smart irrigation technologies for irrigation water management
TRANSCRIPT
Smart Irrigation Technologies for Irrigation Water
Management
Christopher Neale, Director of Research
Daugherty Water for Food Global Institute
Outline and Topics Covered
• Geospatial technologies for managing variable rate irrigation
• Infield Sensors
• Estimation of Evapotranspiration and Biomass using Satellite Remote Sensing
• Global Evapotranspiration Product
Variable Rate Irrigation (VRI)
Credit: Valmont
Sector/speed control: Zone control:
Video Demonstration:https://www.youtube.com/watch?v=u831kWRGVUk
Total Cost: ~$2,000(free with some new panels)
Total Cost: ~$18,000+
Disclaimer: UNL doesn’t endorse any particular brand of irrigation equipment.
Potential Benefits of VRI
• Avoid putting chemicals on waterways (create an “avoidance area”)
• Reduce pumping
• Energy savings
• Reduce nitrate leaching
• Prevent pivot from getting stuck
• Reduce yield losses due to over-irrigation
• Reduce water application rates on steep slopes (reduce runoff/erosion)
• Reduce over-application with corner arms
• Increased yield with given water allocation
=> Optimize the use of natural precipitation
VRI Zoning: Cosmic-ray Neutron Probes, huh?
and EC techniques
Dr. Trenton Franz, School of Natural Resources - UNL
CropMetrics, North Bend, NE
Precision AgricultureMapping done in 2 hours for both EC and VWC on 3/10/16, Brule Water Lab 1
Aerial Image SSURGO Database
EC Map VWC Map
Field Experiment - Methods
• Field Sites: ARDC, Mead NE and West Central Water Resources Field Laboratory near Brule, NE
ECa maps provided by J. Luck and T. Franz.
9
Soil Water Monitoring
Wireless NetworkSatellite
Communication
Data LoggerSensors
Excellent Recent Developments to
Improve Management to Conserve Water
and Energy, and Protect Water Quality
12
Tempest UAV Typical test flight pattern Thermal Infrared Natural Color
Biological Systems Engineering Department @ UNL
Normalized Difference Vegetation Index (NDVI) Image of Variable Rate Irrigation
Center Pivot at Mead, NE.
Image from the June 30th, 2016 flight
8 cm pixel resolution. NDVI uses red and
Near-infrared bands from the onboard
Micasense camera
14
RGB Image of VRI Pivot at Mead.
Image of September 8th, 2016 flight
8 cm pixel resolution.
Red, Green and Blue Bands
The Hybrid ET model 1
HS
H = HS+ HC
HC
TS
TAC
TA
TC RX
TRAD() = f(TS,TC, C())
Prognostic Modified FAO-564
type water balance of the root zone
ETa
P Irr.
DP CR
RO
FC
PWP
Diagnostic SVAT SchemeThe Two-Source Energy Balance Model (TSEB)2,3
Series Resistance FormulationLE = Rn – G – H
Modified with reflectance -based basal crop coefficient (Kcbrf)5
2 Norman and Kustas (1995), 3Li , et al.(2005)
1Neale et al. (2012), Soil water content estimation using a remote sensing based hybrid evapotranspiration modeling approach. In Advances in Water Resources.
4 Allen et al. (1998), 5Neale et al. (1989), Campos et al. (2016)
ETa = Kc .ET0
Kc = Kcbrf . Ka + Ke
TRAD(q) ~ fc(q)Tc + [1-fc(q)]Ts
(two-source approximation)Norman, Kustas et al. (1995)
Provides information onsoil/plant fluxes and stress
TRAD(q) ~ fc(q)Tc + [1-fc(q)]Ts
Accommodates off-nadirthermal sensor view angles
Treats soil/plant-atmosphere coupling differences explicitly
Two-Source Energy Balance Model (TSEB)
RN
System and Component Energy Balance
= H + lE + G
RNC = HC + lEC
RNS = HS + lES + G
= = =
+ + +
TS
TCTAERO
SY
ST
EM
CA
NO
PY
SO
IL
Derived fluxes
Derived states
TRAD
0
2
4
6
8
10
02/09/02 04/09/02 06/09/02 08/09/02 10/09/02 12/09/02
ET (
mm
/day
)
Time (days)
SVAT
Image
Prognostic model
FAO 56
Assimilation of ET + Update of Soil Moisture
Diagnostic model
Kcb
The Hybrid Model1
1Neale et al. (2012), Soil water content estimation using a remote sensing based hybrid evapotranspiration modeling approach. In Advances in Water Resources. Volume 50, Pages 152-161, ISSN 0309-1708, 10.1016/j.advwatres.2012.10.008.
