The Full Season: Nine Crops, One Pipeline, and the Index Crossover Science That Drives the Schedule

March 18, 2026 · 18 min read ·

ssnmndvindre

Penrose Field Notes · SSNM Series · Article III of III

The first two articles in this series built a complete diagnostic and processing system. Article I established why two sensors — multispectral and RGB — are required to separate nitrogen deficiency from water stress and deliver a prescription that acts on the right problem. Article II traced the full data pipeline from raw imagery through radiometric calibration, H3 hexagonal indexing, WebODM processing, N-rich strip calibration, and final shapefile export to the field terminal. Both articles used the same six-crop frame that dominates Magic Valley’s irrigated acreage: Russet Burbank potato, sugarbeet, winter wheat, grain corn, barley, and dry beans.

This article does three things. First, it expands the crop set from six to nine, adding the three crops that complete an honest picture of what Magic Valley farmers actually grow. Second, it maps every crop’s growth cycle to specific drone deployment windows — not as a generic recommendation, but as a scheduling calendar anchored to the NDVI-to-NDRE index crossover science and crop-specific nitrogen demand curves. Third, it describes how the pipeline from Article II absorbs these additions: where new sensor data flows in and how the expanded crop set surfaces as a configurable parameter in zone delineation rather than a hard-coded assumption.

The goal is not a comprehensive agronomic guide to each crop. That literature exists. The goal is to specify exactly when to fly, which sensor configuration to deploy, and what the pipeline does with the result — for every major crop on a Magic Valley center-pivot operation.

Three Crops That Complete the Dataset

The original six crops represent Magic Valley’s row-crop core. But two major agricultural categories are missing: the region’s dominant forage economy and its highest-value specialty crop. Leaving them out produces a scouting calendar with a significant market gap.

Alfalfa is the most consequential addition by volume. Idaho ranks first nationally in alfalfa dry hay production by tonnage, with 940,000 harvested acres and 3.9 million tons of dry hay (USDA NASS, 2024). Twin Falls County alone carries 84,037 hay acres — the county’s single largest crop category (2017 Ag Census). Alfalfa underpins Magic Valley’s dairy sector, which holds approximately 72–75% of Idaho’s dairy cows. What makes alfalfa agronomically interesting for drone scheduling is its 3–4 annual cutting cycles: unlike every annual crop in the original six, alfalfa resets its scouting calendar with each cut, creating multiple independent monitoring windows per season. The drone-to-prescription logic must account for cutting timing, not just calendar date.

Silage corn warrants separate treatment from grain corn even though both crops share the same species and vegetative growth stages. Idaho harvests 265,000 acres of corn for silage, typically ranking in the top five to eight nationally, and silage represents roughly 70% of the state’s total corn acreage (USDA NASS, 2024). Idaho leads the nation in silage yield at 30–31 tons per acre. The agronomic difference is decisive: silage corn is harvested 2–4 weeks before grain corn at the half-to-two-thirds milk line, targeting whole-plant moisture content and fiber digestibility rather than kernel dry-down. The critical late-season drone flight shifts accordingly — instead of monitoring for grain maturity and stalk quality, it monitors for whole-plant quality and optimal harvest timing, with NDVI decline used as a proxy for plant senescence relative to the target moisture window.

Onion completes the set as the highest-value specialty crop on the Snake River Plain. Idaho produces approximately 10,100 acres of dry bulb onions valued at $123.3 million (Idaho Department of Agriculture, 2022). Primary production concentrates in the Treasure Valley, 150 miles west, but the supply chain runs through Twin Falls and production acreage extends across the south-central corridor. Onion is a strong candidate for precision drone management because of three converging factors: its shallow root system makes it acutely sensitive to both water deficit and excess nitrogen; it is highly vulnerable to thrips and Iris Yellow Spot Virus, which drone RGB surveys can flag before infestation spreads; and its nitrogen management has a hard cutoff — all N application must stop by mid-July to prevent thick necks and storage failure (PNW 546, Sullivan et al., 2001). That cutoff creates an urgency structure similar to sugarbeet: the prescription window closes early, and late data is monitoring data, not action data.

