LIT_114: Reher et al. (2024) — arable agrivoltaic land-loss & LER, Belgium field trial

LIT_114: Reher et al. (2024) — arable agrivoltaic land-loss & LER, Belgium field trial

Source

doi:10.1016/j.apenergy.2024.122679 — original publication (opens in a new tab; the file is not redistributed)

Reher et al. (2024) — arable agrivoltaic land-loss & LER (Belgium field trial)

The measured PROXY behind the agri-PV land-retention fraction (partial resolution of RT_108b) — TRANSFERABLE, no NZ value

Peer-reviewed field study (Q1 journal Applied Energy 359:122679, DOI 10.1016/j.apenergy.2024.122679; KU Leuven). Two arable agrivoltaic pilot sites in Belgium (Grembergen: vertical bifacial + horizontal single-axis tracking; Lovenjoel: elevated/overhead), sugar beet and wheat over two seasons each. Its load-bearing figure for Neobiome is the measured “Land Loss (%)” (footprint loss from buffer/driving-safety zones, kept separate from shading yield loss): 11% near-ground (vertical & tracked) → ~0.89 land retained; 8% elevated → ~0.92 retained (Table 3). ⚠ Transferable proxy, data_quality: medium — Belgium (temperate maritime), a reasonable NZ analogue but not an NZ measurement (same convention as LIT_060 / LIT_107). It brackets the existing corpus figure Interview II [INT_002] (Hungary: <5% loss → ~0.95 retained) → carry a design-dependent range ~0.89–0.95 retained, not a fixed constant. Reher is measured, so a stronger anchor than a vendor/interview claim.

Summary

A peer-reviewed field study (Reher et al., Applied Energy 2024, KU Leuven) reporting two consecutive-season arable agrivoltaic (AV) trials in Belgium: an interspaced site at Grembergen with vertical bifacial and horizontal single-axis tracking rows growing sugar beet, and an elevated / overhead (“high-clearance”) site at Lovenjoel growing winter wheat. The study measures crop yield, canopy light (PAR), land equivalent ratio (LER), electricity yield, and — the figure Neobiome needs — “Land Loss (%)”: the fraction of field area taken out of cultivation by the AV structure’s buffer / driving-safety zones, reported separately from the shading-driven yield reduction.

For Neobiome the value is narrow and specific: it supplies a measured land-retention fraction for agri-PV, the parcel-conservation input tracked as RT_108b. The near-ground interspaced systems (vertical, tracked) each lose 11% of the field to buffer strips (~0.89 retained); the elevated overhead system loses 8% (~0.92 retained) because its posts sit on a sparser grid. That result is mildly counterintuitive — the taller, overhead system conserves more cultivable land than near-ground rows — and it brackets the corpus’s only prior figure (Interview II [INT_002], Hungary, <5%). The paper also reports measured LER of 1.00–1.22 for the arable systems, materially below the theoretical +60–80% land-use-efficiency gain in the German Fraunhofer guide (OT_050) — a useful reality-check on temperate-arable AV. Belgium context: a transferable proxy, tagged accordingly (no NZ calibration claimed).