Methods – SETMI Interface• Spatial EvapoTranspiration
Modeling Interface (SETMI)
• Operates in ArcGIS Environment
Geli, H. M. E. and C.M.U. Neale, (2012), Spatial evapotranspiration modeling (SETMI), Proc. IAHS 352, Remote Sensing and Hydrology (September 2010), ISSN 0144-7815
Multispectral Imagery from Satellite and UAV (Variable Spatial –and Temporal)
VRI Prescriptions from SETMI
ECa Map (High Spatial)
Weather Data (High Temporal)
SETMI –Based VRI
Model
Previous Irrigation Maps (Resolution Dependent on VRI)
Soil Moisture (Low Spatial, High Temporal)
New VRI Prescription and Schedule
Reflectance-based Kcb & Water BalanceWright,
1982
Corn,
1982 The Kcb
represents the
average ET
from plant
transpiration
with a dry soil
background
and no
limitation of
soil water in
the root zone
of the crop
(From FAO56)
ETa = Kc ETr
Kc = (Kcb*Ka +Ks)
Neale, C. M. U., W.C. Bausch and D.F. Heermann. 1989. Development of reflectance based crop coefficients for corn.
Transactions of the ASAE,Vol. 32(6):1891-1899
Reflectance-based crop Coefficients
Are obtained by linearly relating the NDVI or SAVI of bare soil with the NDVI or SAVI at
effective full cover the point of maximum ET on a crop coefficient curve
SAVI = (NIR – Red) (1+L)/ (NIR + Red + L)
Effective full cover occurs at LAI varying from 2.7 to 3.5 depending on the crop and with
percent cover around 80%
SAVI and NDVI are vegetation indices estimated from Red and Near-Infrared bands of
satellite, airborne sensor or ground radiometers
Neale et al., 1989; Bausch and Neale,1989
Bausch, 1993
Reflectance based Kcb Transformations
Rationale for re-examining Kbcrf for Corn
• Original research in the mid 1980’s was based on very different varieties with a more planophile leaf structure, shorter plants and planted to a lower plant population reaching lower maximum LAI values in the field
• New hybrid varieties have an erectophile upper canopy, are taller and planted at a higher density reaching higher LAI values in the field
New Reflectance based Kcb for Corn and Soybeans
Application for ET and Irrigation requirements for Corn and Soybeans
RMSE< 1.1 mm RMSE< 1.0 mm RMSE<1.5 mm
Actual Irrigation
Irrigation requirements
Actual Irrigation
Irrigation requirements
New Reflectance based Kcb for Corn and Soybeans
Reflectance-based Crop Coefficients Redux: For Operational Evapotranspiration Estimates In The Age Of High Producing
Hybrid Varieties. 2017. Isidro Campos; Christopher M.U. Neale; Andrew E. Suyker; Timothy J. Arkebauer; Ivo Z. Gonçalves.
Agricultural Water Management.. Vol. 187, Pages 140-153. http://dx.doi.org/10.1016/j.agwat.2017.03.022
CWP = Yieldact / ETact (kg m−3)
where Yieldact is the actual marketable crop yield (kg ha−1)
ETact is the actual seasonal crop water consumption by
evapotranspiration (m3 ha−1)
Crop Water Productivity
Yield/Grain
production (Y)
ET & abiotic
stress (Water, Tª)
Based on soil water balance
remote sensing driven and
hybrid model
Site specific
Harvest index (HI)
Existing models on
Yield partitioning and
experimental data
Biomass (B)
Normalized water
productivity (WP*)
Normalization of WUE for
atmospheric demand
High to Medium resolution EO data:Landsat Constellation
MODIS Constellation
Agro-meteorological dataExisting ground-networks
and satellite based estimates
Yield estimates from remote sensing
Y=B*HI
Light use
efficiency
On the conservative behavior of biomass water productivity (Steduto et al. 2007)
WP*b: normalized water productivity
Biomass, accumulated biomass
Tc: Crop transpiration (adjusted for water stress).
∆e: Lear-to-air vapor pressure deficit, approximated by VPD
Analyzing the relationship between SAVI & crop production
WP*b: normalized water productivity
Biomass, accumulated biomass
Tc: Crop transpiration (adjusted for water stress).
