Explore the vast range of titles available.

Crop yield estimates over large areas are conventionally made using weather observations, but a comprehensive understanding of the effects of various environmental indicators, observation frequency, and the choice of prediction algorithm remains elusive. Here we present a thorough assessment of county-level maize yield prediction in U.S. Midwest using six statistical/machine learning algorithms (Lasso, Support Vector Regressor, Random Forest, XGBoost, Long-short term memory (LSTM), and Convolutional Neural Network (CNN)) and an extensive set of environmental variables derived from satellite observations, weather data, land surface model results, soil maps, and crop progress reports. Results show that seasonal crop yield forecasting benefits from both more advanced algorithms and a large composite of information associated with crop canopy, environmental stress, phenology, and soil properties (i.e. hundreds of features). The XGBoost algorithm outperforms other algorithms both in accuracy and stability, while deep neural networks such as LSTM and CNN are not advantageous. The compositing interval (8-day, 16-day or monthly) of time series variable does not have significant effects on the prediction. Combining the best algorithm and inputs improves the prediction accuracy by 5% when compared to a baseline statistical model (Lasso) using only basic climatic and satellite observations. Reasonable county-level yield foresting is achievable from early June, almost four months prior to harvest. At the national level, early-season (June and July) prediction from the best model outperforms that of the United States Department of Agriculture (USDA) World Agricultural Supply and Demand Estimates (WASDE). This study provides insights into practical crop yield forecasting and the understanding of yield response to climatic and environmental conditions.

Leaf Area Index (LAI) is a key variable that bridges remote sensing observations to the quantification of agroecosystem processes. In this study, we assessed the universality of the relationships between crop LAI and remotely sensed Vegetation Indices (VIs). We first compiled a global dataset of 1459 in situ quality-controlled crop LAI measurements and collected Landsat satellite images to derive five different VIs including Simple Ratio (SR), Normalized Difference Vegetation Index (NDVI), two versions of the Enhanced Vegetation Index (EVI and EVI2), and Green Chlorophyll Index (CI(sub Green)). Based on this dataset, we developed global LAI-VI relationships for each crop type and VI using symbolic regression and Theil-Sen (TS) robust estimator. Results suggest that the global LAI-VI relationships are statistically significant, crop-specific, and mostly non-linear. These relationships explain more than half of the total variance in ground LAI observations (R2 greater than 0.5), and provide LAI estimates with RMSE below 1.2 m2/m2. Among the five VIs, EVI/EVI2 are the most effective, and the crop-specific LAI-EVI and LAI-EVI2 relationships constructed by TS, are robust when tested by three independent validation datasets of varied spatial scales. While the heterogeneity of agricultural landscapes leads to a diverse set of local LAI-VI relationships, the relationships provided here represent global universality on an average basis, allowing the generation of large-scale spatial-explicit LAI maps. This study contributes to the operationalization of large-area crop modeling and, by extension, has relevance to both fundamental and applied agroecosystem research.

As global water scarcity intensifies, identifying agricultural practices that enhance sustainable water management is critical. Temporal crop diversification-rotating multiple species over time-has been proposed to improve soil health and water retention based on field-scale experiments. However, widespread adoption remains limited on farms, in part due to unverified benefits at larger scales. Here, we assess the influence of crop diversification on agricultural water-use efficiency (WUE, ratio of gross primary productivity to evapotranspiration) along a spectrum of monoculture to complex species rotations in California. Leveraging new high-resolution remote sensing datasets, we show that crop diversification is a key driver of agricultural WUE, and increasing the number of species planted in the previous 6 years from two to four increases WUE by ∼20% after accounting for differences between crops. Our results provide spatially explicit, large-scale quantification of crop diversification’s improvements to WUE, with direct implications for climate adaptation. More broadly, our framework offers a tool to evaluate other sustainable practices and guide policy and farm-scale decision-making.

