ERA5-Land post-processed daily statistics from 1950 to present ( open, raster)
Daily gridded observational dataset for precipitation, temperature, E-OBS (open , gridded ),
| | Topography | [**National DEM Austria** ](https://www.data.gv.at/katalog/en/dataset/b5de6975-417b-4320-afdb-eb2a9e2a1dbf#additional-info)(10 m) | Derivation of LS-factor (slope, flow accumulation) | Copernicus DEM - Global and European Digital Elevation Model [(open – raster 30m, 10m for selected users)](https://dataspace.copernicus.eu/explore-data/data-collections/copernicus-contributing-missions/collections-description/COP-DEM) | | Land use and crop data | [**INVEKOS** ](https://www.data.gv.at/katalog/dataset?tags=INVEKOS)parcel dataset | Agricultural fields and crop information Input for C-factor and definition of arable exposure units | [CORINE Land Cover 2018 (vector/raster 100 m), Europe, 6-yearly](https://land.copernicus.eu/en/products/corine-land-cover/clc2018) | | Soil properties | [Austrian national soil maps](https://geometadatensuche.inspire.gv.at/metadatensuche/inspire/ita/catalog.search#/search?any=soil) | Input for K-factor (soil erodibility) | ESDAC datasets (500 m [K](https://esdac.jrc.ec.europa.eu/content/soil-erodibility-k-factor-high-resolution-dataset-europe)-[R](https://esdac.jrc.ec.europa.eu/content/rainfall-erosivity-european-union-and-switzerland) factor, 100 m [C](https://esdac.jrc.ec.europa.eu/content/cover-management-factor-c-factor-eu), and [P](https://esdac.jrc.ec.europa.eu/content/support-practices-factor-p-factor-eu)-factors) | | Agricultural structure and socioeconomics | Austrian [**Farm Structure Survey**](https://www.statistik.at/en/about-us/surveys/agriculture-and-forestry/farm-structure-survey) | Vulnerability indicators (farm structure, income dependency, demographics) at municipal level | [Eurostat Geospatial data from agricultural census](https://ec.europa.eu/eurostat/web/experimental-statistics/geospatial-data-agricultural-census) | Table 2– used data, an alternative dataset to replicate the assessment outside the study area, when available * ***Climate change effects*** > *The rainfall erosivity factor (R) is available from* [*ESDAC*](https://esdac.jrc.ec.europa.eu/content/rainfall-erosivity-european-union-and-switzerland) *as a baseline dataset for the EU and Switzerland, derived from high-resolution pluviography data and interpolated to a 500 m grid (Panagos et al., 2015). In addition to the baseline, ESDAC also provides future projections of R under climate change, calculated from bias-corrected regional climate model outputs (EURO-CORDEX). These projections indicate a general increase in rainfall erosivity towards 2050, especially in central and northern Europe, reflecting the expected intensification of extreme precipitation. Users can therefore choose between present-day R for baseline assessments and climate-adjusted R layers for scenario analyses in RUSLE applications.*  Figure 4 - Example layer showing reference parcels and digitized landscape elements defined by Agrarmarkt Austria under the EU Horizontal CAP Regulation, annually updated, enabling precise linkage of soil erosion modelling results with field-scale agricultural units.  Figure 5 - large-scale assessments, vulnerability proxies are available from Eurostat’s experimental geospatial agricultural census datasets (multi-resolution grids, 1–80 km), providing spatially harmonised indicators on farm number, cultivated areas, farmer demographics (age, gender), livestock, labour, and production methods. While coarser in scale, these can serve as vulnerability layers when regional data are unavailable. The soil erosion workflow is implemented using a **RUSLE-based modelling approach**, adapted to the regional scale of Lower Austria. The model itself is not a standalone application but is coded and executed within a spatial analysis environment. **RUSLE is essentially a multiplicative model: once all factor layers (R, K, LS, C, P) are prepared,** the application reduces to a raster-based calculation in GIS, as will be detailed later in the methodological section. To handle geospatial data, preprocess inputs, and visualize outputs, a **GIS platform such as QGIS** is used. This provides the necessary functionality to manage raster and vector datasets, derive topographic parameters from the DEM, and integrate climate and land-use data. The combination of RUSLE calculations with GIS processing ensures that erosion risk can be **mapped consistently** across agricultural areas and aggregated at municipal and state levels for decision support. | **Tool** | **Type** | **Role** | | -------------------------------------------- | ---------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | RUSLE (Revised Universal Soil Loss Equation) | Open / empirical model |Core model to estimate average annual soil loss under current and future climate scenarios.
