Andre Daccache

The benefits of agrivoltaics (AV) are documented, with empirical evidence for their suitability for collocated energy generation, crop cultivation, and water savings, even for staple crops. What is lacking is a holistic modeling framework to design and optimize AV systems for specific crop types, micro‐environments, and panel configurations. To this end, a multiphysics model that simultaneously solves momentum, energy, and mass exchanges among interacting components, and determines the implications of AV on photosynthesis, water use, power generation, and worker thermal stress is introduced. The model is tested against field data from California and Minnesota. Its applications are illustrated for a representative AV scenario using weather data from central New Jersey, showcasing the multifaceted benefits of AV for a tomato farm under photovoltaic panels. In this scenario, the model predicts a 5.6 reduction in average daytime panel temperature (compared to a common photovoltaic installation), recovering 15% of the thermal energy losses. Despite a 47% reduction in light for the setup modeled here, net carbon assimilation declined by 31%. Daytime crop leaf temperature cooled by 1.84, and these reductions offset some of the decline in carbon assimilation. Another benefit is the reduction in transpiration losses by 22.4% compared to an open crop field. Finally, the average perceived temperature experienced by workers was reduced by approximately 4.46 during working hours. While these are illustrative results for a particular crop and climate, they demonstrate how the model enables integrated assessment and optimization of AV benefits across varying climates, crops, and photovoltaic technologies.

High-resolution surface soil moisture is important for precision agriculture, disaster management, and hydrological modeling. However, existing satellite-derived products are often limited by coarse spatial resolution, restricted regional coverage, or computationally demanding physical modeling frameworks. To address these limitations, we present OASSM-10 (On-demand AI-driven Surface Soil Moisture at 10 m), an on-demand 10 m surface soil moisture dataset generated using a modality-aware Transformer. The proposed architecture uses a cross-attention mechanism to integrate dynamic multi-sensor observations, including Sentinel-1 Synthetic Aperture Radar (SAR) and Sentinel-2/Landsat optical-thermal data, with static environmental covariates such as topography, soil texture, climate zones, and land cover. Evaluated against the International Soil Moisture Network (ISMN), OASSM-10 showed stable predictive performance on an independent test set, with an RMSE of 0.0880 ± 0.0007 m³/m³, ubRMSE of 0.0878 ± 0.0006 m³/m³, Bias of -0.0066 ± 0.0015 m³/m³, and R² of 0.3889 ± 0.0098. The modality-aware Transformer also reduced the conditional bias commonly observed in standard deep learning regressions, allowing the predictive distribution to retain a broader and more realistic range of soil moisture conditions. Comparative assessments against established products, including the 30 m SMAP-HydroBlocks and the field-scale S2MP, indicate that OASSM-10 provides competitive performance in capturing high-resolution moisture dynamics. The spatial detail and practical value of OASSM-10 are illustrated through applications in precision agriculture, drought monitoring, and flood assessment. To avoid the storage and distribution challenges of a static wall-to-wall global 10 m product, OASSM-10 is delivered through an open-source interactive web tool (https://gssm-10.com/) that allows users to generate and export localized high-resolution soil moisture maps for user-specified coordinates and dates. This design supports scalable, on-demand soil moisture mapping that is globally deployable from 2016 to the present.

Irrigation is essential for stabilizing yields and sustaining food production, yet it faces increasing challenges, especially in water‐scarce regions. Cosmic‐ray neutron sensors (CRNS) are a promising tool in irrigation management as they can continuously monitor soil water content (SWC) over large areas. However, observational studies sometimes reported inconsistent CRNS responses across different irrigation methods and environments, limiting the broader transferability of this technique. In this study, CRNS measurements in hose reel raingun (potato), central pivot (winter wheat), and flood irrigation (alfalfa) were evaluated, and their interpretation was aided by neutron transport modeling. Overall, CRNS proved more consistent at estimating field‐scale SWC than point‐scale and profile sensors. Also, replacing the conventional N0 method with the Universal transfer solution approach substantially improved SWC estimates when this was below 0.10 cm ³ cm ⁻³ , avoiding unrealistically low values. In flood and pivot irrigation, CRNS showed clear responses to irrigation and neutron transport simulations with the Monte Carlo code “URANOS” reproduced these dynamics well. In contrast, hose reel raingun irrigation produced weaker responses that were less consistent with URANOS, especially when irrigation occurred far from the CRNS, likely due to the co‐occurrence of precipitation and the small soil volume affected by irrigation. Interestingly, a simplified analytical solution produced results comparable to URANOS, suggesting potential for rapid applications. Overall, the combination of CRNS measurements and neutron transport modeling proved key in interpreting CRNS observations across irrigation systems. Increasingly integrating observations, modeling, and recent methodological advances will likely make CRNS more transferrable, strengthening its role and adoption in irrigation management.

