Search

Highlights • A systematic framework is presented for parametric uncertainty quantification. • Sensitive parameters were identified with qualitative and quantitative methods. • Sensitive parameters were calibrated with adaptive surrogate-based optimization. Abstract Quantifying and reducing uncertainties in physics-based hydrological model parameters will improve model reliability for hydrological forecasting. We present an uncertainty quantification framework that combines the strengths of stepwise sensitivity analysis and adaptive surrogate-based multi-objective optimization to facilitate practical assessment and reduction of model parametric uncertainties. Framework performance was tested using the distributed hydrological model Coupled Routing and Excess Storage (CREST) for daily streamflow simulation over ten watersheds. By identifying sensitive parameters stepwisely, we reduced the number of parameters requiring calibration from twelve to seven, thus limiting the dimensionality of calibration problem. By updating surrogate models adaptively, we found the optimal sets of sensitive parameters with the surrogate-based multi-objective optimization. The calibrated CREST was able to satisfactorily simulate observed streamflow for all watersheds, improving one minus Nash-Sutcliffe efficiency (1−NSE) by 65–90% and percentage absolute relative bias (|RB|) by 60–95% compared to the default. The validation result demonstrated that the calibrated CREST was also able to reproduce observed streamflow outside the calibration period, improving 1−NSE by 40–85% and |RB| by 35–90% compared to the default. Overall, this uncertainty quantification framework is effective for assessment and reduction of model parametric uncertainties, the results of which improve model simulations and enhance understanding of model behaviors. [ABSTRACT FROM AUTHOR]

This study investigates the future U.S. PM 2.5 pollution under multiple emissions scenarios, climate states, and long-range transport (LRT) effects using the regional Community Multi-scale Air Quality (CMAQ) model integrated with a regional climate model. CMAQ with fixed chemical lateral boundary conditions (LBCs) successfully reproduces the present-day PM 2.5 pollution and its major species in rural and suburban areas, but has some discrepancies in urban areas such as the Los Angeles Basin, where detailed emissions and meteorology conditions cannot be resolved by the 30 km grid. Its performance is slightly worsened when using dynamic chemical LBCs from global chemical transport model (CTM) simulations, which provide cleaner conditions into the CMAQ lateral boundaries. Under future Intergovernmental Panel on Climate Change (IPCC) emission scenarios, CMAQ projects large PM 2.5 reductions (∼40% for A1B and ∼20% for A1Fi scenario) in the eastern United States, but slight to moderate increases (∼5% for A1B and ∼10% for A1Fi) in the western United States. The projected increases are particularly large (up to 30%) near the Mexico-U.S. border, suggesting that Mexico is a major source for future U.S. PM 2.5 pollution. The effect from climate change alone is estimated to increase PM 2.5 levels ubiquitously (∼5% for both A1B and A1Fi) over the United States, except for a small decrease in the Houston, Texas area, where anthropogenic non-methane volatile organic compounds (NMVOCs) emissions dominate. This climate penalty, however, is substantially smaller than effects of emissions change, especially in the eastern United States. Future PM 2.5 pollution is affected substantially (up to −20%) by changes in SO 2 emissions and moderately (3–5%) by changes in NO x and NH 3 emissions. The long-range transport (LRT) effects, which are estimated by comparing CMAQ simulations with fixed and dynamic LBCs, are regional dependent, causing up to 10–20% decrease over the western United States in future summertime PM 2.5 pollution. Therefore, it is important to consider the relative contributions of emissions scenarios, climate conditions, and LRT to the major PM 2.5 components in future U.S. air quality regulation. [ABSTRACT FROM AUTHOR]