Key claims

- claim: "Measured land loss (footprint loss) by AV layout — the land-retention figure. VERBATIM (Table 3, row 'Land Loss (%)'): vertical bifacial (sugar beet, Grembergen) = '11%' (2021 and 2022); horizontal single-axis tracking (sugar beet, Grembergen) = '11%' (2021 and 2022); elevated/overhead (wheat, Lovenjoel) = '8%' (2022 and 2023); reference ground-mounted PV (GM-PV) = 'N.A.'. Retained cultivable fraction (1 − loss) therefore = ~0.89 for both near-ground interspaced systems and ~0.92 for the elevated system. [verified verbatim vs Table 3, pdftotext -layout]"
  source_location: "Table 3 'Overview of the different performance indicators of the 3 agrivoltaic pilot sites and the reference ground mounted PV (GM-PV) system', row 'Land Loss (%)' (p.10)"
- claim: "Definition of land loss — footprint loss from buffer zones, kept SEPARATE from shading yield loss. VERBATIM (§2.2.5 Land Equivalent Ratio, methods): 'Given that, by installing an agrivoltaic system, not the entire land area can be used for agriculture, buffer zones are accounted for in the LER through the land loss (LL) factor (fraction of uncultivated land due to the AV system).' Nomenclature (LL): 'Land loss % due to AV system.' The buffer is quantified elsewhere as a '0.5 m on each side' strip around the modules; the LL factor is thus a purely geometric footprint loss, applied in the LER independently of the crop-yield (shading) reduction, which is captured by the separate agricultural-yield term. [verified verbatim, pdftotext -layout]"
  source_location: "§2.2.5 'land equivalent ratio (LER)' (methods); nomenclature list (p.2); buffer dimension in §2.1 site description"
- claim: "Abstract restatement of land loss alongside yield loss (the two are reported separately). VERBATIM (Abstract): 'Beet grown in between vertical or tracked AV systems achieved a reduced average yield of 11–19% for the cultivation area, with a land-loss of 11%, and an electricity yield of 376–560 MWh/ha across two seasons.' and 'Wheat was cultivated under an elevated horizontal AV system in 2022 and 2023. Yields were reduced by 33% in 2022 and 46% in 2023 in addition to 8% land loss.' So land loss (11% / 8% footprint) and shading yield loss (11–19% beet; 33–46% wheat) are distinct components — confirming the land-retention fraction is the footprint term only. [verified verbatim, pdftotext -layout]"
  source_location: "Abstract (p.1)"
- claim: "Cause of the land loss (driving-safety buffers) and the counterintuitive near-ground > elevated ordering. VERBATIM (§4 Discussion): 'both experimental sites were faced with a substantial land loss component (Table 3), decreasing overall system efficiency. Growers and agricultural contractors raised several concerns about driving safety distances to the modules, especially during wet and slippery conditions and when working perpendicular to the field's slope or during harvesting maneuvers.' Conclusions: 'in order to minimize land losses (buffer zones), further optimization of farming vehicle driving systems as well as mitigation of soil compaction during construction are needed.' Interpretation (from the Table 3 values, not a verbatim quote): the elevated/overhead system loses LESS land (8%) than the near-ground interspaced rows (11%) — overhead posts occupy a sparser ground grid, whereas interspaced rows demand wider driving-safety strips. [quotes verified verbatim; ordering is an interpretation of Table 3]"
  source_location: "§4 Discussion (land-use-efficiency limitations); Conclusions"
- claim: "Measured LER for arable AV is only 1.00–1.22 — a reality-check on theoretical land-use-efficiency gains. VERBATIM (Abstract): 'A land equivalent ratio (LER) of 1.00 and 1.18 for the vertical system and 1.17 and 1.22 for the tracked system were recorded in 2021 and 2022, respectively.' and the elevated wheat system 'achieved a LER of 1.09 in 2022 and 0.97 in 2023.' So measured whole-system LER spans 0.97–1.22 (slightly net-positive at best) for open-field arable AV under a temperate maritime climate — materially below the ~1.6–1.8 (+60–80%) theoretical land-use-efficiency figure carried from OT_050. [verified verbatim, pdftotext -layout]"
  source_location: "Abstract (p.1); §3.3.2 'Land equivalent ratio'; Table 3 row 'LER'"

Neobiome Intelligence relevance

SUPPLIES the agri-PV land-retention fraction (RT_108b), as a measured PROXY. The parcel-conservation input — the share of a field that stays cultivable after an agri-PV build — reads directly from this paper’s measured “Land Loss (%)”: ~0.89 retained for near-ground interspaced layouts (vertical bifacial, single-axis tracked), ~0.92 for elevated/overhead. Combined with the corpus’s only prior datapoint (Interview II [INT_002], Hungary, <5% loss → ~0.95 retained), the defensible model treatment is a design-dependent range of ~0.89–0.95 retained, not a single constant. Reher is field-measured and peer-reviewed, so it is the stronger anchor for the lower/middle of that range; int_002’s <5% (an interview/vendor figure for a stow-tracker interspace system) sits at the optimistic end.

Keep land loss SEPARATE from yield derate. The paper is explicit that footprint land loss (buffer/driving-safety zones, 11% / 8%) is a distinct term from the shading-driven crop-yield reduction (11–19% beet; 33–46% wheat), each entering the LER separately. The engine must not double-count: the land-retention fraction (~0.89–0.95) is a geometric parcel-conservation multiplier; the under-panel yield derate is a separate multiplier sourced elsewhere (pasture proxy). This source is the clean evidence that the two are independent.

