, and others. These procedures, including coincidence summary, proximal alignment, statistical tests, percent difference, level slicing, map similarity, and clustering can be used to assess similarities in data patterns.

Yet another type of spatial data mining is the use of predictive models that use crop biomass cover (straw biomass production-dependent variable) and the soil nutrient values (soil texture, soil carbonates, topography, hydrology, water levels, runoff) that can be used as independent variables to “map-ematically” quantify data patterns. As thousands of map locations are analyzed, predictable spatial relationships among crop biomass and the driving variables could be uncovered.

For example, crop residue production might be correlated to potential for reduction of erosion and surface runoff, and other soil and water conservation outcomes. Scientists and practitioners can analyze the numerical relationships of spatial patterns inherent in mapped data using surface modeling and spatial data mining. It is anticipated that these approaches can be used to better explain the complex interrelationships within spatial distributions, thereby strengthening our understanding and decision-making capabilities for sustainable agriculture and natural systems.

Spatial Analysis of Continuous Geographic Space

Spatial analysis operations have a similar potential to integrate site-specific management actions with site-specific conservation practices and also with off-site conservation practices that can contribute to watershed sustainability. Instead of focusing on “numerical” relationships used in spatial statistics, spatial analysis operations focus on “geographic context,” such as proximity and inter-connections.

For example, since water will take the steepest downhill path over a terrain surface, surface flow over an elevation map can be modeled and used in determining an erosion potential map. In a manner analogous to real-world overland flow, spatial analysis can map-ematically “place a drop of water” at a location on an elevation surface and allow it to pick its path down the surface in a series of steepest downhill steps. As each map location is traversed it gets the value of one added to it. As the paths from other locations are considered, the areas sharing common paths get increasing larger values (one + one + one…) that identify more uphill locations contributing overland flow—increased confluence.

An example of a confluence map is shown in the lower portion of figure 5. The elevation surface (vertically exaggerated) shows the terrain configuration and the inset illustrates a slightly depressed location receiving very high accumulated surface flow from 451 uphill locations. The “Flow map” can be draped over showing the locations where all flow is away (gray tone on ridges), areas with greater confluence of water (blue and green tones) and areas of heavy flows where large amounts of water could potentially form large gullies (red areas).

(Figure reproduced with permission from Journal of Soil and Water Conservation)

Figure 5.
Maps of surface flow confluence and slope are calculated by considering relative elevation differences throughout a project area.

Surface flow is just one factor for determining where erosion is likely to occur, and can be extended to simple “erosion potential” by considering the slope. The slope operation in map analysis is based on relative differences in elevation values within the immediate vicinity of a location on an elevation surface (a 3 x 3 window surrounding each grid location). The most commonly used slope algorithm uses a “fitted plane” to the localized elevation values that minimizes the deviations from the plane to the nine individual elevation values—33.23 percent for the example location in the upper inset in the figure.

Characterizing Erosion Potential

The slope map characterizes the relative energy of water flow at a location, while the confluence values on the flow map identify the volume of flow. It is common sense that as energy and volume increases, so does erosion potential. The maps of slope and flow can be combined to develop a simple erosion potential map. While the sequence of processing shown in the top portion of Figure 6 might appear as an unfamiliar way of thinking with maps, the underlying assumptions are quite straightforward.

(Figure reproduced with permission from Journal of Soil and Water Conservation)

Figure 6.
Effective erosion buffers around a stream expand and contract depending on the erosion potential of the intervening terrain as determined by surface slope and flow accumulation.

The first step in the model classifies slope into three relative steepness classes—1= Gentle, 2= Moderate and 3= Steep for the Slope Classes map. The next step does the same thing for relative flow classes—1= Light, 2= Moderate and 3= Heavy for the Flow Classes map. The third step combines the maps of slope and flow classes for a Slope/Flow map that identifies all existing combinations of slope and flow classes. In combining the two maps, the Flow Classes map is multiplied by 10 then added to the Slope Classes map to create a two-digit code where the first digit identifies the flow class and the second digit the slope class.

For example, on the Flow/Slope map, the category “33 Heavy Flow; Steep” (bright red) identifies areas that are relatively steep (Slope class= 3) and have a lot of uphill locations contributing water (Flow class= 3). Loosened soil under these circumstances is easily washed downhill. However, category “12 Light Flow; Moderate” (light green) identifies locations with much less erosion potential. In fact, deposition (the opposite of erosion) occurs in areas of minimal flow and gentle slope; category “11 Light Flow; Gentle” (dark green).

The final step in determining erosion potential interprets the slope/flow combinations for simplified surface transport Erosion Potential map containing a gradient of susceptibility for erosion from 9= low (green) through 1= high (red). Note that the red areas indicating a lot of potential erosion align with the sides of sloping terrain, whereas the green areas indicating little erosion potential are at the flat tops and bottoms of the terrain surface. Of particular concern are red areas near the edge of a field, or other actively disturbed area, where materials can be easily washed off and enter the waterways.