(see note)
| | [QGIS](https://qgis.org/) | Open-source GIS | Data preprocessing, derivation of LS-factor (directly through SAGA GIS tools), conversion of land use and soil/parcel data into necessary RUSLE factors mapping and visualization of erosion risk. | Table 3 – used tools and role in the Erosion Hazard workflow, all tools are free to use. * **Note on deriving R K, C and P factors.** > *The calculation of soil erodibility (K), cover/management (C) and support practices (P) requires **reclassification of local datasets** (soil maps, crop parcels, or land management layers) using standard **lookup tables**. In the Austrian case, this could be done following the methodology of* [\*\*\* Schmaltz et al. (2023)\*\*\*](https://www.sciencedirect.com/science/article/pii/S0167880923002499)*, where K is assigned to soil units, and C and P are linked to INVEKOS crop and management information.* > > *For replication in regions where only European data are available, **users may apply the pre-computed ESDAC layers (R, K at 500 m; C, P at 100 m)**. However, these products are relatively coarse. To increase accuracy, user may invest time in deriving* [*R*](https://www.sciencedirect.com/science/article/pii/S004896971500011X) [*K*](https://www.sciencedirect.com/science/article/pii/S0048969714001727)*,* [*C*](https://www.sciencedirect.com/science/article/pii/S0264837715001611) *and* [*P*](https://www.sciencedirect.com/science/article/pii/S1462901115000611) *from more detailed meteorological , spoil or parcel datasets when available, using the lookup values provided in in the ref. article from the European Soil Data Centre.* ### Methodology #### Step 1 - Data acquisition and preparation The first step of the workflow is the collection and preprocessing of all input datasets required to parameterize the RUSLE model. **When local and national datasets are available, each factor must be recalculated specifically for the study area.** * Topography (LS-factor): **directly derived from the DEM** using the LS-factor function of SAGA GIS within QGIS. This ensures a spatially explicit representation of slope length and steepness adapted to local terrain conditions. * Rainfall erosivity (R-factor), soil erodibility (K-factor), cover-management (C-factor), and support practice (P-factor): require **reclassification of local data sources.** Values are assigned using lookup tables based on soil maps, parcel and crop information, and national climate data, as described in the previous section. Preprocessing: includes harmonizing coordinate systems, resampling datasets to a common resolution, and rasterizing vector inputs such as field parcels or management practices. As an alternative, where local data are not available, users may rely on the precomputed ESDAC products (R at 500 m, K/C/P at 100 m). These datasets are coarser in resolution but provide a consistent and harmonised European baseline for exploratory assessments and large-scale applications, as illustrated in the example maps provided below.  Figure 6 – example of 10 m Lidar derived DTM from Austrian geoportal(left) and corresponding Ls factor derived in Qgis (right)  Figure 7– examples of Rainfall Erosivity R \[MJ mm ha-1 h-1 yr-1] \[ (left) and Soil Erodibility (K- Factor, right) from ESACD databases. #### Step 2 - Model setup and run. The modelling step corresponds to the direct application of the RUSLE formula, i.e. the **multiplication of all prepared factor layers** via basic *raster calculator* (*map algebra*) in GIS: E=R×K×LS×C×P where E is the estimated average annual soil loss (“Erosion”) in **t ha⁻¹ yr⁻¹**, see also (\[1]) Each factor layer (R, K, LS, C, P) had been previously (Step 2) derived or reclassified from local datasets, harmonised to the same projection and resolution. The resulting erosion map provides the **annual mean soil loss rate**, which represents the standard output unit for RUSLE applications. For methodological details and reference values of the factors, users cloud also consult the following key sources: * [**R-factor**](https://www.sciencedirect.com/science/article/pii/S004896971500011X) **(rainfall erosivity):** Panagos et al., 2015 (*Science of the Total Environment*) – ESDAC dataset for Europe. * [**K-factor**](https://www.sciencedirect.com/science/article/pii/S0048969714001727) **(soil erodibility):** Panagos et al., 2014 (*Geoderma*) – European soil erodibility map. * **LS-factor (topography):** [SAGA-GIS Tool Library](https://saga-gis.sourceforge.io/saga_tool_doc/8.1.1/ta_hydrology_25.html) * [*C*](https://www.sciencedirect.com/science/article/pii/S0264837715001611)**- and** [*P*](https://www.sciencedirect.com/science/article/pii/S1462901115000611)**-factors (cover and practices):** Panagos et al., 2015 (*Environmental Modelling & Software*) – European C and P factor datasets. For simplicity, the soil erosion map (E \[t ha⁻¹ yr⁻¹]) presented below (Figure 8) is derived directly from the [**RUSLE2015 soil erosion dataset**](https://esdac.jrc.ec.europa.eu/content/soil-erosion-water-rusle2015/) published by ESDAC, following the above described methodology #### Step 3 - Analysis and interpretation The erosion hazard map derived from the RUSLE model provides the core spatial output, but its value lies in how the results are analysed and interpreted. In Innovation Lab 1, this step focuses on transforming raw model outputs into information that can support planning and decision-making. First, the raster of mean annual soil loss (*t ha⁻¹ yr⁻¹*) can be **aggregated from small to larger territorial unit ( e.g. municipal and state boundaries** (as in the example of Figure 8), allowing direct comparison across relevant administrative units and integration into existing planning frameworks. This allows local authorities to identify where erosion risk is systematically higher within their jurisdiction. Second, the analysis highlights **erosion hotspots** by applying thresholds to the raster map and overlaying them with cropland exposure. These hotspots indicate priority areas for targeted interventions, including the implementation of multifunctional hedgerows and other Nature-based Solutions. Third, **scenario comparisons** are undertaken. Results can be analysed **under different climate projections** (EURO-CORDEX, ÖKS15) to quantify the expected increase in hazard due to climate change. Finally, the interpretation integrates **socio-economic vulnerability indicators** **at municipal level**, such as farm structure and dependence on agricultural income. This combination of hazard, exposure, and vulnerability highlights areas where physical risk coincides with lower adaptive capacity, providing a risk-oriented basis for prioritisation.    Figure 8– Left: extract from the RUSLE2015 soil erosion dataset (ESDAC), showing estimated average annual soil loss (t ha⁻¹ yr⁻¹) for a selected area in Austria, derived from the harmonised European layers of R, K, LS, C and P factors. Right: overlay of the same erosion map with INVEKOS reference parcels (Agrarmarkt Austria, 2025-1, Table 2), which represent agricultural blocks and digitised landscape elements under the EU CAP regulation. This integration allows soil erosion estimates to be directly linked to field-scale agricultural units. Aggregation example over larger territorial unit is provided in the lower figure. #### Step 4 - NbS testing. The testing of Nature-based Solutions (NbS) focuses on the introduction of **multifunctional hedgerows** in arable land as a measure to mitigate soil erosion. Within the RUSLE framework, the presence of hedgerows does not require changes to the hazard drivers (R, K) but can be represented through adjustments of the **LS- and C-factors**. * **C-factor (cover and management):** hedgerows increase permanent vegetative cover, reduce raindrop impact and promote soil stability (\[2],\[3]). This effect is simulated by assigning lower C values to cropland areas adjacent to or intersected by hedgerows. * **LS-factor (slope length and steepness):** the establishment of hedgerows effectively interrupts slope length and alters runoff pathways, reducing the effective contributing area upslope of a pixel. In GIS terms this can be simulated by shortening slope lengths or by recalculating LS with breaks introduced along hedgerow alignments **modifying DTM to count for new morphology induced by NBS before reprocessing.** The workflow in steps 2 and 3 shall then be repeated with adjusted LS and C values representing hedgerow implementation. The comparison of “with” and “without NbS” maps quantify the potential reduction of annual soil loss, highlighting those hotspots where hedgerows deliver the highest benefit. **Integration of hedges as a nature-based solution for erosion mitigation in the RUSLE model (L-factor)** The objective of this workflow was to represent hedgerows as a nature-based solution within the RUSLE framework by explicitly modifying the slope length (L) factor through terrain- and flow-based analysis. The approach combines terrain analysis, rule-based hedge placement, and hydrological enforcement to quantify how hedges disrupt downslope runoff connectivity. First, a 10 m resolution digital elevation model (DEM) was hydrologically conditioned by filling depressions to ensure continuous flow paths. Specific contributing area (SCA) was used as a proxy for the RUSLE L-factor, which was used to prioritise placement of hedges based on the upslope contributing area (L-factor), thus ensuring that hedges are preferentially placed where erosion risk is highest. Potential hedge locations were generated from contour lines extracted from the DEM, constrained to agriculturally relevant areas (intersecting with arable paddocks), elevation limits (<600 m), and slopes exceeding