Two other candidates were evaluated and excluded. Canola occupies 87,000–97,000 statewide acres but concentrates in northern Idaho’s Palouse, outside Magic Valley’s irrigated cropping system. Hops cover 5,797 acres but cluster in Canyon County with insufficient Magic Valley presence to justify crop-card treatment in this series.

When NDVI Stops Working: The Index Crossover Problem

Articles I and II introduced the NDVI-to-NDRE handoff as an operational necessity. This article explains the physics in enough detail to make the scheduling decisions defensible, because the crossover point is not arbitrary — it is a spectral constraint that directly determines whether a flight produces actionable data or a false sense of coverage.

NDVI uses the red band (approximately 670 nm) and the NIR band (860 nm on the M3M). At moderate canopy density, red reflectance drops rapidly as chlorophyll absorbs incoming radiation across successive leaf layers. The problem is that absorption is nearly complete at only a few leaf layers — after that point, additional biomass produces no further change in red reflectance. The NDVI formula, a normalized difference between NIR and red, compresses the ratio as the denominator approaches its minimum. This produces mathematical saturation independent of any real change in nitrogen status.

NDRE substitutes the red-edge band (730 nm on the M3M) for the red band. Chlorophyll absorption is weaker at 730 nm than at 670 nm, and leaf transmittance is meaningfully higher, allowing radiation to penetrate deeper into the canopy and interact with more leaf layers before saturation occurs. The result is a much wider dynamic range at the LAI values typical of mid-to-late season irrigated crops in the Snake River Plain.

The published crossover threshold is approximately LAI 2–3 m²/m² as a general rule (Haboudane et al., 2004; Si et al., 2023, Remote Sensing of Environment). Mutanga and Skidmore (2004) found that standard NDVI explained only R² = 0.26 for biomass at high canopy density, while narrow-band red-edge indices achieved R² = 0.77 in the same dataset. For nitrogen status specifically, NDRE explained 81% of variance in wheat N uptake across a global UAV meta-analysis, compared to NDVI’s R² range of 0.11–0.65 across crops (Castilho et al., 2024, Scientific Data; Magney et al., 2017, Precision Agriculture). NDRE-based variable-rate fertilization trials in winter wheat reduced total nitrogen use by 5–40% depending on field heterogeneity, without yield penalty (Argento et al., 2021, Precision Agriculture).

The crop-specific crossover points that anchor the scheduling model below are:

Crop	NDVI saturation point	Switch to NDRE at	Key reference
Grain corn / silage corn	V12 growth stage (approx. LAI 3)	After V6–V8 (canopy closure)	Sharma et al., 2015, Sensors
Winter wheat	Booting–heading (Feekes 8–10); NDVI ≥0.8	From jointing onward	Raun et al., 2001; Magney et al., 2017, Precision Agriculture
Potato	Peak canopy ~60 DAP	After full canopy closure	General potato canopy literature; cf. Nguy-Robertson et al., 2014
Sugarbeet	Dense canopy (~70 DAE)	Mid-season high-chlorophyll canopy	No published sugarbeet-specific threshold; estimated from canopy density
Barley	Mid-season after tillering	Heading stage and beyond	Nelsen & Lundy, 2020, Agronomy Journal
Onion	Post-bulbing initiation	Once leaves begin to overlap	No published threshold; estimated LAI 2.5
Alfalfa	3rd+ week of regrowth	Per-cut: after visible canopy closure	No published threshold; use SAVI → NDRE transition logic

Nitrogen Demand Curves and the Scheduling Calendar

The highest-ROI drone flights coincide with each crop’s rapid nitrogen uptake window — the phase when the plant is absorbing 2–5 lb N/acre/day and a spatial prescription can redirect fertilizer from adequately supplied zones to deficient ones before the deficit becomes yield loss. Flying two weeks too early produces data with no prescription leverage. Flying two weeks too late produces data that describes damage already done.

The demand curves below are anchored to University of Idaho Extension data and PNW extension publications. They define the phase boundaries the flight schedule is built around.

