Contents

Search (over 200 pages above & below)

Crop Agriculture

Field Data Bundle (slideshow; details & graphics below)   

Yield Monitor Data  •  Landsat Imagery  •  LiDAR-Based Elevation

Field Data Bundle Details

A Field Data Bundle includes Yield Monitor Data, Landsat Imagery, and LiDAR-elevation based layers (in formats shown on this page) organized in a single folder (vector and raster data). If you have yield maps, they can be cleaned and, if necessary, post-calibrated. A single master point shapefile will include all yield maps, Landsat data, as well as elevation values for each point and can include other topographic values such as curvature. Landsat can be calibrated with yield in different ways if requested. Other data (such as electrical conductivity readings) can be included if sent. Data will be in the correct geographic coordinate system for farm software and can be customized in a way that meets particular needs. Contact GIS Ag Maps if you have question about or are interested in purchasing layers.

 

Yield mapping; 1 - 12; Landsat: 13 - 27; LiDAR: 28 - 43; Yield - LiDAR: 44 - 48; Yield - Landsat: 49 - 55

Yield Monitor Data and Mapping (1 - 12)

Yield maps derived from yield monitor data have useful information but also inherent errors. It is well-documented that these errors should be cleaned prior to use. Information about yield monitor data is included in the Yield Map Cleaning & Mgmt. folder. Also, it is common that yield monitors have not been properly calibrated; linear or non-linear post-calibration can be applied to improve the data and can include the field average, minimum and maximum values, and standard deviation. See the Yield Map Post-Calibration folder within the Yield Yield Map Cleaning & Mgmt. folder for more information.

Raw corn yield map with many errors that need to be cleaned. Classification for all maps unless otherwise specified is with natural breaks. Natural breaks classifies based on similar groups of vales and is one of the correct methods of classifying yield maps. It is incorrect to classify with methods such as equal interval or quantiles. Colors represent, from highest to lowest yield:dark green, green, yellow, orange, and red. (Background imagery for all graphics is from OGRIP [2013].)

Raw corn yield map

 

Map after many cleaning steps; see Corn & Soybean Yield Map Cleaning Examples page for more details.

Filtered corn yield map

 

Clean corn yield map.  Point spacing is 4-meters; symbolized with squares to show 4-meter resolution.

Clean corn yield monitor data map

 

Clean polygon corn yield map

Clean corn yield monitor data polygon map

 

Clean one-meter resolution raster corn yield map (darker green is higher yield; scaled from minimum to maximum value). 

Clean one-meter resoluton raster corn yield map

 

Raw soybean yield map with many errors that need to be cleaned. 

Raw soybean yield map with many errors that need to be cleaned

 

Map after many cleaning steps; see Corn & Soybean Yield Map Cleaning Examples page for more details.

Filtered soybean yield monitor data map

 

Clean soybean yield map. Points are processed to be at the same location as the corn yield map (as shown more in the graphics below the raster map) and are represented as squares to show 4-meter spacing resolution. The clean corn and soybean yield maps are in one shapefile. Yield maps for all seasons for any crop, whether for the whole field or part of the field, should be represented by the same point shapefile; each has its own column of yield amounts in the attribute table so values can be compared and analyzed at the same location over time.

Clean soybean yield map

 

Clean polygon soybean yield map.

Clean corn yield monitor data polygon map

 

Clean one-meter resolution raster soybean yield map (darker green is higher yield; scaled from minimum to maximum value); pixels are at same location as raster corn map above. 

Clean one-meter resoluton raster soybean yield map

 

Yield map points from above are zoomed into. Though the raw yield map points are not at the same location, processing in GIS can result in yield values at the same location for different seasons by joining points from close distances (within the horizontal accuracy error of the data) to the points from the first clean yield map. The shapefile attribute table can be accessed and analyzed using common spreadsheet software to see how yield correlates to each other and to other factor such as elevation and curvature (concave to flat to convex areas).

Corn 

Corn yield map points zoomed into

 

Soybeans 

Soybean yield map points from above are zoomed into

 

Because yield values are at same location, comparisons can be made. The R² correlation between the corn and soybean yield maps shown above is .6651 (highly significant; 7,668 points). (See the About R and R² Correlation page for information about correlation values.) 

 

 Landsat Imagery (13 - 27)

In regards to crop field applications, if applied correctly Landsat imagery can essentially be viewed as yield data with a 30-meter resolution instead of the 4-meter yield map resolution. The Landsat data used needs to be correct for the specific crops to achieve high correlation with yield. Information about applying Landsat for crop field purposes is in the Crop Imaging Background, Landsat Information, and Landsat Correlation to Crop Yield folders. There are many different levels of processing of Landsat that affect the cost; but remember that Landsat is free so you may be able to get all the use out of it you want be accessing the imagery and using it yourself. An advantage of Landsat is that it does not have to be cleaned; a disadvantage is that is does not extend to the field boundary as pixels along the boundary average in values from outside the field and should not be used. The 30-meter resolution may seem large but the size is reasonable for generalizing patterns throughout a typical Corn Belt-size field. The field below has an electrical tower in the center; a decision as to whether pixels should be used if they represent a non-crop feature needs to be made on a field-by-field basis. In this case the pixels that include the tower can be used.