A consistent modeling framework with nested global and regional chemical transport models (CTMs) is used to separate and quantitatively assess the relative contributions to projections of future U.S. ozone pollution from the effects of emissions changes, climate change, long-range transport (LRT) of pollutants, and differences in modeling design. After incorporating dynamic lateral boundary conditions (LBCs) from a global CTM, a regional CTM's representation of present-day U.S. ozone pollution is notably improved, especially relative to results from the regional CTM with fixed LBCs or from the global CTM alone. This nested system of global and regional CTMs projects substantial surface ozone trends for the 2050's: 6–10 ppb decreases under the ‘clean’ A1B scenario and ∼15 ppb increases under the ‘dirty’ A1Fi scenario. Among the total trends of future ozone, regional emissions changes dominate, contributing negative 25–60% in A1B and positive 30–45% in A1Fi. Comparatively, climate change contributes positive 10–30%, while LRT effects through changing chemical LBCs account for positive 15–20% in both scenarios, suggesting introducing dynamic LBCs could influence projections of the U.S. future ozone pollution with a magnitude comparable to effects of climate change alone. The contribution to future ozone projections due to differences in modeling design, including model formulations, emissions treatments, and other factors between the global and the nested regional CTMs, is regionally dependent, ranging from negative 20% to positive 25%. It is shown that the model discrepancies for present-day simulations between global and regional CTMs can propagate into future U.S. ozone projections systematically but nonlinearly, especially in California and the Southeast. Therefore in addition to representations of emissions change and climate change, accurate treatment of LBCs for the regional CTM is essential for projecting the future U.S. ozone pollution. [ABSTRACT FROM AUTHOR]

Abstract: This study develops fine temporal (seasonal, day-of-week, diurnal) and vertical allocations of anthropogenic emissions for the TRACE-P inventory and evaluates their impacts on the East Asian air quality prediction using WRF-Chem simulations in July 2001 at 30-km grid spacing against available surface measurements from EANET and NEMCC. For NO2 and SO2, the diurnal and vertical redistributions of emissions play essential roles, while the day-of-week variation is less important. When all incorporated, WRF-Chem best simulates observations of surface NO2 and SO2 concentrations, while using the default emissions produces the worst result. The sensitivity is especially large over major cities and industrial areas, where surface NO2 and SO2 concentrations are reduced by respectively 3–7 and 6–12 ppbv when using the scaled emissions. The incorporation of all the three redistributions of emissions simulates surface O3 concentrations higher by 4–8 ppbv at night and 2–4 ppbv in daytime over broad areas of northern, eastern and central China. To this sensitivity, the diurnal redistribution contributes more than the other two. [Copyright &y& Elsevier]

Although the Soil and Water Assessment Tool (SWAT) model has been widely used to assess the environmental impacts of growing perennial grasses for bioenergy production, its utility is limited by not explicitly accounting for shoot and root biomass development. In this study, we integrated the DAYCENT model's grass growth algorithms into SWAT (SWAT–GRASS D) and further modified it by considering the impact of leaf area index (LAI) on potential biomass production (SWAT–GRASS M). Based on testing at eight sites in the US Midwest, we found that SWAT–GRASS M generally outperformed SWAT and SWAT–GRASS D in simulating switchgrass biomass yield and the seasonal development of LAI. Additionally, SWAT–GRASS M can more realistically represent root development, which is key for the allocation of accumulated biomass and nutrients between aboveground and belowground biomass pools. These improvements are critical for credible assessment of agronomic and environmental impacts of growing perennial grasses for biomass production. • We developed the SWAT-GRASS model by integrating the grass growth module from the DAYCENT model into SWAT. • We further enhanced SWAT-GRASS to better represent the effects of LAI on biomass accumulation. • The new model achieved better simulation of switchgrass yield across multiple sites. • The new model can explicitly and reasonably represent root and shoot development of switchgrass. • A new tool for comprehensive assessment of agronomic and environmental impacts of growing switchgrass. [ABSTRACT FROM AUTHOR]

Abstract: A global climate chemistry model CAM-Chem is driven by the meteorology output from community climate system model version 3 (CCSM3) to investigate the relative contributions of changes in local anthropogenic emissions (LE) versus changes in remote anthropogenic emissions (RE) to global surface ozone air quality in 2050. On major ozone pollution regions (Europe, the United States, Asia), the effects are examined following three distinct pathways, A1FI, A1B and B1, from the Intergovernmental Panel on Climate Change (IPCC) Special Report on Emissions Scenarios (SRES) to address the uncertainty in projections of future climate and emissions. We find that projected changes in anthropogenic emissions under the A1FI scenario lead to an increase of 5–14 ppb in summertime daily maximum 8-h (DM8H) ozone concentration over U.S. by 2050, of which 48% is contributed by LE changes and 52% is contributed by RE changes. For Europe, the change in local emissions contributes 62% of the surface ozone increase in 2050 summer, while 38% of the increase is attributed to remote emission change. For Asia, changes in LE dominate the 2050 surface ozone increase with a magnitude of 10–30 ppb on summertime DM8H ozone concentration. However, under the A1B and B1 scenarios, contributions from LE changes are much larger than that from RE changes over all three regions except Asia under the B1 scenario, in which the RE changes contribute 31% of total change. The results indicate that for the United States and Europe, pollution control is a local issue under global low emission situations, while it becomes an international issue when fossil fuel use is rapidly increasing. Due to the weak Euro-Asia transport, local emission increase seems to be the main force for Asia''s ozone air quality change under all cases except the low emission scenario B1. Therefore, the strategies for regional air quality control need to be based on global emission situation. [Copyright &y& Elsevier]