Russet Burbank potato takes up 150–250 lb N/acre over the season (UI Extension BUL 840, Stark et al., 2004). The critical Phase II runs from approximately 40 to 100 days after planting, absorbing 3–4 lb N/acre/day at peak (PNW 513, Sullivan et al., 1999). Sixty percent of seasonal N is consumed by day 75. Standard Magic Valley practice splits applications: 25–50% preplant, remainder fertigated at 20–40 lb N/acre every 7–14 days during bulking, guided by petiole NO₃-N monitoring (targets: 15,000–20,000 ppm vegetative, 10,000–15,000 ppm bulking). The drone prescription at early bulking gates which zones receive each fertigation increment.

Grain corn demands 200–300+ lb N/acre for irrigated yields of 180–220 bu/acre in Magic Valley (PNW 615, Brown et al., 2010). The V6-to-R1 window captures 55–60% of total uptake at peak rates of 3–5 lb N/acre/day between V10 and VT (Abendroth et al., 2011, Iowa State; Montana State EB0191). By silking (R1), approximately 63% of total plant N has been accumulated. Sidedress at V6–V8 delivers the largest single application; fertigation through center pivots enables fine-grained management through tasseling.

Silage corn follows an identical demand curve to grain corn through VT but the prescription logic diverges at R3–R4. Grain corn late-season N management targets standability and kernel fill; silage corn late-season management targets whole-plant starch content, fiber digestibility, and moisture uniformity across the field. A spatially variable NDVI decline map at R3 identifies zones that will hit the harvest-moisture target early, enabling strip-by-strip harvest sequencing rather than field-level average decision-making.

Winter wheat accumulates 100–150 lb N/acre, with the jointing-to-heading window (Feekes 6–10) absorbing 60–100 lb N/acre at 2–3 lb N/acre/day over 5–8 weeks (PNW 513). Idaho irrigated rates run 160–330 lb total available N/acre depending on yield goal (UI CIS 373). Nitrogen shortage during jointing is not recoverable with later applications — the tiller differentiation and stem elongation processes that determine yield potential are already complete.

Sugarbeet presents a unique constraint: all nitrogen must be applied before canopy closure because late-season N reduces sucrose percentage in the harvested root. Total uptake runs 150–250 lb N/acre, with an estimated 65–77% absorbed by 90 days after emergence. The rapid canopy-growth window (45–90 DAE, roughly June through mid-July) is the decisive period. Mid-season drone flights over sugarbeet function primarily as monitoring data for the following season’s prescription — with one important exception: the July NDRE flight that identifies zones with excess vegetative biomass, which predicts sucrose penalty at harvest.

Barley mirrors wheat on a compressed timeline: 80–150 lb N/acre total, peak uptake during tillering through heading over 5–7 weeks. The malt barley qualification adds a quality dimension that grain yield mapping misses: excess nitrogen drives grain protein above the 13.0–13.5% rejection threshold set by malt houses (varying by row type). The drone prescription at jointing must target N sufficiency, not N maximization. Under-application costs yield; over-application costs the contract.

Dry beans require 0–150 lb N/acre of applied fertilizer depending on soil N and yield goal (UI CIS 1189) because Rhizobium fixation supplies a substantial portion of their total demand. The critical drone windows are not nitrogen prescription windows but stress detection windows: iron chlorosis on Magic Valley’s calcareous soils produces spectral patterns indistinguishable from N deficiency on NDRE alone, making multi-flight temporal comparison and ground-truth validation particularly important to avoid prescribing iron chelate or supplemental N based on what is actually an irrigation or micronutrient problem.

Alfalfa fixes 150–300 lb N/acre/year and rarely needs supplemental nitrogen fertilizer. Its value to the drone program is different: stand density mapping after winter, cutting-date optimization using NDVI growth rate monitoring, and water stress detection during summer cuts when heat stress and irrigation timing interact. Each 3–4-week regrowth cycle has an internal mini-season: SAVI at green-up, NDVI through the vegetative growth phase, and NDRE in the week before cutting to assess stand health and identify sections that have declined in productivity.

