Peer-Review Publications

Where soil differs with depth, models that carry water-holding properties inferred at the surface down the whole column bias the moisture reaching crop roots. We applied depth-varying properties to a hyper-resolution land surface model resolving individual agricultural fields at 30 meters. Against ground measurements, systematic surface soil moisture biases dropped by about 14%, with error spread and timing unchanged, correcting the mean state rather than daily dynamics.

Brazil is widely recognized as a global hotspot for flash drought, which sets in over weeks rather than a season, yet events are often reported under other names, among them wet-season dry spells, leaving the evidence unconsolidated. We reviewed 92 articles from 2012 to September 2024. From these, 23 of them were Brazil-focused events. Only five of those explicitly examined flash drought, and at least 35 definitions were in use across the reviewed set, hampering cross-study comparison and early warning; the North, Center-West and South remain underrepresented.

Indigenous Lands hold back Amazon deforestation, yet Brazil’s buffer-zone protections cover conservation units and not Indigenous territories, exposing their edges to extraction and drug trafficking. Satellite forest-loss records from 2008 to 2024 were compared inside the Vale do Javari Indigenous Land, noted for its tri-border setting and its large population living in voluntary isolation, and in zones out to 200 kilometers. Clearing there surged from 2017 to 2022, in 2022 running nearly 187.5 times the loss inside the territory, and 64% fell within holdings on the national rural land registry, misused to claim public land. It eased in 2023 and 2024 yet stayed well above the previous decade.

Many countries report crop yields only for large administrative units, so models trained on coarse records must be applied at finer scales, where few samples and scale-dependent relationships degrade accuracy. A machine learning model trained on field-level soybean yields in the US Midwest was fine-tuned with Brazilian state-by-year records, a technique called transfer learning, then used to predict municipal yields from 2001 to 2021. Transfer across scales rather than across regions had rarely been tested. Without any municipal training data, explained yield variance rose from 0.29 to 0.44, though transfer added little where fine-scale data were plentiful.

Nitrate pollution in streams is a significant environmental challenge, affecting water quality and ecosystem health. Hydrological connectivity, the way water moves through the landscape, governs nitrate transport and transformation, but measuring it across entire catchments remains a challenge. Spatial patterns of soil moisture can provide key insights into catchment-scale hydrological connectivity, yet they have not been widely integrated into nitrate prediction models. In this study, we developed a deep learning framework trained on daily streamflow data and SMAP-HydroBlocks, a high-resolution satellite-based soil moisture data set. We tested whether deep learning models can predict nitrate concentrations using spatial soil moisture patterns and applied Explainable AI (XAI) methods to identify which areas contribute most to the model predictions. Our findings reveal that soil moisture patterns significantly improve nitrate predictions, with near-stream areas having the highest predictive power. This study highlights the potential of AI-driven approaches to provide spatially detailed insights into nitrate transport, which could support better water quality management and help identify nutrient pollution hotspots at fine spatial scales.

Land surface models, which simulate vegetation, soil, water and carbon within Earth system models, depend on parameters that often cannot be measured directly. Varying one carbon flux parameter within its uncertainty range can shift projected atmospheric CO2 in 2100 more than the choice of emissions scenario. This review presents two decades of work on adjusting parameters to match observations is consolidated into seven recurring obstacles, each paired with a machine learning opportunity.

About half the rain falling on the Amazon returns to the air from soil and plants, so its response to clearing and warming bears on rainfall across South America. Existing estimates are too coarse to detect fine-scale signal of deforestation. We used a surface energy balance model on satellite imagery resolved dry-season losses at 30 meters across Brazil’s 1.3 million square kilometer arc of deforestation, from 1990 to 2022. Over remaining forest ET rose 11 ± 1.5% while cleared land lost 41 ± 2.1%; since forest still covers most of the region, the area-wide trend was positive, 10 ± 2.1%. The ET rise is atritubeted to forest responding to warmer, drier air, though continued clearing could reverse it.

