Conference Agenda

Effective ventilation assessment in naturally ventilated livestock buildings is essential for reducing environmental impact and improving emission modelling accuracy. This work presents the emerging CISPO (CFD-Informed Sensor Placement Optimization) framework, which integrates high-fidelity numerical simulations, wind-tunnel experiments, and data-driven analysis to identify optimal sensor locations for characterizing ventilation performance. The methodological foundations of CISPO are on the basis of sentinel-grid construction, information-based ranking, and cross-validation using boundary-layer wind-tunnel measurements. Our recent work expands the framework through a comprehensive parametric CFD study covering considerable wind directions and wind speeds combinations to capture the breadth of ventilation regimes observed in practice.

The resulting large-scale CFD dataset provides detailed spatial information on airflow structures, local mean age of air, and exchange rates. Beyond supporting sensor-placement optimization, these data highlight the potential of massive CFD ensembles to reveal essential ventilation patterns that drive pollutant dispersion and emission rates. By combining multi-source modelling, systematic parameterization, and scalable analysis tools, CISPO offers a robust pathway for improving environmental monitoring strategies and enhancing emission modelling capabilities in livestock systems. This contribution outlines the framework, key findings, and future applications toward climate-relevant environmental engineering.

Ammonia (NH₃) emissions from livestock manure management are a major precursor of secondary fine particulate matter and an important source of odor pollution. Conventional emission factors are usually estimated using bottom-up approaches based on ventilation rates and NH₃ concentrations measured at individual farms. However, these methods have limited representativeness across different facility types and management practices and are subject to uncertainty due to complex ventilation systems. This study proposes a top-down approach to estimate NH₃ emission factors using the tracer gas dispersion method combined with inverse modeling. Average emissions from multiple farms were estimated by integrating field observations with three atmospheric dispersion models: AERMOD, CALPUFF, and computational fluid dynamics (CFD). The method was applied to a livestock-dense area in Gimje, South Korea, including 40 pig farms within a 4.5 × 2.8 km² domain. Model results were evaluated using two months of NH₃ measurements. AERMOD produced simplified Gaussian plume patterns, while CALPUFF simulated heterogeneous distributions with localized peaks under variable meteorological conditions. CFD captured detailed plume structures influenced by terrain and buildings. The estimated emission factors ranged from 4.81 to 5.64 kg·animal-1·yr-1, slightly lower than national inventory values, indicating that top-down approaches can complement bottom-up methods to improve emission inventory reliability.

Heat stress is a major constraint on animal welfare and production efficiency in intensive livestock systems, particularly for broilers and pigs with limited thermoregulatory capacity. The temperature–humidity index (THI) is commonly used to indicate thermal stress, but conventional monitoring methods are often limited in spatial coverage and scalability. This study proposes a satellite-based machine learning framework to estimate indoor heat stress in livestock housing using GEO-KOMPSAT-2A data. Based on an “outside-in” concept, satellite-derived temperature, humidity, and solar radiation variables are used to infer indoor microclimates without requiring in-house sensors or building structural information. Unlike computational fluid dynamics approaches, which are computationally intensive and difficult to deploy at scale, the proposed data-driven model captures nonlinear relationships between outdoor weather conditions and indoor environments. The XGBoost algorithm demonstrated strong predictive performance for indoor temperature and humidity across multiple commercial farms. When expressed as THI, relative RMSE ranged from 0.623% to 0.693% in swine facilities and from 0.827% to 1.332% in broiler houses, indicating reliable heat stress estimation. Leveraging the high spatial and temporal resolution of geostationary satellite data, this approach enables continuous, wide-area monitoring to support ventilation planning, cooling strategies, and welfare-oriented livestock management under variable climatic conditions.