Onion takes up 100–200 lb N/acre, with peak demand during bulb enlargement (July 1–August 15). All nitrogen application must stop by mid-July to prevent thick neck formation and storage failure (PNW 546, Sullivan et al., 2001). The precision value of drone monitoring for onions is concentrated in two windows: early-season stand and weed uniformity assessment (onions are extremely poor competitors with weeds before canopy closure), and the bulb-enlargement NDRE flight that identifies zones with irrigation deficits before leaf fall begins.

Nine-Crop Flight Schedule

Three sensor configurations appear throughout the schedule. Understanding their purpose before the crop-by-crop detail keeps the scheduling logic from becoming a list of dates.

RGB-only (Mavic 3 Cine, single platform): Stand counts, emergence uniformity, weed pressure assessment, canopy closure mapping. Appropriate when LAI is too low for spectral saturation to matter, stress discrimination is not needed, and spatial resolution at 2.7 cm GSD matters more than multispectral sensitivity. Rapid, low-cost, and sufficient for early-season decisions.

Single multispectral (M3M only): NDVI pre-closure, NDRE post-closure, CIre validation, zone classification. The workhorse configuration from canopy closure through grain fill or bulking. The N-rich strip NSI is computed from these flights for topdress and fertigation decisions. When paired with the Cine RGB (flown in the same weather window), the two-platform composite provides visual confirmation that spectral anomalies correspond to real canopy differences.

The schedule below represents recommended deployment windows. Actual trigger dates should be confirmed against GDD accumulation and field observations — calendar dates are Magic Valley averages at approximately 42.5°N, 3,000–4,500 ft elevation.

Crop	Flight	Window	Configuration	Primary decision
Potato	1	May 20–June 5	RGB	Stand counts, emergence gaps, replant zones
	2	June 20–July 5	Single MS	Canopy closure verification, pivot uniformity, early NDRE baseline
	3	July 10–25	Single MS	Peak prescription flight — NDRE NSI for VRA fertigation
	4	Aug 5–20	Single MS	Late-bulking quality monitoring, late blight spatial risk
	5	Sept 1–15	Single MS	Vine maturity uniformity, vine-kill timing
Grain corn	1	June 5–20	RGB	V4–V6 stand, planter skip detection
	2	July 1–15	Single MS	V10–V14 NDVI for sidedress N — NDRE if V12 reached
	3	July 20–Aug 5	Single MS	VT–R1 pollination stress, critical yield-set window
	4	Aug 15–30	Single MS	R3–R4 late N, disease pressure, stalk quality
	5	Sept 20–Oct 5	Single MS	R5–R6 kernel black layer, harvest timing
Silage corn	1	June 5–20	RGB	V4–V6 stand (same as grain corn)
	2	July 1–15	Single MS	V10–V14 sidedress N
	3	July 20–Aug 5	Single MS	VT–R1 stress assessment, critical yield-set window
	4	Aug 15–Sept 1	Single MS	R3–R4 whole-plant quality — NDVI decline rate as harvest-timing proxy, moisture zone mapping
Winter wheat	1	Mar 15–Apr 5	Single MS	Spring green-up, winter kill mapping, N topdress decision
	2	Apr 25–May 10	Single MS	Jointing NDRE — last topdress window
	3	May 25–June 10	Single MS	Flag leaf / boot — fungicide timing, NDRE stress assessment
	4	June 20–July 5	Single MS	Grain fill NDRE — final yield prediction
	5	July 15–25	Single MS	Maturity uniformity, harvest timing by zone
Sugarbeet	1	May 10–25	RGB	2–4 leaf stand, early weed pressure
	2	June 15–30	Single MS	Pre-canopy closure — last N application window
	3	July 15–30	Single MS	Full canopy NDRE — identify excess biomass zones (sucrose penalty prediction)
	4	Aug 20–Sept 5	Single MS	Late root growth monitoring, maturity zones
	5	Sept 15–30	Single MS	Pre-harvest: root quality prediction, zone-by-zone harvest sequencing
Barley	1	Apr 10–25	RGB	Stand, green-up uniformity
	2	May 1–15	Single MS	Tillering NDRE — early N status, late topdress
	3	June 10–25	Single MS	Boot / heading — malt protein risk mapping, NDRE stress
	4	July 5–15	Single MS	Grain fill — final quality monitoring
	5	July 25–Aug 5	Single MS	Maturity and harvest timing
Dry beans	1	June 15–30	RGB	V2–V4 stand, weed pressure, Fe chlorosis early scan
	2	July 10–20	Single MS	Flowering — stress discrimination (Fe chlorosis vs. N vs. irrigation)
	3	Aug 1–15	Single MS	Pod fill — final stress discrimination, disease/virus risk
	4	Aug 25–Sept 10	Single MS	Maturity, harvest-aid timing
Alfalfa (per-season cycle)	1	Apr 15–30	Single MS	Spring green-up, winter stand loss mapping
	2	May 15–25	Single MS	Pre-first-cut NDVI growth rate, cutting-date optimization
	3	June 20–July 5	Single MS	Post-1st-cut regrowth — NDRE summer stress assessment
	4	Aug 1–15	Single MS	Post-2nd-cut regrowth — NDRE peak heat stress assessment
	5	Sept 20–Oct 5	Single MS	Fall stand evaluation, winterization timing
Onion	1	May 1–20	RGB	2–4 leaf stand, weed mapping (onions are poor competitors)
	2	June 10–25	Single MS	Pre-bulbing — early stress detection, thrips pressure scan
	3	July 10–25	Single MS	Bulb enlargement — NDRE N status before cutoff, stress mapping
	4	Aug 5–20	Single MS	Maturity — leaf-fall uniformity, irrigation cutoff timing