Landsat processing and correlation to yield examples

Landsat near infrared (NIR) pixels within a field that correlate highly to yield (levels as shown in this website) can be applied to generalized crop condition at the time of imagery and future yield patterns. Landsat can be calibrated to soybean of corn yield.  

Landsat near infrared (NIR) pixels within a field that correlate highly to yield

 

Landsat can be extended to field boundary based on interpolation or resampling for a more coherent map.

Landsat can be extended to field boundary based on interpolation or resampling for a more coherent map

 

Landsat-derived values can be joined from close distances with the yield point maps so there is a corresponding value in the attribute table and a correlation can be made.

Landsat-derived values can be joined from close distances with the yield point maps

 

Corresponding clean soybean yield map for same season; because Landsat and yield values are at same locations, correlations can be made: R² = .7580 (highly significant; 7,668 points [same yield map as shown in yield cleaning example above]). When the data is processed as shown here, Landsat NIR explains 75.80% of soybean yield variability.

Corresponding clean soybean yield map for same season; because Landsat and yield values are at same locations, correlations can be made: R² = .7580 (highly significant; 7,668 points [same yield map as shown in yield cleaning example above]).

 

Example of Landast NIR correlation with soybean yield for another season

Landsat near infrared (NIR) pixels within a field

Landsat near infrared (NIR) pixels within a field

 

Landsat to field boundary extent based on interpolation from centroids

Landsat to field boundary extent based on interpolation from centroids

 

Landsat-derived values at yield points locations (from above)

Landsat-derived values at yield points locations

 

Corresponding clean soybean yield map for same season; because Landsat and yield values are at same locations, correlations can be made: R² = .7003 (highly significant; 7,668 points). When the data is processed as shown here, Landsat NIR explains 70.03% of soybean yield variability.

Corresponding clean soybean yield map for same season; because Landsat and yield values are at same locations, correlations can be made: R² = .7003 (highly significant; 7,668 points). When the data is processed as shown here, Landsat NIR explains 70.03% of soybean yield variability.

 

The simplest level of Landsat usage is to open data that correlates to the specific crop well and overlay a field boundary shapefile. Landsat imagery and GIS software to view and process data are free and can be accessed in the Free section to the left. An historic assessment of field yield patterns can be assessed by viewing proper imagery for past seasons. For all graphics below, lighter shades represent higher values and are scaled at ± 2 standard deviation from the mean (an entire scene was opened and the field boundary was overlayed).

Landsat 8 (launched in February of 2013); 8/23/13, NIR (band 5); soybeans. NIR solely should be used for correlations with soybean yield. See details in the Correlation to Soybean Yield page for a correlation comparison for NIR and indices.

Landsat 8 (launched in February of 2013); 8/23/13, NIR (band 5); soybeans

 

Landsat 5; 8/9/11, NIR (band 4); soybeans (there was no correct imagery for corn in 2012)

Landsat 5; 8/9/11, NIR (band 4); soybeans

 

Landsat 5; 6/19/10 NNIR; corn. NNIR is less common than NDVI but is the preferred index; although correlation results between NNIR and NDVI are very similar (NDVI follows for comparison purposed).  See details in the Correlation to Corn Yield page for index correlation comparisons. NNIR includes the green band in addition to the red and NIR bands (the higher the crop condition the lower all visible bands are, including the green) and in written as: (NIR / [NIR + red + green]).

Landsat 5; 6/19/10 NNIR; corn

 

Landsat 5; 6/19/10 NDVI ([NIR - red] / [NIR + red]); corn.

Landsat 5; 6/19/10 NDVI ([NIR - red] / [NIR + red]); corn

 Landsat 5; 8/3/09, NIR (band 4); soybeans

Landsat 5; 8/3/09, NIR (band 4); soybeans

 

Landsat 5; 7/15/08 NNIR; corn

Landsat 5; 7/15/08 NNIR; corn

Landsat 5; 7/15/08 NDVI; corn

Landsat 5; 7/15/08 NDVI; corn

 

LiDAR-Based Elevation and Layers (28 - 43)

LiDAR-based elevation rasters can be acquired for free from different sources. Some states supply free, publicly available data; in many cases, state data becomes available on the USGS National Map Viewer. The USGS further processes and smoothes state data prior to making it available; the National map LiDAR-based data has an approximate 3-meter resolution upon download (data at the state level commonly has a finer resolution of about 1 - 2 meters). See the LiDAR-Based Elevation and Layers folder for details about LiDAR including what areas it is available for. If LiDAR is not available for a particular area, other elevation data (such as RTK data) can be used to develop the layers shown below.

The simplest way of applying the data is to open it in free GIS and overlay a field boundary. Depending on the source of the data, it can be have different spatial characteristics. Below, data is shown for the state of Ohio (OGRIP, 2013); and has a 2.5 foot resolution (the raster was derived from approximately 2-meter post spacing LiDAR points).  

