MANAGING VARIABLE RATE IRRIGATION WITH A HYBRID REMOTE SENSING

EVAPOTRANSPIRATION MODELJ. Burdette Barker1, Christopher M. U. Neale2, Derek M. Heeren1

1Dept. of Biological Systems Engineering, University of Nebraska-Lincoln, 2Daughetry Water for Food Institute, University of Nebraska

A combined energy balance and water balance remote sensingevapotranspiration (ET) model is being evaluated for use in developingvariable rate irrigation (VRI) prescriptions for center pivots. The modelwas tested for four maize and soybean fields in eastern Nebraska using2013 Landsat 8 satellite imagery. Modeled ET compared well with eddycovariance measurements for summer images, but poorly for fallimages. The water balance ET compared well with the energy balanceET for irrigated fields, but not for dryland. Image frequency is thoughtto be a contributing factor. In 2015 the model will be used to generateirrigation prescription maps for a VRI system located in one of the studyfields. Unmanned aircraft aerial imagery will be used as the modelinput to improve image frequency. Water use (calculated from neutronprobe readings) and yield under this VRI management technique will becompared with those under traditional irrigation management.

We thank Dr. Andy Suyker of the University of Nebraska-Lincoln, who provided eddy covariance flux data; Dr. Hatim Geli and Mr. Clayton Lewis of Utah State, who provided computer code for the SETMI model; and Mr. Mark Schroder, ARDC director, who provided crop management records. Weather data was provided by the High Plains Regional Climate Center. We also appreciate financial support from the Daugherty Water for Food Institute and the University of Nebraska-Lincoln.

Allen, R. G., L. S. Pereira, D. Raes, and M. Smith. 1998. Crop evapotranspiration: Guidelines for computing crop water requirements, Irrigation and Drainage Paper 56. In Irrigation and Drainage Papers. Rome, Italy: United Nations FAO.

Neale, C. M. U., W. C. Bausch, and D. F. Heermann. 1989. "Development of Reflectance-Based Crop Coefficients for Corn." Transactions of the ASAE 32 (6):1891-1899.

Neale, C. M. U., H. M. E. Geli, W. P. Kustas, J. G. Alfieri, P. H. Gowda, S. R. Evett, J. H. Prueger, L. E. Hipps, W. P. Dulaney, J. L. Chavez, A. N. French, and T. A. Howell. 2012. "Soil water content estimation using a remote sensing based hybrid evapotranspiration modeling approach." Advances in Water Resources 50:152-161. doi: 10.1016/j.advwatres.2012.10.008.

Norman, J. M., W. P. Kustas, and K. S. Humes. 1995. "A two-source approach for estimating soil and vegetation energy fluxes in observations of directional radiometric surface-temperature." Agricultural and Forest Meteorology 77 (3-4):263-293. doi: 10.1016/0168-1923(95)02265-y.

ABSTRACT PRELIMINARY RESULTS

BACKGROUND AND MODEL DESCRIPTION

FUTURE RESEARCH SUMMARY

REFERENCES

ACKNOWLEDGEMENTS

A hybrid remote sensing ET and water balance model is being testedfor use in variable rate, or site-specific irrigation management.Preliminary results using satellite imagery compared well withmeasurements for the mid summer image date, but poorly for threeother dates. Inaccuracies in modeled relationships early and late in theseason, errors in imagery atmospheric correction, and the lowfrequency of cloud-free images are thought to be causes of the poorperformance. Future research will focus on using aerial imagery andsatellite imagery as inputs to improve the frequency of images. Themodel will be used to generate VRI prescriptions for a research fieldnear Mead, Neb. in the 2015 and 2016 growing seasons.

Figure 7. Two-source energy balance model latent heatflux for the four Landsat 8 image dates. Right field isirrigated maize, left is irrigated soybean. Poor temporalfrequency of images is illustrated by the timeline.Background aerial image is from USDA-NAIP 2014.