High Mountain Asia supplies rivers that billions depend on, yet gauges are sparse and earlier assessments covered mainly high-altitude sub-catchments, leaving change along whole channels unmapped. Daily flow was reconstructed for 114,147 reaches from 2004 to 2019 by combining hydrologic modeling with satellite-derived discharge, resolving reach by reach what coarser work saw only in aggregate. Discharge rose significantly in 9.7% of reaches, skewed toward smaller tributaries, at a mean 8.3% per year, and 8% of reaches with existing or planned dams showed stream power rising 2.9% per year

Much of what is known about smallholder farming rests on farmers’ own knowledge, and gaps between them and instruments used for measurements are usually treated as error. Reported field size, harvest and season timing from 640 maize farms in Zambia were checked against GPS boundaries, weighed harvests and satellite rainfall. The discrepancies were structured rather than random: small fields were overstated and large ones understated, while low harvests were understated and high ones overstated. Read instead as conceptual mismatch, they point to farmers and researchers holding different mental models.

The Pantanal, the world’s largest tropical wetland, dried unusually over 2000-2021, with climate variability and land clearing in its headwaters both proposed as causes. Land-cover maps, gauged discharge and records of rainfall and evaporative demand back to 1951 were combined in a water balance analysis for the Upper Paraguay River Basin, reaching three decades further back than the land-cover record. We found that the runoff change across two wet and two dry periods, 96% traced to climate and only 4% to land cover, while wetland area shrank about 30% after 1985. The recent dry period was hotter and more arid than the 1965-1975 drought and may worsen rather than revert to humid conditions.

Optical satellite methods infer soil wetness from surface reflectance and greenness, filling the gap left by coarse resolution microwave sensors. Calibrated once for a whole region, they are biased toward the dominant land cover; calibrating separately by land cover class offers a way around that. We show that vegetation-type specific calibration on California’s Central Valley cut average soil moisture estimation error from 0.09 to 0.05 cubic meters per cubic meter and resolved individual fields and water pathways.

Brazil holds 16% of the world’s freshwater, yet its water systems are still not fully understood. To understand how rainfall and rivers interact across the country, we studied the hydrologic signatures of 735 catchments. Through catchment classification we identified six major water behavior groups, mostly based on how wet or dry the regions are. In drier areas, climate drives streamflow; in wetter ones, soil and land features matter more. This helps map and predict when and where floods or droughts are likely to happen and it also guides smarter water storage and usage in vulnerable regions. Ultimately, this research supports better planning for a changing climate.

This study aims to better understand flash droughts worldwide and their varying characteristics and impacts. We surveyed the experiences of people affected by flash droughts and then examined a wide range of literature, including non-English and nonacademic sources. This helped us understand how flash droughts can differ from those commonly studied in the United States and China. We identified and described five types of flash droughts, some of which may not be detected by current drought measurement methods. It is crucial to include all types of flash droughts in drought monitoring systems and management plans, as they are expected to become more common due to global warming. We can then better prepare for and reduce the impacts of this growing threat.

Aflatoxin and fumonisin contamination of corn costs the United States up to an estimated $1.66 billion a year, yet the forecasting models used elsewhere have no US counterpart. Machine learning models were trained on 1,772 mycotoxin measurements from Illinois corn between 2003 and 2021, paired with daily rather than monthly weather, satellite greenness and county soil properties, which sets them apart from an earlier attempt in the state. High contamination events were rare. Even so, the best models reached 96% and 92% balanced accuracy on withheld data. Carbonate-rich soils tracked lower aflatoxin and wetter soils higher fumonisin, relationships that remain correlational.

This study develops the version 2 of the Köppen-Geiger climate classification maps at 1-kilometer resolution from 1901 to 2099 and seven future climate pathways. We screened climate models by how plausibly they warm and dropping the least realistic, which narrows projected change and its spread. Roughly 5% of land outside Antarctica is shifited a major climate class since 1901-1930, and 5% to 13% more is projected by 2071-2099 depending on emissions scenario.

Nationwide accounts of Brazilian land use have tended to follow a single cover class or biome, leaving unclear how much deforestation fell on long-standing vegetation rather than regrowth. Annual 30-meter satellite land cover maps for 1985-2020 were reprocessed across six Brazil biomes, separating regrown from long-standing vegetation to quantify the country’s land-use transitions at this resolution for the first time. Long-standing vegetation took 72% of clearing and regrowth 17%; of cleared land, 48% was pasture by 2020 and 9% soybean. Pasture peaked nationally in 2006, since giving way to crops, and every biome showed net clearing despite a quarter of cleared land having regrown.