Accurate and timely access to the production area of crop seeds allows the seed market and secure seed supply to be monitored. Seed maize and common maize production fields typically share similar phenological development profiles with differences in the planting patterns, which makes it challenging to separate these fields from decametric-resolution satellite images. In this research, we proposed a method to identify seed maize production fields as early as possible in the growing season using a time series of remote sensing images in the Liangzhou district of Gansu province, China. We collected Sentinel-2 and GaoFen-1 (GF-1) images captured from March to September. The feature space for classification consists of four original bands, namely red, green, blue, and near-infrared (nir), and eight vegetation indexes. We analyzed the timeliness of seed maize identification using Sentinel-2 time series of different time spans and identified the earliest time frame for reasonable classification accuracy. Then, the earliest time series that met the requirements of regulatory accuracy were compared and analyzed. Four machine/deep learning algorithms were tested, including K-nearest neighbor (KNN), support vector classification (SVC), random forest (RF), and long short-term memory (LSTM). The results showed that using Sentinel-2 images from March to June, the RF and LSTM algorithms achieve over 88% accuracy, with the LSTM performing the best (90%). In contrast, the accuracy of KNN and SVC was between 82% and 86%. At the end of June, seed maize mapping can be carried out in the experimental area, and the precision can meet the basic requirements of monitoring for the seed industry. The classification using GF-1 images were less accurate and reliable; the accuracy was 85% using images from March to June. To achieve near real-time identification of seed maize fields early in the growing season, we adopted an automated sample generation approach for the current season using only historical samples based on clustering analysis. The classification accuracy using new samples extracted from historical mapping reached 74% by the end of the season (September) and 63% by the end of July. This research provides important insights into the classification of crop fields cultivated with the same crop but different planting patterns using remote sensing images. The approach proposed by this study enables near-real time identification of seed maize production fields within the growing season, which could effectively support large-scale monitoring of the seed supply industry.

Coupling crop growth models and remote sensing provides the potential to improve our understanding of the genotype x environment x management (G × E × M) variability of crop growth on a global scale. Unfortunately, the uncertainty in the relationship between the satellite measurements and the crop state variables across different sites and growth stages makes it difficult to perform the coupling. In this study, we evaluate the effects of this uncertainty with MODIS data at the Mead, Nebraska Ameriflux sites (US-Ne1, US-Ne2, and US-Ne3) and accurate, collocated Hybrid-Maize (HM) simulations of leaf area index (LAI) and canopy light use efficiency (LUECanopy). The simulations are used to both explore the sensitivity of the satellite-estimated genotype × management (G × M) parameters to the satellite retrieval regression coefficients and to quantify the amount of uncertainty attributable to site and growth stage specific factors. Additional ground-truth datasets of LAI and LUECanopy are used to validate the analysis. The results show that uncertainty in the LAI/satellite measurement regression coefficients lead to large uncertainty in the G × M parameters retrievable from satellites. In addition to traditional leave-one-site-out regression analysis, the regression coefficient uncertainty is assessed by evaluating the retrieval performance of the temporal change in LAI and LUECanopy. The weekly change in LAI is shown to be retrievable with a correlation coefficient absolute value (|r|) of 0.70 and root-mean square error (RMSE) value of 0.4, which is significantly better than the performance expected if the uncertainty was caused by random error rather than secondary effects caused by site and growth stage specific factors (an expected |r| value of 0.36 and RMSE value of 1.46 assuming random error). As a result, this study highlights the importance of accounting for site and growth stage specific factors in remote sensing retrievals for future work developing methods coupling remote sensing with crop growth models.