Planting 4/29-5/1

Harvest 10/15 -10/17

06/03

07/21

09/23

10/09

Figure 6. Comparison of TSEB and water balance modeledET for three points. Water balance ET is prior to selfcalibration for that date.

The model was tested using the four cloud-freeLandsat 8 images over maize and soybean researchfields at the University of Nebraska’s AgricultureResearch and Development Center (ARDC) nearMead, Neb. The TSEB model performed well for Rn

(RMSE = 20 W m⁻²), but poorly for other fluxes (RMSE= 88 to 126 W m⁻²), see Figure 4. The model________

Reflectance-based cropcoefficients comparedwell with time adjustedFAO-56 crop coefficients(Figure 5). Themagnitude of model selfcalibration is illustratedfor three points in Figure

Figure 4. Comparison of TSEB modeled energy fluxes witheddy covariance fluxes for all four image dates, solidsymbols are for 7/21/2013, and the solid line is unity.

performed best forJuly 21, 2013. Thepoor performancefor June 3 may bean artifact ofimagery atmosph-eric corrections.The model may bepoorly parameter-rized for the lateseason and sene-scence.

6. The model req-uired the most adju-stment for the dry-land maize. Imagefrequency is a prob-able cause of themagnitude of theadjustments. Figure7 is a timeline ofTSEB LE for the

Irrigated maize and soybean fields illustrating the gapsin available imagery and variability in ET.

Figure 8. Soil mapping and aerial imagery from unmannedaircraft will be used to provide data input into the hybridmodel for VRI management in 2015 and 2016. Aircraftimage from nuaire.unl.edu.

Figure 9. The model will be implemented using a centerpivot with individual nozzle control in a research field atthe University of Nebraska’s ARDC facility near Mead,Neb. Image at right courtesy of T. Lo.

The model will be implemented tomanage irrigation in an 130 acreresearch field at the ARDC facility.The research field is planted to halfmaize – half soybeans. Unmannedaircraft from the University ofNebraska’s NU-AIRE program(nuaire.unl.edu) will be used tocollect multispectral imagery for themodel. Management zones will bedelineated within the field andtreatments of VRI and conventionalirrigation will be imposed within the

Variable rate irrigation (VRI) technology enables water users to applyvarying rates of water to zones throughout a single field. Determiningirrigation requirements (or prescriptions) at these small spatial scales isa critical step in assessing the benefits of VRI. A hybrid remote sensingevapotranspiration (ET) model (Neale et al. 2012) is being evaluated foruse in developing VRI prescriptions. The model is a water balance (WB)model coupled with a thermal-infrared-based two-source energybalance (TSEB) model (Norman et al. 1995) (Figure 1). The WB modeluses reference ET and vegetation index based crop coefficients (Kcbrf)(Neale et al. 1989) to run a spatial soil water balance (Figure 2).

Reference to commercial products or trade names is made with the understanding that no discrimination is intended of those not mentioned and no endorsement by the University of Nebraska-Lincoln is implied for those mentioned.

ET = LE = Rn - G + H

Figure 1. Simplified surface energy balance applied toplants and soil separately in the TSEB.

Dr, i = Dr, i-1 − (P − RO)i − Ii − CRi + ET, i + DPi

Figure 2. Simplified water balance, where CR, and RO areassumed to be zero and i is the present time step.

A depiction of the modelimplementation is shown inFigure 3. On Image dates ETis found by both methods.The WB is updated dailybetween images. On imagedates Kcbrf-WB-ET is rec-onciled with TSEB-ET, whichis considered to be moreaccurate.

Figure 3. Representation of the hybrid model operation for VRIprescription development.

zones. The model performance will be evaluated in terms of the accuracy of the hybrid model’s calibrated water balance over time, total crop water use (from neutron probe measurements), and yield as compared with conventional irrigation management.

Figure 5. Plot of SETMI modeled Kcbrf with a typicalKcb curve following FAO-56 (Allen et al. 1998) forflux tower locations in irrigated maize andsoybeans.