Attributing a smallholder harvest to the farmer’s decisions rather than to soil and weather has been difficult, since field records are scarce. A process-based crop model simulated maize yields across all 72 of Zambia’s districts from 1979 to 2016, varying cultivar, fertilizer rate and planting date, which resolves those three choices separately for the first time nationally. Management explained 53% of yield variance overall, from 27% in the drier south to 82% in the wetter north, where fertilizer alone accounted for 72%.

We assessed how advanced soil properties maps can improve soil moisture modeling using a Land Surface Model. The results showed that using contemporary soil properties maps and pedotransfer functions to estimate soil properties improved soil moisture modeling performance, especially when soil properties data is available at different soil layers.

While potential maize yields in Zambia are 9 t per hectare (t/ha), the average farmer produces only 1–2. We evaluate the contribution of responses to weather risk to that gap by decomposing the yield gap in maize in Zambia. Risk aversion explained 24% of the average gap, varying widely by location, implying 6.75 tons per hectare attainable without added risk against 8.84 when maximizing profit.

In this study we show how a surrogate model can accurately predict the spatial structure of Land Surface Model’s (LSM) hydrological output fields. For a target level of complexity, the optimal LSM tile configuration was determined using a multi-objective Pareto efficiency analysis. We found that approximately 100 tiles can effectively reproduce a quasi-fully distributed LSM setup with 1% of the computational cost. This method provides a path forward to efficiently determine the optimal tile configurations for LSMs while simultaneously considering the spatial heterogeneity and spatial accuracy of hydrologic processes.

Soil water is a major uncertainty in crop models, and correcting it against measured soil moisture can reduce uncertainty on yield gain estimation. We tested 19 site-years of corn and soybean at five US Midwest sites, 2011 to 2019, and tested such corrections using in-situ sensors and satellite soil moisture, including SMAP-HydroBlocks, a 30-meter product not previously used in a crop model. Sensors improved yield estimates in 63% of site-years, mostly under water stress. The satellite data constrained the soil profile far more weakly yet still cut yield error by a median 17.2%.

Drought management is currently informed by a variety of approaches, mostly responding to drought crisis when it happens. Toward more effective and integrated drought management, we introduce a conceptual drought diagnosis framework inspired by diagnostic concepts from the field of medicine. This framework comprises five steps: 1. Initial diagnostic assessment; 2. Diagnostic testing; 3. Consultation; 4. Communication of the diagnosis; and 5. Treatment and prognosis. This framework can help drought managers to be more proactive in enabling drought-affected regions to become more drought resilient in the future.

We introduce SMAP-HydroBlocks (SMAP-HB), a high-resolution satellite-based surface soil moisture dataset at an unprecedented 30-m resolution (2015–2019) across the conterminous United States. SMAP-HB was produced by using a scalable cluster-based merging scheme that combines high-resolution land surface modeling, radiative transfer modeling, machine learning, SMAP satellite microwave data, and in-situ observations. We evaluated the resulting dataset over 1,192 observational sites. Its largest benefit of SMAP-HB is the high spatial detail and improved representation of the soil moisture spatial variability and spatial accuracy with respect to SMAP products.

Although there have been significant advances in river routing and sub-grid heterogeneity (i.e., tiling) schemes in Earth system models over the past decades, there has yet to be a concerted effort to couple these two concepts. This paper aims to bridge this gap through the development of a two-way coupling between tiling schemes and river networks in the HydroBlocks land surface model. The scheme is implemented and tested over a 1 arc degree domain in Oklahoma, United States.

Drought monitoring and yield prediction often rely on coarse-scale hydroclimate data or (infrequent) vegetation indexes that do not always indicate the conditions farmers face in the field. Consequently, decision-making based on these indices can often be disconnected from the farmer reality. Our study focuses on smallholder farming systems in data-sparse developing countries, and it shows how field-scale soil moisture can leverage and improve crop yield prediction and drought impact assessment.

We evaluated the largest and most diverse set of surface soil moisture products ever evaluated in a single study. We found pronounced differences in performance among individual products and product groups. Our results provide guidance to choose the most suitable product for a particular application.