Abstract: We investigate the effects of projected global changes in climate and human-related emissions for the year 2050 relative to 2000 for trends in the potential risk of hazardous ozone pollution episodes using a global climate chemistry model, CAM-Chem, driven by meteorology output from Community Climate System Model 3. Three distinct pathways from the Intergovernmental Panel on Climate Change Special Report on Emissions Scenarios A1FI, A1B and B1 are considered to address the range and uncertainty in projected climate and emission changes. The projected changes in extreme climate conditions are likely to intensify the associated extreme weather conditions that foster the risk of high ozone pollution episodes over many parts of the world. It is found that the changes in regional high surface ozone concentration by 2050 largely depends on changes in the anthropogenic emissions of ozone precursors. Our analysis under projected climate and emissions on the frequency of “hazardous ozone days” in which the peak ozone concentration exceed the limit in the summer of 2050, based on 8 and 1-h standards, show that the risk of hazardous ozone pollution episodes will likely increase in developing regions, but changes of risk in developed regions depend on scenarios. For three major pollutant regions, north America, Europe and East Asia under 8-h definition, the hazardous risk in all regions increases under the A1FI scenario with a potential rate of 39–79 days/summer by 2050, but it is likely to decrease over North America and Europe under the A1B and B1 scenarios. The risk on Europe under the A1B and B1 scenarios can be ignored, but a significant rate of 6–71 days/summer still occur on other regions. The relative variability in projected frequency of hazardous ozone days increase by using the 1-h definition, although it shows the highest risk of 17–59 days/summer under the A1FI scenario. The higher variability can be understood through statistical analysis of cumulative frequency distribution of daily peak ozone concentrations, which shows that the changes of surface ozone level are more significantly presented on the high value part. [Copyright &y& Elsevier]

Simulation of summertime U.S. surface ozone diurnal cycle is influenced by the model representation of planetary boundary layer (PBL) mixing, spatial resolution, and precursor emissions. These factors are investigated here for five major regions (Northeast, Midwest, Southeast, California, and Southwest) by using the Model for Ozone And Related chemical Tracers version 2.4 (MOZART-2.4), with important modifications, to conduct sensitivity experiments for summer 1999 with three PBL mixing schemes, two horizontal resolutions and two emissions datasets. Among these factors, the PBL mixing is dominant. The default non-local scheme well reproduces the observed ozone diurnal variation, where the timing for the afternoon maximum and the morning minimum is within 1h of the observed; biases for the minimum are less than 5ppb except over the Southeast; and the ozone maximum–minimum contrast (OMMC) is within 10ppb of observations except for the overprediction by 18.9ppb over the Northeast. In contrast, the local scheme significantly overestimates the OMMC by 10–34ppb over all regions as ozone and precursors are trapped too close to the ground. On the other hand, the full-mixing assumption underestimates the OMMC by 0–25ppb, except over the Northeast, as the nighttime ozone decline is greatly underpredicted. As compared to PBL mixing, the effects of horizontal resolutions and precursor emissions being used are smaller but non-negligible. Overall, with the non-local mixing scheme, relatively high horizontal resolution (∼1.1°) and updated emissions data, the modified MOZART is capable of simulating the main features of the observed ozone diurnal cycle. [Copyright &y& Elsevier]