Why July Is the Operational Bottleneck

The flight schedule reveals a pattern that calendar dates alone obscure. The first two weeks of July are when potato enters early-mid bulking (the highest-value SSNM flight of the year), corn crosses the V12 canopy closure line triggering the NDVI-to-NDRE switch, sugarbeet reaches full canopy just as the N application window closes permanently, and winter wheat grain fill overlaps with flag leaf fungicide timing. Five crops simultaneously require flights that include NDRE, and the window for actionable prescriptions is measured in days rather than weeks.

This is not a hypothetical scheduling challenge — it is the reason a single platform cannot cover a diversified Magic Valley farm at the correct resolution. The capacity math is straightforward: a single M3M at standard 80/75 overlap covers 400–500 acres per operational day (accounting for battery logistics, weather windows, and panel calibration). A 1,000-acre operation with even modest diversification across four crops will generate more flight demand in the first two weeks of July than one platform can execute. Planning around that bottleneck — by staggering crop mixes and prioritizing the highest-value flights when conflicts arise — is as much a business constraint as an agronomic one.

The priority order when flights must be triaged: potato at early bulking (highest N demand rate, tightest action window, highest value per acre), onion at bulb enlargement (hard nitrogen cutoff, no second chance), sugarbeet at full canopy (last chance to identify zones with sucrose penalty risk), then corn, wheat, and barley in approximate sequence.

Pipeline Expansion: Absorbing the New Crops

The data pipeline described in Article II was built around six row crops and two sensor platforms: the M3M multispectral and the Mavic 3 Cine RGB. The nine-crop expansion adds three new crops. The pipeline structure does not change; the crop-type configuration is extended to parameterize the new entries.

Crop-Type Configuration as a Pipeline Parameter

The nine-crop expansion makes crop type an explicit runtime parameter rather than an implicit assumption. Each crop has a configuration entry that specifies the active index for the current growth stage (NDVI or NDRE), the NSI thresholds that trigger a prescription, and the N-rich strip polygon for that field and season. A simplified schema:

{
  "crop": "potato",
  "growth_stage": "early_bulking",
  "active_index": "NDRE",
  "nsi_threshold": 0.95,
  "n_strip_polygon": "path/to/strip.geojson",
  "application_zones": {
    "min_zone_area_acres": 2.0,
    "max_zones": 5,
    "clustering_method": "jenks"
  }
}

Alfalfa requires a modified schema because it is managed by cutting cycle rather than by days after planting. The growth_stage field for alfalfa accepts regrowth_week_1 through regrowth_week_4, and the active index transitions from SAVI at week 1 (sparse canopy on bright soil) to NDVI at weeks 2–3, to NDRE at week 4 when the canopy has closed. This per-cycle state machine is the only structural addition the new crops require.