Terrestrial ecosystems annually absorb ∼30${\\sim} 30$ % of anthropogenic C emissions. The degrees to which contemporary CO2${\\text{CO}}_{2}$and climate trends drive this absorption are uncertain, as are the governing mechanisms. To reduce uncertainty, we use Bayesian model‐data integration (CARbon DAta MOdel fraMework) to retrieve a terrestrial biosphere reanalysis where Earth Observations optimally inform mechanistic model processes: observations include satellite‐ and inventory‐based constraints on distributions and change in terrestrial C (including live biomass, dead organic C, and land‐atmosphere CO2${\\text{CO}}_{2}$exchanges) and underlying mechanisms (including photosynthesis, deforestation, water storage anomalies, and fire). We find that the impact of 2001–2021's atmospheric CO2${\\text{CO}}_{2}$increase on terrestrial C (+39.4 PgC) opposes and far outweighs the impact of climate trends over this period (−${-}$ 10.5 PgC). Globally, C gains are mostly attributable to live biomass growth (+31.2 PgC), while CO2${\\text{CO}}_{2}$ ‐induced dead organic C gains (+7.8 PgC) are compensated by climate‐induced losses (−${-}$ 8.8 PgC). The distribution of compensating dead C changes induces an aggregate shift in dead C from high‐ and mid‐latitudes (−${-}$ 3.5 PgC) to tropical ecosystems (+2.6 PgC). We additionally find global residence time reductions attributable to CO2${\\text{CO}}_{2}$(−${-}$ 2.6%) and climate (−${-}$ 1.3%) reflected across latitudes, irrespective of reservoir C changes. In aggregate, these changes reveal an acceleration and redistribution of terrestrial C stores in response to CO2${\\text{CO}}_{2}$and climate trends, which together reflect a gradual but fundamental reorganization of the terrestrial C cycle. Tracking this reorganization—through robust and continual diagnosis of ecosystem function—is essential for accurately resolving the compensating dynamics governing the strength and resilience of the terrestrial C sink. Plain Language Summary This study explores how land ecosystems responded to rising CO2${\\text{CO}}_{2}$and changing climate from 2001 to 2021. By analyzing satellite data with a model of how ecosystems cycle carbon, we find that rising CO2${\\text{CO}}_{2}$levels have helped land ecosystems store more carbon, especially in living plants. In contrast, climate changes have reduced the amount of carbon stored, particularly in non‐living material such as soil and plant debris. The balance of these effects depends on the region. For example, the tropics gained more dead carbon, and the mid‐ and high‐latitudes lost more dead carbon, than these regions would have without recent climate or CO2${\\text{CO}}_{2}$changes; this implies a gradual shift of global dead carbon from high to low latitudes. We also find that the average time carbon stays in ecosystems is shortening. Overall, our results show that carbon gains and losses affect different ecosystem components and vary across regions. Understanding how these responses interact is key to understanding the biosphere's current capacity to absorb and store carbon, and how that capacity is changing. Key Points CO2${\\text{CO}}_{2}$ ‐driven terrestrial C gains are ∼4 times greater than climate‐driven losses during 2001–2021 and are mainly in live biomass reservoirs In the global sum, CO2${\\text{CO}}_{2}$ ‐induced dead organic C gains are symmetrically compensated by climate‐induced losses Despite net C storage gains, terrestrial C residence times are declining across scales in response to both CO2${\\text{CO}}_{2}$and climate trends

Estimating gross primary productivity (GPP) over space and time is fundamental for understanding the response of the terrestrial biosphere to climate change. Eddy covariance flux towers provide in situ estimates of GPP at the ecosystem scale, but their sparse geographical distribution limits larger-scale inference. Machine learning (ML) techniques have been used to address this problem by extrapolating local GPP measurements over space using satellite remote sensing data. However, the accuracy of the regression model can be affected by uncertainties introduced by model selection, parameterization, and choice of explanatory features, among others. Recent advances in automated ML (AutoML) provide a novel automated way to select and synthesize different ML models. In this work, we explore the potential of AutoML by training three major AutoML frameworks on eddy covariance measurements of GPP at 243 globally distributed sites. We compared their ability to predict GPP and its spatial and temporal variability based on different sets of remote sensing explanatory variables. Explanatory variables from only Moderate Resolution Imaging Spectroradiometer (MODIS) surface reflectance data and photosynthetically active radiation explained over 70 % of the monthly variability in GPP, while satellite-derived proxies for canopy structure, photosynthetic activity, environmental stressors, and meteorological variables from reanalysis (ERA5-Land) further improved the frameworks' predictive ability. We found that the AutoML framework Auto-sklearn consistently outperformed other AutoML frameworks as well as a classical random forest regressor in predicting GPP but with small performance differences, reaching an r2 of up to 0.75. We deployed the best-performing framework to generate global wall-to-wall maps highlighting GPP patterns in good agreement with satellite-derived reference data. This research benchmarks the application of AutoML in GPP estimation and assesses its potential and limitations in quantifying global photosynthetic activity.