Eo: Reference ETo
Drawbacks of the normalization based on VPD∆e < VPD if high transpiration rates of advection
VPD > ∆e if water stress (high canopy temperature)
Effects of the data availability and VPD formulation
Simulation of biomass production
WP*B: Normalized water productivity (g/m2)Kcb: Transpiration coefficientKSW: Water stress coefficientKST: Air temperature water stress coefficient
Analyzing the relationship between SAVI & crop production
Water stressbased on thesoil water balance
Temperature stressbased on the air temperature
WP*
Crop Yield and Water Productivity Estimation:• 𝑾𝑷𝑩
∗ =𝑩𝒊𝒐𝒎𝒂𝒔𝒔
𝒏𝒊=𝟏
𝑻𝒂𝒅𝒋
𝑬𝑻𝒐
=𝑩𝒊𝒐𝒎𝒂𝒔𝒔
𝒏𝒊=𝟏𝑲𝒄𝒃,𝒂𝒅𝒋
• 𝐵𝑖𝑜𝑚𝑎𝑠𝑠 = 𝑊𝑃𝐵∗ × 𝑛
𝑖=1𝐾𝑐𝑏 × 𝐾𝑆𝑇 × 𝐾𝑆𝑊
• 𝑌𝑖𝑒𝑙𝑑 = 𝐻𝐼 × 𝐵𝑖𝑜𝑚𝑎𝑠𝑠 = 𝐻𝐼 ×𝑊𝑃𝐵∗ × 𝑛
𝑖=1𝐾𝑐𝑏 × 𝐾𝑆𝑇 × 𝐾𝑆𝑊
Water Productivity And Crop Yield: A Simplified Remote Sensing Driven Operational Approach. 2017.
Isidro Campos; Christopher M.U. Neale; Timothy J. Arkebauer, Andrew E. Suyker; Ivo Z. Gonçalves. Agricultural and Forest Meteorology.
https://doi.org/10.1016/j.agrformet.2017.07.018
Analyzing the relationship between SAVI & crop production
Experimental values ofHarvest Index
for corn and soybeansin Mead, Ne
• Mean values around 0.5 for bothcrops and irrigation managements
• Strong variability depending onmeteorological conditions???
• Difficulty to determine empiricallybecause of the dynamics ofbiomass accumulation
Orange: Corn Green: Soybeans
Interpolation of SAVI data for the assessment of ΣKcb
• Strong variability of crop growth (ET, Biomass and yield)
Differences in ET (ΣKcb) result in variability of crop yield at field scale
Differences in WP* for corn and soybean result in differences in yield
Comparison of measured and modeled yield for corn and soybeans (2011-2012)
• Relative good but noisy
agreement for both
analyzed campaigns
• Higher production in
2012, extremely dry
year (flash drought in
the mid-western US)
• Sensitive to the great
range of corn production
in 2012
Comparison of measured and modeled yield for corn
plots in 2011 (blue) and 2012 (red)
CWP = Yieldact / ETact (kg m−3)
where Yieldact is the actual marketable crop yield (kg ha−1)
ETact is the actual seasonal crop water consumption by
evapotranspiration (m3 ha−1)
Crop Water Productivity
Nebraska
Upper
Republican
Center Pivots
Location of the Upper Republican River Basin in southwest
Nebraska
Example of Application in Nebraska
Christopher Neale
Water for Food Institute
University of Nebraska
Christopher Hain
Earth System Science Interdisciplinary Center, University of
Maryland, NOAA-NESDIS
Martha C. Anderson
USDA-Agricultural Research Service, Hydrology and Remote
Sensing Laboratory
Supplementing ALEXI Capabilities with Polar Orbiting Sensors
Time of Day
Lan
d S
urf
ac
e T
em
pe
ratu
re
Local NoonSunrise
Morning LST Rise: ALEXI Window
VIIRS Nighttime
LST
VIIRS Daytime
LST
A technique has been developed and evaluated using GOES data to train a regression model to use day-night LST differences from MODIS to predict the morning LST rise needed by ALEXI. The
regression model can provide reasonable estimates of the mid-morning rise in LST (RMSE ~ 5 to 8%) from the twice daily VIIRS LST observations.