Counterintuitive design signal for the layout choice. Elevated/overhead AV conserves more cultivable land (8% loss) than near-ground interspaced rows (11% loss), because overhead posts sit on a sparse grid while interspaced rows need wide machinery-safety strips. So a community optimising for retained farmland leans overhead; one optimising for capital cost leans near-ground (the paper’s elevated system had ~1.8 €/Wp CAPEX and 176 €/MWh LCOE vs ~1 €/Wp and 88–117 €/MWh for the near-ground systems — cost figures noted for context, not a Neobiome cost input, which comes from the NZ cost source).

Reality-check on the OT_050 land-use-efficiency headline. OT_050 carries a theoretical land-use-efficiency gain of +60–80% (LER ~1.6–1.8). Reher’s measured arable LER is only 1.00–1.22 (and 0.97 in one wheat year), because it is a real open-field arable trial in a cool maritime climate with construction soil-compaction and light-limited shading — not the horticulture/optimum case behind the theoretical figure. This does not contradict OT_050 (different systems and metrics) but it bounds how far the +60–80% headline should be pushed for temperate-arable staples: treat ~1.6–1.8 as an upper-bound / best-case, not a planning expectation for arable NZ crops.

⚠ Transferable proxy (Belgium), no NZ value. Same handling as LIT_060 / LIT_107: the geometric land-loss figure transfers (footprint geometry is climate-independent; buffer widths follow machinery, not weather), but no NZ-specific recalibration is claimed. A measured NZ agri-PV land-loss figure would come from the NZ grazing-agrivoltaic cost/design source.

Research targets

Research gaps

• RT_108b (land-retention fraction) — PARTIALLY RESOLVED by this source. The agri-PV land-retention input now has a measured proxy: ~0.89 (near-ground vertical/tracked) to ~0.92 (elevated), bracketed with Interview II [INT_002] (~0.95) into a design-dependent ~0.89–0.95 range. RT_108b is addressed at the proxy tier; a measured NZ land-loss figure remains the only outstanding piece and is expected from the NZ grazing-agrivoltaic source, so no new RT is opened (RT_108 stays open as the umbrella until the parameters are wired into the model). See the shared RT_108 row.

Notes

• Authoritative publisher PDF. The raw is the genuine Applied Energy article (not an AI synthesis); it was located and downloaded by a retrieval pass. The load-bearing figures were re-verified verbatim against the downloaded PDF (pdftotext -layout), so the values are not merely AI-extracted.
• Why data_quality: medium despite verification. The medium rating reflects transferability, not extraction uncertainty: Belgium (temperate maritime) is a reasonable NZ analogue but not an NZ measurement. Same convention as the other transferable proxies in the corpus (LIT_060, LIT_107).
• Layout naming. The paper’s three layouts map to the agri-PV taxonomy on agro_pv: “Vertical” = vertical bifacial (near-ground / <2.1 m); “Tracking” = horizontal single-axis tracked (near-ground); “Elevated” = high-clearance overhead. The int_002 stow-tracker interspace system is a near-ground variant.
• Cost/LER figures are context only. CAPEX (~1 vs 1.8 €/Wp), LCOE (88–176 €/MWh) and LER (0.97–1.22) are recorded for completeness and the OT_050 reality-check; the Neobiome agri-PV cost input is sourced from the NZ (North Canterbury) grazing-agrivoltaic source, not from this Belgian paper.

Retrieval provenance

• Upstream source: Applied Energy 359:122679 (Elsevier / ScienceDirect), DOI 10.1016/j.apenergy.2024.122679 — the authoritative peer-reviewed article.
• Prepared by: retrieval pass (agri-PV model-integration retrieval run) on 2026-07-25, targeting the agri-PV land-retention fraction.
• Prompt / target: the exact retrieval prompt was not preserved verbatim; the recorded target was “retrieve the agri-PV land-retention (land-loss) fraction; source = Reher et al. 2024, Applied Energy 359:122679, Table 3 ‘Land Loss (%)’, reporting footprint loss separately from shading yield loss.”
• Verification: the retrieved file is the genuine publisher PDF, and every figure applied here was re-verified verbatim against that PDF (pdftotext -layout) at draft — so this exceeds the “AI-extracted, unverified” baseline. Held at data_quality: medium on transferability grounds (Belgium proxy, not NZ), not extraction uncertainty.

Connections

Links to

Sources (3): LIT_060 · LIT_107 · OT_050

Technologies (1): Agro-PV (Dual-Use Solar + Agriculture)

Referenced by

EDT domains (2): D01: Renewable Energy & Storage Systems · D02: Smart Food Systems & Agriculture

Technologies (1): Agro-PV (Dual-Use Solar + Agriculture)