a defined threshold (>4°). Contour segments were split into individual line features. Additional filtering enforced minimum hedge length (500 m) and spatial thinning rules to avoid unrealistically dense hedge networks. Two scenarios were produced by varying slope (4°, 5°), L-factor (>40, >100), and spacing thresholds (100 m, 150 m). To account for the hydrological effect of hedges, selected hedge lines were rasterised and “burned” into the DEM as elevated linear barriers. This modified DEM was subsequently filled again to remove artefacts and used to recompute flow accumulation and SCA. The resulting hedge-adjusted SCA raster represents a reduced effective slope length downslope of hedges and was directly integrated as an adapted L-factor in RUSLE for the two hedge scenarios. Overall, the method provides a spatially explicit and process-based representation of hedges as runoff-interrupting features, allowing erosion mitigation effects of hedge scenarios to be quantified consistently within the RUSLE modelling framework. ## Hazard – Agricultural drought. ### Description and context The second hazard assessed in Innovation Lab 1 is **agricultural drought**, understood here as the insufficient availability of soil water to sustain crop growth during the vegetation period. In the context of Lower Austria, drought has become an increasingly relevant climate risk alongside water-induced soil erosion. While the region is traditionally well supplied with rainfall, recent summers have shown more frequent dry spells and heat periods, affecting yields and farm income. Within the CRA, drought is approached through an **indicator-based method** that combines climate variables (precipitation and temperature) with soil and land use characteristics to estimate water deficit conditions in cropland. Climate change scenarios (EURO-CORDEX, ÖKS15) are used to explore how frequency and severity of droughts may evolve under future conditions. The objective is to provide **spatially explicit drought hazard maps**, comparable with the erosion analysis, and to identify where agricultural areas are most exposed to prolonged soil moisture deficits. These results allow stakeholders to evaluate the relative importance of “too little water” in comparison to “too much water” and to consider adaptation options at both municipal and regional planning levels. | | | | | | ------------- | ------------------------------------------------------------------------------ | ------------ | ------------------------------------------------------------------ | | **Dimension** | **Indicator(s)** | **Unit** | **Purpose** | | Hazard | Potential crop yield loss in rainfed conditions (derived from ETₐ / ETₙ ratio) | % yield loss | Quantify drought stress on crops under baseline and future climate | *Table 4 – key indicators tracked-Agricultural Drought Hazard* ### Data sources The assessment of agricultural drought in Innovation Lab 1 is explicitly based on the [**CLIMAAX Risk Workflow for agricultural systems**](https://handbook.climaax.eu/notebooks/workflows/DROUGHTS/02_agriculture_drought/AGRICULTURE_Risk_workflow_description.html). By design this CRA methodology follows an **indicator-based approach** that combines climate drivers, soil water holding capacity, crop exposure, and vulnerability information to derive **potential yield losses under rainfed conditions**. In this section we therefore focus only on the **datasets required to apply the workflow**, both at Austrian and European scale. The full methodological description is not repeated here but is entirely referred to the official [**CLIMAAX Risk Workflow for agricultural systems**](https://handbook.climaax.eu/notebooks/workflows/DROUGHTS/02_agriculture_drought/AGRICULTURE_Risk_workflow_description.html). For the regional application in Lower Austria, the CRA uses **ÖKS15 climate scenarios** (1 km), **EURO-CORDEX simulations** (12.5 km), and national meteorological observations as the basis for calculating reference evapotranspiration (ET₀) and effective evapotranspiration (ETₐ). Soil water holding capacity is derived from **national soil maps**, while exposure is defined through the **INVEKOS parcel dataset** with crop-specific attributes. **National agricultural statistics** provide information on crop production and economic value, while irrigation distribution and socio-economic data from the **Farm Structure Survey** represent the main vulnerability layers. Elevation and climate zoning are also included to reflect topographic and thermal conditions influencing crop growth. To ensure replicability outside Austria, equivalent **European open datasets** can be used, such as **Copernicus CDS** (precipitation, temperature, evapotranspiration indices), **ESDAC** (soil hydraulic properties), **MapSPAM/FAO-IIASA GAEZ** (crop maps, yields, irrigation, climate zones), the **Copernicus