• The impacts of monsoon rainfall on food insecurity differed by earthquake exposure. • Low rainfall was linked with greater food insecurity in non-impacted areas. • More rainfall was positively linked with food insecurity in affected areas. • This is likely due to rainfall increasing landslides in earthquake-affected regions. Food insecurity is a key global health challenge that is likely to be exacerbated by climate change. Though climate change is associated with an increased frequency of extreme weather events, little is known about how multiple environmental shocks in close succession interact to impact household health and well-being. In this paper, we assess how earthquake exposure followed by monsoon rainfall anomalies affect food insecurity in Nepal. We link food security data from the 2016 Nepal Demographic and Health Survey to data on shaking intensity during the 2015 Gorkha earthquake and rainfall anomalies during the 2015 monsoon season. We then exploit spatial variation in exposure to the earthquake and monsoon rainfall anomalies to isolate their independent and compound effects. We find that earthquake exposure alone was not associated with an increased likelihood of food insecurity, likely due in part to effective food aid distribution. However, the effects of rainfall anomalies differed by severity of earthquake exposure. Among households minimally impacted by the earthquake, low rainfall was associated with increased food insecurity, likely due to lower agricultural productivity in drought conditions. Among households that experienced at least moderate shaking, greater rainfall was positively associated with food insecurity, particularly in steep, mountainous areas. In these locations, rainfall events disproportionately increased landslides, which damaged roads, disrupted distribution of food aid, and destroyed agricultural land and assets. Additional research on the social impacts of compound environmental shocks is needed to inform adaptation strategies that work to improve well-being in the face of climate change. [ABSTRACT FROM AUTHOR]

This study investigates the role of convection representation across the gray zone in the multigrid nesting Weather Research and Forecasting (WRF) model based on two extreme precipitation events in Shanghai during the warm season. The precipitation forecast sensitivity and physical processes are examined using three convective treatments: (1) the traditional Kain-Fritsch (KF), (2) scale-aware Grell-Freitas (GF) cumulus parameterization, and (3) explicit convection (EC) solution and different nesting configurations consisting of grid spacings of 27, 15, 9, 5, 3, and 1 km using single, double, and triple nestings. At coarser grids (> 10 km), GF produces wet biases with delayed rainfall peaks, while KF produces dry biases with largely underestimated rainfall peaks. At a 9-km grid spacing, these biases were largely reduced, and using cumulus parameterizations (CUPs) shows specific advantages over the EC in the diurnal variation in extreme precipitation. GF performs better in simulating primary rainfall peak intensity, but KF performs better in simulating diurnal timing, albeit with underestimated intensity. The utilization of CUPs is beneficial at the 5-km grid in simulating both extreme intensity and diurnal variations, but it shows mixed effects at 3 km. When the 1-km grid with explicit convection was nested directly into their outermost 15-km or 9-km grids using KF CUP, avoiding treatment in the gray zone (5 km or 3 km), the primary rainfall peak at noon was best reproduced, but a secondary peak with a weak forcing was not detected. This peak was identified with local moisture divergence and a stable atmosphere. • Cumulus parameterization is beneficial with 9- and 5-km grids. • Kain-Fritsch can predict the diurnal rainfall peak better than the Grell-Freitas scheme for extreme rainfall in Shanghai. • Double nesting with a large ratio outperforms traditional triple nesting when forecasting the extreme rainfall in Shanghai. [ABSTRACT FROM AUTHOR]

The ability of climate models to capture extreme precipitation events is crucially important, but most of the existing models contain significant biases for the simulation of extreme precipitation. To understand the causes of these biases, we used five different cumulus parameterization schemes in the regional Climate–Weather Research and Forecasting (CWRF) model to investigate its performance and biases in the simulation of extreme precipitation events in China. In general, the ensemble cumulus parameterization (ECP) scheme was the most skillful in reproducing the spatial distribution of the 95th percentile daily precipitation (P95) and the other four schemes either overestimated (the Kain-Fritsch Eta and Tiedtke schemes) or underestimated (the Betts-Miller-Janjic and New Simplified Arakawa-Schubert schemes) P95. Compared with the observational data, ECP scheme significantly improved the simulation of extreme precipitation in China and had the highest correlation and the smallest root-mean-square error in most areas and seasons. To clarify the underlying physical processes of P95 simulation biases, we established a regression model of extreme precipitation based on ECP scheme. This showed that P95 in North China is mainly affected by moisture convergence, planetary boundary layer height and lifting condensation level (relative importance 18–32%). In Central China, the vertical upward motion of water vapor, sensible heat flux and planetary boundary layer height (relative importance 18–30%) are main factors associated with P95. In South China, the vertical upward motion and horizontal transport of water vapor are predominant (relative importance 26–37%). In addition, the net surface energy, surface and atmospheric radiation flux, total precipitable water, convective available potential energy and cloud water path also have a high correlation with P95 (the second most important factor; relative importance 14–31%). The influence of each factor on the simulation of P95 is different when using different cumulus parameterization schemes and the interaction among the different factors determines the ability of CWRF model to simulate extreme precipitation. These results provide important references for future model evaluations and improvements. • Five cumulus schemes were compared for their effects on China P95 using CWRF. • The ensemble cumulus parameterization outperformed the other schemes in simulating P95. • The underlying physical processes and mechanisms of P95 biases were explored. [ABSTRACT FROM AUTHOR]