Precipitation frequency and the intensity of the heaviest events underpin infrastructure design and climate model evaluation, yet earlier near-global climatologies drew on just one of satellites, model reconstructions or gauges, with only 16% of global land placed within 25 kilometers of a gauge. We developed a machine learning method to combined all three, mapping both quantities from gauge records for 1979-2018 onto a 0.1-degree global grid. Against withheld gauges, the resulting PPDIST dataset explained 78% of the variance in daily occurrence and 72% in peak intensity, outperforming reconstruction- and satellite-based estimates.

Soil moisture is a direct measure of agricultural drought, yet operational global indices rely on rainfall or model estimates rather than observed soil water in near-real time. We developed the first operational global soil moisture index from merged near-real-time satellite data, mapping daily percentiles at a 7-8 day lag. Over the Canadian Prairies it tracked the dry 2012-2013, wet 2016 and severe 2017 drought.

Helicopters can experience a phenomenon called brownout when flying close to a dusty surface the uplifting of dust in the air can remarkably restrict the pilot’s visibility area. We show how realistic simulations of brownout need billions of grain particles. Tracking each grain as an independent trajectory through a precomputed airflow allows the work to be spread across multiple GPUs showed to be effective in reducing the simulation time substantially.

Farmers’ climate perceptions guide adaptation, but previsouly they have been tested against weather stations rather than soil moisture, which matters more for crops. We used surveys of 1,171 Zambian farm households after the 2016 harvest recalled onset dates for recent and earlier seasons. In our analysis each farmer’s own onset rule was compared with local rainfall and modeled soil moisture. Some 88% reported rains arriving 21.9 days later than a decade ago, yet the mean perceived onset for the latest season nearly matched the physically derived mean. Recall bias sets in as early as a year after harvest, and planting rules track maize planting dates, so policy should target current variability over future change.

Governments forecast harvests partly from farmers’ mid-season yield expectations, which may be biased. Data from field surveys are often late and costly. Zambia’s maize forecast and post-harvest surveys were compared by district over six seasons from 2001 to 2012, alongside yield models built from weather and soil data. Forecasts averaged above the harvests later recorded, with a typical district-level error of 780 kilograms per hectare, 56% of mean recorded yield. A machine learning model cut that by 13.3%, beating the survey even on pre-planting weather. A major limitations is that models fitted to past weather-yield relationships, may underperform on unprecendented climate extremes.

Köppen-Geiger classifications, which categorize the world into cliamte types such as tropical, arid or polar, underpin ecology and climate impact work, but the global versions in wide use were drawn at about 10 kilometers or coarser resolution, without any uncertainty quantification. We developed new maps at about 1 kilometer for 1980-2016 by merging several precipitation and temperature datasets and quantifying its confidence value. A companion map projects 2071-2100 under a high-emissions scenario. Against 22,078 weather stations the new map classified 80.0% correctly, versus 66.1 to 73.4% for the earlier maps.

This study represents the most comprehensive global-scale precipitation dataset evaluation to date. We evaluated 13 uncorrected precipitation datasets using precipitation observations from 76 086 gauges, and 9 gauge-corrected ones using hydrological modeling for 9053 catchments. Our results highlight large differences in estimation accuracy, and hence, the importance of precipitation dataset selection in both research and operational applications.

A very strong El Niño was widely expected to ease the western United States drought in winter 2015/16, as in 1982/83 and 1997/98. Forecasts issued on 1 January 2016 covered winter precipitation and temperature and spring snowpack, from both empirical and dynamical methods. A common but poorly supported assumption, that anomalies scale with El Niño strength, was tested. Precipitation proved less sensitive to the signal than assumed. Only 46% of precipitation forecasts had the correct sign, against 87% for temperature and 67% for snowpack.

Coarse climate models predict that clearing Amazon forest reduces rainfall and evapotranspiration, as the water plants and soil return to the air. Finer-scale models mostly predict the opposite, and observations had not settled it, earlier satellite work being brief or confined to part of the basin. Satellite records of both for 2000 to 2012 were compared at 25 kilometers between cleared land, the forest edge beside it and intact forest. Evapotranspiration over cleared land rose 17%, to 1299 millimeters a year, and rainfall 6%, favoring the finer-scale models.