Open-Source Stack Summary

The complete pipeline from flight to prescription runs on the following open-source components, all of which have active maintenance and documented agricultural use cases:

OpenDroneMap (WebODM 3.6.x) — Photogrammetric reconstruction, multispectral orthomosaic generation, radiometric calibration. The primary processing engine. Self-hostable on GCP, AWS, or bare metal with ClusterODM for parallel job execution across multiple fields.

GDAL / OGR (3.9.x) — Raster reprojection, co-registration, band math, Cloud-Optimized GeoTIFF creation, shapefile export. The plumbing layer that connects every other component.

Python ecosystem — rasterio (raster I/O), geopandas (vector operations), numpy (array math), scikit-learn (k-means / Jenks clustering), rasterstats (H3 cell zonal statistics), h3-py (hexagonal grid generation). The analytical core.

PostGIS / PostgreSQL — Multi-season spatial database storing field boundaries, H3 cell statistics, prescription history, and strip polygon references. Enables year-over-year zone stability analysis.

QGIS (3.x LTR) + CSIRO PAT plugin — Visualization, manual zone editing, prescription review before export. The agronomist review layer before any prescription reaches the controller.

FIELDimageR (github.com/OpenDroneMap/FIELDimageR) — R package for orthomosaic-to-index-to-plot extraction with built-in soil masking (fieldMask()), published in The Plant Phenome Journal (Matias et al., 2020). Useful as a validation layer against the Python pipeline.

ADAPT / ISOv4Plugin (github.com/ADAPT/ADAPT) — Agricultural Data Application Programming Transform for ISOXML read/write. Handles the prescription file format translation for ISOBUS-compliant terminals.

For prescription file delivery: shapefiles in WGS84 (EPSG:4326) with a rate attribute column remain the most universally compatible format across John Deere GreenStar, Raven Slingshot, Trimble, and Ag Leader terminals. ISOXML (ISO 11783-10) is the correct interoperability standard for ISOBUS terminals from Case IH, New Holland, and AGCO, but v3/v4 version mismatches between terminals require per-client testing. The practical ceiling for zone complexity is 3–5 management zones — fewer misses within-field variability, more exceeds most controllers’ polygon-processing capacity at field speed.

The one constraint the open-source stack does not resolve is the index-to-rate calibration problem. No algorithm can derive the correct N rate from an NDRE value alone without knowing the relationship between that index and soil nitrogen supply for that specific field, soil series, and season. The N-rich strip NSI methodology from Article II is the calibration anchor. The strip’s NDRE value becomes the denominator; the field’s NDRE value is the numerator; the ratio drives the rate lookup table that has been calibrated against field observations and tissue test data. That calibration is what separates a prescription from a map.

The Full-Season View

Across a diversified nine-crop Magic Valley operation — potato, grain corn, silage corn, winter wheat, sugarbeet, barley, dry beans, alfalfa, and onion — the framework generates somewhere between 20 and 40 flights per season depending on the crop mix and total acreage. The distribution is not uniform. March through May sees six to eight flights, mostly single-platform, mostly early-season stand assessment and N topdress decisions on winter wheat. June through mid-August is the dense window: every week from mid-June to early August has at least two crops requiring NDRE-based multispectral coverage, and the first two weeks of July represent the peak conflict period where potato, corn, sugarbeet, and wheat all demand simultaneous high-priority flights.

The economic case for that flight density rests on a straightforward comparison. A meta-analysis of 85 empirical studies found approximately 22.3% ROI improvement and 18.5% net profit increase attributable to variable-rate management (Lan & Ban, 2025). Variable-rate nitrogen delivered +€163.8/ha in wheat from a 7.2% marginal return improvement (Tsitouras et al., 2025, Agronomy) and +30–101% gross margin improvement in simulation-based VRA trials (Gobbo et al., 2022, Precision Agriculture). The multispectral framework — anchored to the N-rich strip NSI methodology — provides the spatial precision that uniform-rate application cannot: the certainty that a low-NDRE zone flagged for nitrogen application is receiving the right rate, and that the correction can be validated in the next flight window against the strip reference.