• A unified agricultural modeling framework is implemented in DLEM v4.0. • Simulated results agree well with site-scale LAI, biomass and yield measurements. • Regional simulations can well reproduce spatial-temporal patterns of crop production. • DLEM v4.0 can be used to support agricultural climate adaptation and mitigation. Agricultural decision-making by different interest groups (e.g., farmers, development agents and policy makers) usually takes place on different scales (e.g., plot, landscape and country). Currently, tools to assist decision-making are either dedicated to small-scale management guidance or large-scale assessment, which ignore the cross-scale linkages and interactions and thus may not provide robust and consistent guidance and assessment. Here, we developed an advanced agricultural modeling framework by integrating the strengths of conventional crop models in representing crop growth processes and management practices into a terrestrial biosphere model (TBM), the Dynamic Land Ecosystem Model (DLEM), to meet the cross-scale application needs (e.g., adaptation and mitigation). Specifically, dynamic crop growth processes, including crop-specific phenological development, carbon allocation, yield formation, biological nitrogen fixation processes, and management practices such as tillage, cover cropping and genetic improvements, were explicitly represented in DLEM. The new model was evaluated against site-scale observations and the results showed that the model performed generally well, with an average normalized root mean square error of 19.91% for leaf area index and 17.46% for aboveground biomass at the seasonal scale and 14.42% for annual yield. Then the model was applied to simulate corn, soybean, and winter wheat productions in the conterminous United States from 1960 to 2018. The spatial patterns of simulated crop productions were consistent with ground survey data. Our model also captured both the long-term trends and interannual variations of the total national productions of the three crops. This study demonstrates the significance of fusing conventional crop modeling techniques into TBMs to establish a unified modeling framework, which holds the potential to address climate impacts, adaptation and mitigation across varied spatiotemporal scales. [ABSTRACT FROM AUTHOR]

The weak and sustained tropical cyclone (TC) Rumbia (2018) and its associated heavy precipitation after landfall are simulated by the Weather Research and Forecasting model using three microphysics parameterizations (MP) and two planetary boundary layer (PBL) schemes. Key factors regulating the heavy rainfall distribution of landed TC at three stages (landfall, inland slow-moving and recurving stage) and their sensitivities to multi-physics parameterizations are examined. Results show that heavy rainfall distribution is largely regulated by the intensity of TC itself at landfall, but more affected by environmental factors at the two inland stages, including the environmental vertical wind shear, the associated upper-level wind divergence and low-level moisture convergence. Different MP and PBL schemes mainly affect the simulation of TC thermodynamic structure and key environmental factors, leading to their different forecast skills at three stages. Specifically, the Yonsei University (YSU) PBL scheme systematically outperforms the Mellor-Yamada-Nakanishi-Niino (MYNN) scheme in simulating stronger TC and heavy rainfall intensity. With the advantageous YSU PBL scheme, the Ferrier MP scheme only shows a slight advantage in simulating the intensity of TC and heavy rainfall at landfall. However, the WRF single-moment 6-class (WSM6) and Thompson scheme produce more graupel or snow, and better simulate the key environmental factors, showing their respective advantages in simulating the heavy rainfall structure and location at the two inland stages. This implies that the use of nonlocal YSU PBL scheme and the MP schemes with sophisticated ice processes shows superiority in simulating the postlandfall heavy rainfall induced by weak TCs. • Postlandfall rainfall of weak TC is highly affected by environmental factors. • Ice species and radial distribution of diabatic heating affect TC structures and rainfall. • The MYNN PBL scheme underestimates TC intensity and heavy rainfall compared to YSU. [ABSTRACT FROM AUTHOR]