Accurate estimation and forecasting of reference evapotranspiration (ETo) is fundamental for irrigation scheduling and agricultural water management, particularly in regions facing growing hydrologic stress. This study integrates multiple observational and modeling datasets—including high-quality eddy covariance measurements over four well-watered alfalfa sites, California Irrigation Management Information System (CIMIS) weather-station records, and numerical weather prediction outputs from the Global Forecast System (GFS)—to evaluate widely used ETo estimation and forecasting methods across California’s diverse climatic regions. The ETo estimation and forecasting methods considered are Penman–Monteith (PM), the operational CIMIS formulation, Priestley–Taylor (PT), and Slob–de Bruin (SdB), and calibrated versions of both PT and SdB. Results show that while all methods capture seasonal ETo dynamics, PM and CIMIS systematically overestimate ETo in hot, dry regions, and SdB consistently underestimates it. In contrast, the PT method reproduces observed ETo with the smallest bias (≈1–6%) and lowest RMSE (≈7–14%) across all sites. When driven by GFS forecasts, PT retains its better performance, producing stable accuracy across 1–7-day lead times (RMSE ≈ 9–19%; bias within ±3%). In contrast, PM errors amplify sharply under forecast forcing, with RMSE reaching 60% in arid regions due to its sensitivity to forecast uncertainties in relative humidity and wind speed. Sensitivity analyses confirm that PT’s reduced input requirements render it less vulnerable to meteorological forecast errors than PM. A statewide comparison across CIMIS stations further shows that PM and CIMIS substantially overestimate long-term ETo in inland and southern California, whereas SdB underestimates moderately relative to PT. Local calibration of PT and SdB using limited datasets did not yield meaningful improvements over their standard parameter values for forecasting purposes. Collectively, these findings demonstrate that the standard Priestley–Taylor formulation (α = 1.26) provides relatively more accurate, robust, and operationally reliable approach for both estimating and forecasting ETo across California’s climatic gradients. This has important implications for irrigation advisory systems, water-use planning, and statewide strategies to improve agricultural water-use efficiency under limited water availability.

The societal benefits of agrivoltaics are well supported, with ample empirical evidence for their suitability for collocated energy generation, crop cultivation, and water savings, even for staple crops. What is lacking is a holistic modeling framework to design and optimize AV systems for specific crop types, micro-environments, and panel configurations. To this end, a multiphysics model that solves momentum, energy, and mass exchanges among interacting components, and determines the implications of AV on photosynthesis, power generation, and worker thermal comfort, is introduced. The model is tested against field data from California, and Minnesota. Its applications are illustrated for a representative AV scenario using weather data from central New Jersey, showcasing the multifaceted benefits of AV for a tomato farm under photovoltaic panels. In this scenario, the model predicts a 5.6 °C reduction in average daytime panel temperature (compared to a regular PV installation), reducing overheating-related energy losses by 15.2%. Despite a 47% reduction in light for the setup modeled here, net carbon assimilation declined only by 31%. Daytime crop leaf temperature declined by 1.84 °C, and these reductions offset some of the decline in carbon assimilation. Another benefit is the reduction in transpiration losses by 22.4% compared to an open crop field. Finally, the average perceived temperature experienced by workers improves by approximately 5.2 °C during working hours. While these are illustrative results for a particular crop and climate, they demonstrate how the model enables integrated assessment and optimization of agrivoltaic benefits across varying climates, crops, and PV technologies.

Satellite-driven soil moisture monitoring systems currently available fail to meet the spatial resolution requirement for a wide range of applications. This limitation is particularly critical for agricultural water management, assessing risks associated with extreme events, and hydrological modeling. This work aims to address the spatial limitations of satellite soil moisture remote sensing by developing GSSM-10, a global 10-meter resolution surface soil moisture dataset, using multi-sensor datasets integrated within an ensemble machine learning framework. These datasets encompass diverse data types—active microwave, multispectral, thermal infrared, and land elevation—offering a robust and comprehensive approach to estimating surface soil moisture (SSM). The ensemble model incorporates TabNet, Random Forest (RF), and Extreme Gradient Boosting (XGBoost). The model was trained on ground-truth data collected from the International Soil Moisture Network (ISMN). The ensemble model demonstrated robust performance, achieving an R² of 0.8344, a bias of –0.0001, an RMSE of 0.0433 m³/m³, and an ubRMSE of 0.0433 m³/m³ in 5-fold cross-validation. When evaluated on a held-out test set, the model achieved similar levels of accuracy, with an R² of 0.8591, a bias of –0.0002 m³/m³, and an RMSE/ubRMSE of 0.0401 m³/m³. An interactive web platform has been developed for data access, visualization, and download, enabling broad adoption by researchers, practitioners, and policymakers. By providing globally consistent, high-resolution SM estimates, GSSM-10 represents a significant advancement in satellite-based soil moisture monitoring for environmental and agricultural applications.

By 2050, the demand for food must increase by 50% to satisfy the needs of the expected 9 billion people. Precision agriculture, primarily driven by technology, aims to produce more with less input (i.e., water) by managing in-field soil and plant variability. Point measurement with traditional soil moisture sensors calls for a large number of sensors, thus increasing costs. The low spatial resolution and the limited measurement depth of the space-borne microwave sensors are not pertinent to field-scale and agricultural applications. Cosmic Ray Neutron Sensors (CRNS) emerge as an alternative technology for large-scale measurements of moisture (SM) content and biomass water equivalent (BWE). However, a lot of uncertainties are still in assessing the measurement footprint in space and depth. This work will explore the potential of combining high-resolution microwave L-band measurements using a UAV platform with CRNS to evaluate the spatial and temporal moisture content distribution. An experimental setup of one eddy covariance station and 2 Finapp CRNS (with and without Gadolinium) was installed on an alfalfa field (Davis, California). TDR measurements were taken before and after irrigation at two different depths to calibrate both CRNS and Microwave L-band measurements. An on-site multispectral sensor (Arable) was used to monitor canopy development and to validate CRNS’ BWE and the microwave’s vegetation optical depth (VOD). The calibrated values were also compared with satellite measurements. Both CRNS and high-resolution microwave measurements showed promising results, especially when used in conjunction. However, research is still needed before unlocking the full commercial potential and mass adoption in the agricultural sector.