These are good simple precision conservation techniques that can be used to identify potential hot spots for runoff and sediment and agrochemicals transport out of the field. Producers may want to cover these high sensitive edge areas with grasses or buffers along the edge of the fields or use other viable practices.

Identifying Effective Erosion Buffers

Traditionally, protective buffers based on simple geographic distance from a stream are used to protect sensitive waterways from sediment and chemical loading. However, the Erosion Potential map can be used to identify effective erosion buffers around waterways that respect the intervening terrain (bottom portion of Figure 6).

These variable-width buffers reach farther under high erosion conditions and not as far under low erosion conditions. The algorithm for deriving effective distance moves away from the stream as a series of wave fronts, noting the relative erosion potential at each step. An area of high erosion potential causes the wave front, to extend farther in geographic space than an area of low erosion potential. The result is an Erosion Buffer map that constricts and expands as a function of the intervening conditions.

This simplified example does not take into consideration plant density, soil type, drainage, infiltration, saturation, hard pans, soil depth, or other important variables affecting erosion severity. However, it is sufficient to illustrate the basic elements of the GIS modeling approach encapsulated in Figure 6. The flowchart is used to summarize the model’s logic with each map representing a logical step and each line representing an analysis operation. At each step, an analytical operation is employed from GIS’s robust “toolbox” of capabilities for assessing spatial relationships.

Future Directions

The bulk of analysis used in Precision Farming relies on “static coincidence modeling” using a stack of geo-registered map layers, and for the most part, is in place. However, the Precision Conservation frontier is shifting the focus to “dynamic flow modeling” that tracks movement over space and time in three-dimensional geographic space. This quantum leap awaits both technological and scientific advances.

A wholesale revamping of the data structure underlying digital maps is needed which will shake at the very core of geotechnology—the Cartesian coordinate system itself, a spatial referencing concept introduced by mathematician Rene Descartes over 400 years ago.

The current 2D square for geographic referencing is fine for “static coincidence” analysis over relatively small land areas, but woefully lacking for “dynamic 3D flows.” It is likely that Descartes’ 2D squares will be replaced by hexagons (analogous to the pentagon patches forming a soccer ball) that better represent our curved earth’s surface. Current three-dimensional referencing using cubes will be replaced by nesting polyhedrals for a consistent and seamless representation of three-dimensional geographic space. This change in referencing extends the current six-sides of a cube for flow modeling to the twelve-sides (facets) of a polyhedral—radically changing flow and cycle algorithms, as well as our historical perspective of mapping.

The radical changes on the technical front will be matched on the disciplinary sciences underlying our understanding of spatial relationships and sustainability models. The new spatial tools can more precisely identify where to locate riparian buffers, sediment ponds, nutrient management farms and other ecological engineering practices to most effectively reduce environmental impacts from hot spots across the watershed.

These technologies can be used to simultaneously consider variable hydrology and temporal flows to identify the best locations for the implementation of conservation practices at the watershed and sub-watershed levels. These technologies can also be used to design better buffers to manage flows at field borders, to identify the best locations for phosphorous recovery devices, and to locate potential denitrification trap sites.

These new approaches can contribute to better management of variable surface and underground flows across grass waterways, buffers, riparian buffers, ditches, wetlands, and watersheds. New advances even show that there is potential to integrate management of rangeland animal behavior with management practices that account for spatial and temporal variability to enhance conservation of soil and water resources.

With continued increases in population growth and increased demands of land resources for food and biofuel production, maximizing agricultural production is increasingly necessary. Advances in geotechnology can be used to synchronize best management practices that maximize yields while reducing unnecessary inputs and losses of sediment and other chemicals to the environment. As new technological advances continue to emerge, adaptations of Precision Farming and Conservation techniques by land owners, managers, farmers, and extension personnel will be widely implemented.

These new technologies can contribute to higher efficiency of resource management, economical returns, and environmental sustainability. Precision Farming and Precision Conservation will play significant roles in maximizing and sustaining agricultural yields while contributing to global sustainability in the 21st century.

Joseph K. Berry is Keck Scholar in Geosciences, University of Denver, Colorado, Jorge A. Delgado is Soil Scientist, USDA-ARS, Soil Plant Nutrient Research, Fort Collins, Colo., and Rajiv Khosla is GPS/ Assistant Professor, Colorado State University, Fort Collins, Colo.

Editor’s Note:
This is the second installment in a two-part series about the use of precision land management techniques. Read the first installment here .


1) Precision Conservation for Environmental Sustainability , J. of Soil and Water Conservation, Nov/Dec 2003, 58:6 332-339. J.K. Berry, J. A. Delgado, R. Khosla and F.J. Pierce.

2) Applying Spatial Analysis for Precision Conservation Across the Landscape , J. of Soil and Water Conservation, Nov/Dec 2005, 60:6 22-29. J.K. Berry, J. A. Delgado, R. Khosla and F.J. Pierce.

3) Advances in Precision Conservation , Journal of Advances in Agronomy, Winter 2008, accepted-in press. J.A. Delgado and J.K. Berry.

Leave a Reply

Your email address will not be published. Required fields are marked *