DEM**, and **Eurostat agricultural census data**. This dual setup allows both accurate local analysis and transferability to other regions. | **Data type** | **Source (Austria)** | **Role in workflow** | **Open/EU alternative** | | ------------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | Climate projections (precipitation, temperature) | [**ÖKS15** ](https://data.hub.geosphere.at/group/oks15)(Austrian scenarios, 1 km), | Input for ET₀, ETₙ, ETₐ and drought hazard calculation | Copernicus [**EURO-CORDEX** ](https://cds.climate.copernicus.eu/datasets/projections-cordex-domains-single-levels?tab=overview)(12.5 km /0.11°) | | Elevation | [**National DEM Austria** ](https://www.data.gv.at/katalog/en/dataset/b5de6975-417b-4320-afdb-eb2a9e2a1dbf#additional-info)(10 m) | Derivation of topographic effects and climate zoning | Copernicus DEM - Global and European Digital Elevation Model [(open – raster 30m, 10m for selected users)](https://dataspace.copernicus.eu/explore-data/data-collections/copernicus-contributing-missions/collections-description/COP-DEM) | | Thermal climate zones | *Global dataset used* | Crop-specific growing conditions and evapotranspiration |FAO-IIASA GAEZ climate zones
Thermal climate zones of the world (FGGD)
| | Soil available water capacity (AWC) | *Global dataset used* | Soil buffering capacity against drought | [Hengl and Gupta (2019)](https://zenodo.org/records/2629149) | | Crop distribution | [INVEKOS](https://www.data.gv.at/katalog/dataset?tags=INVEKOS) parcel dataset | Definition of crop types, derive Crop production map for Austria | [CORINE Land Cover 2018 (vector/raster 100 m), Europe, 6-yearly](https://land.copernicus.eu/en/products/corine-land-cover/clc2018) | | crop-specific information | *Global dataset used* | specific parameters needed for the hazard assessment | [Climaxx Handbook](https://handbook.climaax.eu/notebooks/workflows/DROUGHTS/02_agriculture_drought/crop_table.html) | | Crop production \[ton] | *Global dataset used* | crop production to calculate exposure (first exposure dataset) | [MapSPAM repository](https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/SWPENT) | | crops aggregated value | *Global dataset used* | second exposure dataset | FAO-GAEZ [Global Agro-Ecological Zones](https://www.fao.org/gaez/en/) | | Irrigation availability | [Austrian farm structure survey](https://www.statistik.at/en/statistics/agriculture-and-forestry) | cropland full-irrigation availability to define vulnerability | GAEZ v5 [Share of irrigated land](https://data.apps.fao.org/catalog/iso/5e11e7c5-9088-4f1a-80df-04ef026bd726) | Table 4 – used data, an alternative dataset to replicate the assessment outside the study area, when available ### Methodology (addition) As mentioned above the CLIMAAX agricultural drought workflow served as a basis for this modelling. For the general methodology refer to the CLIMAAX handbook. The workflow was modified so that instead of a periodic daily average of the climatic parameters, daily resolved parameters were used to calculate ET0. This should ensure that particularly dry years are not averaged out too much. This results in annual average yield loss values for the studied periods and scenarios (Historical 1971-2000; RCP8.5 2041-2070; RCP4.5 2071-2100; RCP8.5 2071-2100). The yield loss calculation was carried out for eight different crops, which are of great importance in agricultural production in Lower Austria. In total, an ensemble of six EURO-CORDEX models was used for each scenario and period described.  Figure 9 - example boxplot for resulting mean yield loss values per crop across the studied climate scenarios. ### NbS testing **Hedge scenarios for wind-effect modelling for agricultural drought.** *Identifying the prevailing wind direction for optimal hedge placement* The first step involves determining the prevailing wind direction to place hedges at right angles to this direction. For the case of the Wagram Lab, long-term [annual wind data](https://doi.org/10.60669/6yna-mw14) (1991–2024) from two meteorological stations (Langenlebarn and Stockerau) were analysed. Daily occurrences of wind direction were aggregated by cardinal and intercardinal directions (e.g., N, NE, E) and used to identify the dominant wind direction relevant for hedge effects in the context of agricultural drought. Based on the prevailing wind direction across the two stations (Langenlebarn: southwest, Stockerau: west), the dominant wind direction of 247.5° (WSW) was selected as most relevant for the study area (Figure 10). Hedges with an orientation of SSE-NNW were therefore considered as most optimal to reduce the effect of wind on evapotranspiration in the context of agricultural drought.  