LiDAR-Based Elevation and Derived Layers 

USGS National Map Viewer (USGS, 2013) LiDAR-based elevation raster for the same extent

 USGS National Map Viewer (USGS, 2013) LiDAR-based elevation raster for the same extent

 

Elevation data can be extracted to the extent of a field; below OGRIP data is shown.

Elevation data can be extracted to the extent of a field 

Elevation contours can be produced from rasters; OGRIP half-foot contours shown below.

Elevation contours can be produced from rasters

 

National Map data has been smoothed which evident in the more coherent contours; though the pixel size of the rasters increases from 2.5 feet to about 3 meters. 

National Map data has been smoothed which evident in the more coherent contours; though the pixel size of the rasters increases from 2.5 feet to about 3 meters.

 

Though the contours appear different, National Map and OGRIP elevation are nearly the same.  For the correlation below, the closest OGRIP elevation point was joined to the National Map (the average distance of a point was about one foot). National Map data is on the x-axis and OGRIP data is on the y-axis.  The average absolute error from the regression line is 0.44 inches (18,168 points).

Though the contours appear different, National Map and OGRIP elevation are nearly the same.

 

 OGRIP elevation to field extent (same as above)

OGRIP elevation to field extent

 Main flow accumulation lines from above elevation

Main flow accumulation lines from above elevation

 

Elevation data can be further smoothed, as shown below, to produce layers 

Elevation data can be further smoothed, as shown below, to produce layers

 

Main sinks (areas where ponding can occur) from the further smoothed data

Main sinks (areas where ponding can occur) from the further smoothed data

 

Sinks with hollow symbology

Sinks with hollow symbology

 

Main basin divides (produced from original [non-smoothed] data)

Main basin divides (produced from original [non-smoothed] data)

 

Curvature from further smoothed 2.5 foot raster (darker is more concave; lighter is more convex)

Curvature from further smoothed 2.5 foot raster (darker is more concave; lighter is more convex)

 

Relatively High Convex Areas

Relatively High Convex Areas

 

Curvature raster with relatively high convex areas, flow accumulation, sinks, and basin boundaries

Curvature raster with relatively high convex areas, flow accumulation, sinks, and basin boundaries

 

Imagery with relatively high convex areas, flow accumulation, sinks, and basin boundaries 

Imagery with relatively high convex areas, flow accumulation, sinks, and basin boundaries

 

Yield Maps with LiDAR Layers (44 - 48)

Elevation-Based Layers can be Overlayed onto Yield and Landsat Maps

Corn yield from yield cleaning example near top of page

Elevation-Based Layers can be Overlayed onto Yield and Landsat Maps (Corn yield from yield cleaning example near top of page)

Soybean yield from yield cleaning example near top of page

Soybean yield from yield cleaning example near top of page

Soybean yield from correlation example

Soybean yield from correlation example

 

Landsat NIR from first correlation to yield example

Landsat NIR from first correlation to yield example

 

Landsat NIR from second correlation to yield example

Landsat NIR from second correlation to yield example

 

Landsat Maps with LiDAR Layers (49 - 55)

As shown above, the simplest level of Landsat usage is to open data that correlates to the specific crop well and overlay a field boundary shapefile. The elevation layers can be overlayed onto to raw Landsat and be applied, in part, as a drainage assessment. For all graphics below, lighter shades represent higher values and are scaled at ± 2 standard deviation from the mean (an entire scene was opened and the field boundary was overlayed). (Same Landsat raw imagery as shown above in this page; see above for more details.)

 

Landsat 8 (launched in February of 2013); 8/23/13, NIR soybeans

Landsat 8 (launched in February of 2013); 8/23/13, with LiDAR elevation layers over NIR soybeans

 

Landsat 5; 8/9/11, NIR; soybeans (there was no correct imagery for corn in 2010) 

Landsat 8; 8/9/11, with LiDAR elevation layers over NIR soybeans

 

Landsat 5; 6/19/10 NNIR; corn 

Landsat 8; 8/9/11, with LiDAR elevation layers over NIR soybeans

 

Landsat 5; 6/19/10 NDVI; corn 

Landsat 8; 6/19/10 with LiDAR elevation layers over NIR soybeans

 

Landsat 5; 8/3/09, NIR; soybeans

Landsat 5; 8/9/11, NIR; soybeans with LiDAR elevation layers

 

Landsat 5; 7/15/08 NNIR; corn 

Landsat 5; 6/19/10 NNIR; corn with LiDAR elevation layers

 

Landsat 5; 7/15/08 NDVI; corn 

Landsat 5; 6/19/10 NDVI; corn with LiDAR layers

 

References

OGRIP. 2013. Ohio Geographically Referenced Information Program. Last updated: 2013. Cited at: http://ogrip.oit.ohio.gov/. 

USGS. 2013. The National Map Viewer. United States Geological Survey. Last updated: 2013. Cited at: http://viewer.nationalmap.gov/viewer/