The pipeline described across these three articles is not a research prototype. Each component — WebODM, GDAL, rasterio, PostGIS, QGIS — is in active production use across agricultural and geospatial applications worldwide. The integration is the work. Assembling it, calibrating it against local Idaho soil series and crop physiology, validating prescriptions against petiole and tissue tests in the first season, and building the multi-year dataset that makes zone delineation stable from year to year — that is the engineering investment that turns a drone and an open-source stack into a site-specific nutrient management system.

What this series has established: the diagnostic framework, the data pipeline, and the scheduling logic. The next question — how that pipeline performs across multiple seasons against actual yield and quality data — is an empirical one, and one that the instrumented fields of Magic Valley are particularly well-positioned to answer.

References

Abendroth, L.J. et al. (2011). Corn Growth and Development. Iowa State University Extension PMR-1009.
Argento, F. et al. (2021). Site-specific nitrogen management in winter wheat supported by low-altitude remote sensing and soil data. Precision Agriculture, 22, 364–382.
Brown, B. et al. (2010). Fertilizer Guide: Irrigated Corn in Southern Idaho. University of Idaho Extension PNW 615.
Castilho, D. et al. (2024). A global dataset for assessing nitrogen-related plant traits using drone imagery in major field crop species. Scientific Data, 11, 561.
Ebmeyer, H. & Hoffmann, C.M. (2021). Efficiency of nitrogen uptake and utilization in sugar beet genotypes. Field Crops Research, 273, 108270.
Gobbo, S. et al. (2022). Precision agriculture from satellite and drone images for variable-rate nitrogen management. Precision Agriculture, 23, 1922–1948.
Haboudane, D. et al. (2004). Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop canopies. Remote Sensing of Environment, 90(3), 337–352.
Idaho Department of Agriculture (2022). 2022 Idaho Agricultural Statistics.
Lan, Y. & Ban, H. (2025). A meta-analysis of precision agriculture technology adoption and farm-level economic impacts. Sustainability, 17(24), 11223.
Magney, T.S. et al. (2017). Response of high frequency Photochemical Reflectance Index (PRI) measurements to environmental conditions in wheat. Remote Sensing of Environment, 173, 84–97.
Matias, F.I. et al. (2020). FIELDimageR: An R package to analyze orthomosaic images from agricultural field trials. The Plant Phenome Journal, 3(1), e20005.
Montana State University Extension (2011). Nutrient Uptake and Timing. EB0191.
Mutanga, O. & Skidmore, A.K. (2004). Narrow band vegetation indices overcome the saturation problem in biomass estimation. International Journal of Remote Sensing, 25(19), 3999–4014.
Nelsen, T.C. & Lundy, M.E. (2020). Nitrogen management and NDRE in barley for malt quality. Agronomy Journal, 112(3), 1881–1893.
Nguy-Robertson, A. et al. (2014). Estimating foliar chlorophyll in Zea mays, Triticum aestivum, Glycine max, and Solanum lycopersicum using spectroradiometer and chlorophyll meter data. Photosynthetica, 52(2), 181–192.
Raun, W.R. et al. (2001). In-season prediction of potential grain yield in winter wheat using canopy reflectance. Agronomy Journal, 93(1), 131–138.
Si, Y. et al. (2023). Evaluating the saturation effect of vegetation indices in forests using 3D radiative transfer simulations and satellite observations. Remote Sensing of Environment, 295, 113665.
Stark, J.C. et al. (2004). Nutrient Management Guidelines for Russet Burbank Potatoes. University of Idaho Extension BUL 840.
Sullivan, D.M. et al. (1999). Nutrient Management for Potato Production. PNW 513. Oregon State, Washington State, University of Idaho Extension.
Sullivan, D.M. et al. (2001). Nutrient Management for Onion Production. PNW 546. Oregon State, Washington State, University of Idaho Extension.
Tsitouras, A. et al. (2025). Variable-rate nitrogen in winter wheat: yield response and economic analysis. Agronomy, 15(2), 312.
USDA National Agricultural Statistics Service (2024). 2024 State Agriculture Overview: Idaho.