Remote sensing estimation of evapotranspiration (ET) directly quantifies plant water consumption and provides essential information for irrigation scheduling, which is a pressing need for California vineyards as extreme droughts become more frequent. Many ET models take satellite-derived Leaf Area Index (LAI) as a major input, but how uncertainties of LAI estimations propagate to ET and the partitioning between evaporation and transpiration is poorly understood. Here we assessed six satellite-based LAI estimation approaches using Landsat and Sentinel-2 images against ground measurements from four vineyards in California and evaluated ET sensitivity to LAI in the thermal-based two-source energy balance (TSEB) model. We found that radiative transfer modeling-based approaches predicted low to medium LAI well, but they significantly underestimated high LAI in highly clumped vine canopies (RMSE ~ 0.97 to 1.27). Cubist regression models trained with ground LAI measurements from all vineyards achieved high accuracy (RMSE ~ 0.3 to 0.48), but these empirical models did not generalize well between sites. Red edge bands and the related vegetation index (VI) from the Sentinel-2 satellite contain complementary information of LAI to VIs based on near-infrared and red bands. TSEB ET was more sensitive to positive LAI biases than negative ones. Positive LAI errors of 50% resulted in up to 50% changes in ET, while negative biases of 50% in LAI caused less than 10% deviations in ET. However, even when ET changes were minimal, negative LAI errors of 50% led to up to a 40% reduction in modeled transpiration, as soil evaporation and plant transpiration responded to LAI change divergently. These findings call for careful consideration of satellite LAI uncertainties for ET modeling, especially for the partitioning of water loss between vine and soil or cover crop for effective vineyard irrigation management.

Gross primary productivity (GPP) is the largest carbon flux in the Earth system, playing a crucial role in removing atmospheric carbon dioxide and providing carbohydrates needed for ecosystem metabolism. Despite the importance of GPP, however, existing estimates present significant uncertainties and discrepancies. A key issue is the underrepresentation of the CO2 fertilization effect, a major factor contributing to the increased terrestrial carbon sink over recent decades. This omission could potentially bias our understanding of ecosystem responses to climate change. Here, we introduce CEDAR-GPP, the first global machine-learning-upscaled GPP product that incorporates the direct CO2 fertilization effect on photosynthesis. Our product is comprised of monthly GPP estimates and their uncertainty at 0.05° resolution from 1982 to 2020, generated using a comprehensive set of eddy covariance measurements, multi-source satellite observations, climate variables, and machine learning models. Importantly, we used both theoretical and data-driven approaches to incorporate the direct CO2 effects. Our machine learning models effectively predict monthly GPP (R2 ∼ 0.72), the mean seasonal cycles (R2 ∼ 0.77), and spatial variabilities (R2 ∼ 0.63) based on cross-validation at flux sites. After incorporating the direct CO2 effects, the predicted long-term GPP trend across global flux towers substantially increases from 3.1 to 4.5–5.4 gC m−2 yr−1, which aligns more closely with the 7.7 gC m−2 yr−1 trend detected from eddy covariance data. While the global patterns of annual mean GPP, seasonality, and interannual variability generally align with existing satellite-based products, CEDAR-GPP demonstrates higher long-term trends globally after incorporating CO2 fertilization and reflected a strong temperature control on direct CO2 effects. The estimated global GPP trend is 0.57–0.76 PgC yr−1 from 2001 to 2018 and 0.32–0.34 PgC yr−1 from 1982 to 2018. Estimating and validating GPP trends in data-scarce regions, such as the tropics, remains challenging, underscoring the importance of ongoing ground-based monitoring and advancements in modeling techniques. CEDAR-GPP offers a comprehensive representation of GPP temporal and spatial dynamics, providing valuable insights into ecosystem–climate interactions. The CEDAR-GPP product is available at https://doi.org/10.5281/zenodo.8212706 (Kang et al., 2024).

Gross primary productivity (GPP) is the largest carbon flux in the Earth system, playing a crucial role in removing atmospheric carbon dioxide and providing carbohydrates needed for ecosystem metabolism. Despite the importance of GPP, however, existing estimates present significant uncertainties and discrepancies. A key issue is the underrepresentation of the CO.sub.2 fertilization effect, a major factor contributing to the increased terrestrial carbon sink over recent decades. This omission could potentially bias our understanding of ecosystem responses to climate change.