TRAD(q) ~ fc(q)Tc + [1-fc(q)]Ts
(two-source approximation)Norman, Kustas et al. (1995)
Provides information onsoil/plant fluxes and stress
TRAD(q) ~ fc(q)Tc + [1-fc(q)]Ts
Accommodates off-nadirthermal sensor view angles
Treats soil/plant-atmosphere coupling differences explicitly
Two-Source Energy Balance Model (TSEB)
The ALEXI model runs the TSEB
Development of a High-Resolution (375-m) VIIRS ET Product
1. Mid-morning change in Land Surface Temperature
Development of a High-Resolution (375-m) VIIRS ET Product
2. Leaf Area Index and Fraction of Green Vegetation Cover (fc)
Development of a High-Resolution (375-m) VIIRS ET Product
3. Land Surface Albedo
• Only available VIIRS product is at 750-m – mapped to 375-m grid – used
to calculate surface reflectivity in VIS/NIR spectrum as needed by ALEXI
Development of a High-Resolution (375-m) VIIRS ET Product
4. Incoming Solar Radiation
• Only available from geostationary platforms – Meteosat (3-km) / will use
CFS-4 daily insolation for scaling to daily fluxes as back-up data source
[currently used for other ALEXI applications; not ideal due to model-
based estimate but model/RS-based insolation tend to converge when
considering monthly-annual time scales]
Development of a High-Resolution (375-m) VIIRS ET Product
5. Meteorological Surface Fields (e.g., air temperature; wind speed; surface
pressure; incoming LW)
• Climate Forecast System Reanalysis (hourly; 0.50º)
6. Morning Profile of Potential Temperature
• Climate Forecast System Reanalysis (hourly; 0.50º)
CFS-R fields are currently being used for all data fusion results (30-m ET) so we
don’t expect any issues with this dataset.
Development of a High-Resolution (375-m) VIIRS ET Product
7. Landcover / Vegetation Type
• Only available VIIRS product is at 1-km – insufficient for 375-m product;
• So we’ll use 30-m Landsat-based classification (Chinese dataset) – we’ve
downloaded all tiles over MENA region and have gridded to 375-m
domain and calculated % of each land class in each VIIRS pixel.
• What happens when Landsat-based classification is not representative of
VIIRS EVI/NDVI information?
• We’re developing a processing check to ensure changes to
agricultural practices which can be determined by VIIRS vegetation
time series to ensure we’re processing pixels which may be classified
as “barren” but now include agricultural pixels.
Development of a High-Resolution (375-m) VIIRS ET Product
8. Cloud Mask
• Only available VIIRS product is at 750-m – mapped to 375-m grid.
• Once we acquire more years of VIIRS data we’ll develop additional
climatological-based QC metrics to remove cloud contamination that is
“missed” in VIIRS cloud mask.
Development of a High-Resolution (375-m) VIIRS ET Product
Example location
*Shading indicates 1-km percentage of cropland from global synthesis of several RS-based land use maps
Development of a High-Resolution (375-m) VIIRS ET Product
Current MODIS Capability (1000-m) VIIRS Capability (375-m)
PROPOSED PRODUCTION TIMELINE FOR THE MENA REGION:
• ALEXI software has been ported to run on the HCC super computer at University of Nebraska-Lincoln
• September 2017: start producing the daily product at Nebraska
• September – December 2017: test and verify the product with ground data with partners in the MENA region and other countries: Spain, India, Brasil, USA
• January 2018: Launch the product through a website and make it available to registered users and collaborators
Disaggregation of VIIRS 375 m product to Higher Resolutions
• Disaggregation and data fusion downscaling to
higher resolutions will be done using the DisALEXI
model
• The 375 m product is the upper boundary
condition
• A new version of DisALEXI developed using open-
source tools (e.g., Python) is being developed for
this purpose called PyDisALEXI (no licensed
software required)
• We will train the cooperating agencies in the
different MENA countries to use this software and
downscale ET to agricultural and irrigated areas
0.00
3.75
7.50
11.25
15.00
Nile Delta Irrigation
Input data: ALEXI daily ET
VIIRS daily ET mm/d
• Daily ET calculated at
VIIRS 375 m data using
the ALEXI model.
0.00
3.75
7.50
11.25
15.00
Nile Delta Irrigation
Initial results: Landsat daily ET
Landsat daily ET mm/d
• Landsat Daily ET
downscaled from ALEXI
using the PyDisALEXI
model.
The Future is Bright – Think of the Possibilities
High Producing HybridCrop Varieties
Encapsulated Fertilizer
Monitoring with Unmanned Vehicle
Auto Steer Vehicles
ET from Remote Sensing
Soil/Plant Monitoring Satellite Communication
Variable Rate Irrigation
Now we have to put it all together
Look at Recent Developments
Courtesy of Derrel Martin