Figure 10 - Occurrences of wind direction (%) based on weather station measurements at Langenlebarn and Stockerau (1991-2024). *Extraction of wind-perpendicular hedge lines from paddock boundaries* The second step involves generating hedge geometries directly from agricultural paddock boundaries to represent plausible hedge line scenarios. In this example, the land parcels from the Integrated Administration and Control System (IACS or InVeKoS) are used. Parcels classified as arable land were first converted into boundary line segments. These boundaries were split and merged based on changes in segment orientation (tolerance = 45°) to obtain continuous line segments with generally consistent azimuths. Segment azimuths were then compared against directions perpendicular to the prevailing wind direction (±90°), allowing for a specified angular tolerance (30° or 45° for different scenarios). From these candidate segments, only those located on the western side of paddocks (i.e. facing the prevailing wind) were retained. Additional filtering was applied to ensure that, for narrow paddocks, only the most representative boundary segment, i.e. the longest, was kept. Finally, hedge candidates were filtered by minimum length (40 m) and spatially thinned using a band-based approach to enforce a minimum spacing between hedges (150 m or 300 m for different scenarios). Within each north–south band, overlapping or closely spaced candidates were resolved by retaining the longer segment. The resulting hedge lines form the spatial input for the four hedge scenarios used in subsequent analyses. The four hedge scenarios are given by defining two angular tolerances for hedge orientation relative to wind direction (±30° and ±45°) with two minimum spacing constraints between hedges (150 m and 300 m). *Spatial modelling of wind shelter effects of hedges* The third and final step involves the approximation of hedge effects on evapotranspiration. Change factors for potential evapotranspiration (ET0) were derived to represent the sheltering effect of hedgerows on near-surface microclimatic conditions in the lee of the hedge. Following the empirical findings of Orfánus and Eitzinger (2010) – a study also undertaken in Lower Austria, ET0 was assumed to increase with distance from the hedge, reflecting a gradual reduction in wind protection and shading effects. Observed ET0 values at different distances (8 m, 20 m, and 80 m) under both low- and high-demand atmospheric conditions were normalised relative to ET0 at 80 m, which was defined as the reference condition (change factor = 1.0). Across demand regimes (low vs. high), this resulted in relative change factors of approximately 0.4–0.5 close to the hedge and values approaching unity at 80 m distance. Based for each hedge scenario, as well as the actual hedges planted by the Agrarbezirksbehörde (ABB), a raster-based change factor was calculated to represent wind shelter effects on evapotranspiration in the lee of hedges. For each raster cell (10 m), the shortest distance to the nearest hedge was computed, and cells located downwind of hedges were identified using angular relationships between wind direction and hedge orientation. A linear distance-decay function was applied to a factor of change, with maximum shelter at the hedge (0.5) and a gradual reduction to no effect beyond a defined distance of 80 m. The result is a spatially explicit change-factor raster that can be used to modify evapotranspiration in subsequent modelling steps (Figure 11).  Figure 11 - Example of a hedge scenario with change factors for evapotranspiration in the Wagram Lab. Given the site specifics of the field measurements undertaken by Orfánus and Eitzinger (2010), this approach assumes an average hedgerow height of 8 m, consistent shelter effects across soil types, a uniform prevailing wind direction, and a maximum effective shelter distance of 80 m despite potential site-specific variability. A key limitation is the implicit assumption that yield responses are primarily driven by soil moisture availability during dry summer conditions. While this assumption is appropriate for drought years, it likely overestimates the positive effects of reduced ET0 under average climatic conditions, as it does not explicitly account for potential negative effects related to reduced photosynthetically active radiation (PAR) or below-ground competition. Tree and hedge effects are expected to be beneficial under very dry conditions but neutral or negative under normal conditions (Koch et al., 2024; Majaura et al., 2025; Reuse & Langhof, 2025). This limitation is partly mitigated within the CLIMAAX framework, as the relative importance of ET increases under dry conditions, thereby reducing the influence of shelter effects in wetter or average years. **References** > Majaura, M., Sutterlütti, R., Böhm, C., Freese, D., 2025. High and dry: Barley (Hordeum vulgare) yield benefits from tree presence in a temperate alley cropping system during a drought year. Agroforest Syst 99, 166.