Recurring heatwaves increase heat-stress risk in broiler production, reducing feed intake, suppressing growth, delaying market age, and causing economic losses. Although the temperature–humidity index (THI) is widely used, it mainly reflects instantaneous indoor conditions and does not capture cumulative physiological burden, recovery, or lagged growth responses under repeated heat exposure. This study proposes a system dynamics (SD) modeling framework to quantify cumulative heat-stress load and its propagation to productivity using 5-min indoor temperature and humidity data. A causal loop diagram (CLD) represented feedbacks among indoor climate, THI exceedance, stress load, feed intake, daily gain, body weight, and management actions (ventilation and cooling). The SD model computed THI exceedance above a threshold (ΔTHI) and updated a stress-load stock S (THI units) with first-order recovery (time constant τ, h). During hot periods, outputs showed THI ≥75 for ~6–10 h/day with peak THI ~84–88, cumulative heat dose ~25–45 THI·h/day, and stress load peaking at ~6–12 THI with τ ~8–14 h. These conditions corresponded to ~5–10% lower feed intake and ~7–15% lower daily gain, implying ~1–3 days of market-age delay (or equivalent body-weight shortfall). Model fit was BW RMSE ~50–100 g and MAPE ~3–6%.

Heat stress is a major factor affecting the health, welfare, and zootechnical performance of palmipeds. Climate change is expected to increase the frequency and intensity of heat waves, further exacerbating this issue. In livestock production, thermal comfort indicators such as the Temperature–Humidity Index (THI) are commonly used to assess and anticipate heat stress. While THI formulations exist for several poultry species, no index has been specifically developed for palmipeds.

This study aimed to develop a THI adapted to mule ducks, a key species in foie gras production in South-West France. Zootechnical performance data from 2,297 flocks raised during the growing stage between late spring and early autumn were analyzed. Hourly THI values were used to characterize heat stress exposure at both daily and flock levels, allowing performance to be compared across stress categories.

Results indicate that mule ducks exhibit slightly greater tolerance to high THI levels than Pekin ducks, whose thermal sensitivity is considered similar to that of broilers. This species-specific THI provides a new tool to better estimate short-term heatwave risks at farm level due to acute heat stress and to assess long-term climate risks at the regional scale.

In long-distance cattle transport, the vehicle’s thermal environment may represent a critical challenge contributing to heat stress. To characterize the thermal environment, eight field measurements were conducted during six commercial journeys transporting cattle in the two-deck vehicle Pezzaioli IPB86. Internal temperature and humidity were recorded to validate a Computational Fluid Dynamics (CFD) model, integrated with a cattle thermoregulation model. This simulation accounts for the influence of dynamic animal heat loss, water loss and external weather factors on the internal thermal environment of the vehicle. Based on experiments and simulations, the following conclusions were reached: (1) CFD simulations showed strong agreement with field data, maintaining relative errors below 10%. (2) The trailer microclimate showed high spatial heterogeneity. Using the Coefficient of Variation (CV) to quantify this, the average CV for temperature and humidity differences reached 39.21% and 41.43%, respectively. (3) The front compartments consistently maintained lower temperature and humidity levels than rear sections. Simulations demonstrated that, in this vehicle geometry, the front compartments have better ventilation performance during transit. These findings reveal that cattle in the rear compartments face significantly higher heat stress risks. This research provides a validated framework for predicting thermal environment within the cattle vehicle during commercial transport.

UAV hyperspectral imaging scales field phenotyping for rice heat tolerance, but canopy spectra conflate genotype, development, and environment, limiting direct seed set modeling under natural heat. We present a heat-stress-aware, phenology-aligned framework, RiceHeatMamba, that converts UAV hyperspectral time series into a dynamic endophenotype. Heat exposure is derived from hourly air temperature as a heat stress integral to detect episodes and define matched pre- and post-windows. To accommodate genotype-specific phenological offsets, plot trajectories are aligned by developmental stage, such as days after heading or flowering, enabling comparisons under comparable biological states. Heat response is modeled as reaction and recovery, capturing spectral disruption near peak heat and the rebound and residual deviation afterward. Event-level scores are aggregated into season-level tolerance rankings. RiceHeatMamba uses an efficient state-space sequence model conditioned on exposure and phenological context to learn long-range structure in event-aligned trajectories. In a large within-season seed set dataset with thousands of plot observations, high versus low seed set identification achieves 78.7% precision for high seed set and 77.0% for low seed set. This framework yields an interpretable, actionable endophenotype for identifying susceptible tails and supporting downstream GWAS and genomic selection.