Impacts of meteorological variables and machine learning algorithms on rice yield prediction in Korea

As crop productivity is greatly influenced by weather conditions, many attempts have been made to estimate crop yields using meteorological data and have achieved great progress with the development of machine learning. However, most yield prediction models are developed based on observational data, and the utilization of climate model output in yield prediction has been addressed in very few studies. In this study, we estimate rice yields in South Korea using the meteorological variables provided by ERA5 reanalysis data (ERA-O) and its dynamically downscaled data (ERA-DS). After ERA-O and ERA-DS are validated against observations (OBS), two different machine learning models, Support Vector Machine (SVM) and Long Short-Term Memory (LSTM), are trained with different combinations of eight meteorological variables (mean temperature, maximum temperature, minimum temperature, precipitation, diurnal temperature range, solar irradiance, mean wind speed, and relative humidity) obtained from OBS, ERA-O, and ERA-DS at weekly and monthly timescales from May to September. Regardless of the model type and the source of the input data, training a model with weekly datasets leads to better yield estimates compared to monthly datasets. LSTM generally outperforms SVM, especially when the model is trained with ERA-DS data at a weekly timescale. The best yield estimates are produced by the LSTM model trained with all eight variables at a weekly timescale. Altogether this study shows the significance of high spatial and temporal resolution of input meteorological data in yield prediction, which can also serve to substantiate the added value of dynamical downscaling.

Coupling hydroclimate-hydraulic-sedimentation models to estimate flood inundation and sediment transport during extreme flood events under a changing climate

Extreme flood events are disastrous and can cause serious damages to society. Flood frequency obtained based on historical flow records may also be changing under future climate conditions. The associated flood inundation and environmental transport processes will also be affected. In this study, an integrated numerical modeling framework is proposed to investigate the inundation and sedimentation during multiple flood events (2,5,10, 20, 50, 100, 200-year) under future climate change scenarios in a watershed system in northern California, USA. The proposed modeling framework couples physical models of various spatial resolution: kilometers to several hundred kilometers climatic processes, hillslope scale hydrological processes in a watershed, and centimeters to meters scale hydrodynamic and sediment transport processes in a riverine system. The modeling results show that compared to the flows during historical periods, extreme events become more extreme in the 21st century and higher flows tend to be larger and smaller flows tend to be smaller in the system. Flood inundation in the study area, especially during 200-year events, is projected to increase in the future. More sediment will be trapped as the flow increases and the deposition will also increase in the settling basin. Sediment trap efficiency values are within 37.5-65.4% for the historical conditions, within 32.4-68.8% in the first half of the 21st century, and within 34.9-69.3% in the second half of the 21st century. The results highlight the impact of climate change on extreme flood events, the resulting sedimentation, and reflected the importance of incorporating the coupling of physical models into the adaptive watershed and river system management.

Multi-model projections of precipitation extremes in Southeast Asia based on CORDEX-Southeast Asia simulations

This study examines the projected precipitation extremes for the end of 21st century (2081-2100) over Southeast Asia (SEA) using the output of the Southeast Asia Regional Climate Downscaling/Coordinated Regional Climate Downscaling Experiment - Southeast Asia (SEACLID/CORDEX-SEA). Eight ensemble members, representing a subset of archived CORDEX-SEA simulations at 25 km spatial resolution, were examined for emission scenarios of RCP4.5 and RCP8.5. The study utilised four different indicators of rainfall extreme, i.e. the annual/seasonal rainfall total (PRCPTOT), consecutive dry days (CDD), frequency of extremely heavy rainfall (R50mm) and annual/seasonal maximum of daily rainfall (RX1day). In general, changes in extreme indices are more pronounced and covering wider area under RCP8.5 than RCP4.5. The decrease in annual PRCPTOT is projected over most of SEA region, except for Myanmar and Northern Thailand, with magnitude as much as 20% (30%) under RCP4.5 (RCP8.5) scenario. The most significant and robust changes were noted in CDD, which is projected to increase by as much as 30% under RCP4.5 and 60% under RCP8.5, particularly over Maritime Continent (MC). The projected decrease in PRCPTOT over MC is significant and robust during June to August (JJA) and September to November (SON). During March to May (MAM) under RCP8.5, significant and robust PRCPTOT decreases are also projected over Indochina. The CDD changes during JJA and SON over MC are even higher, more robust and significant compared to the annual changes. At the same time, a wetting tendency is also projected over Indochina. The R50mm and RX1day are projected to increase, during all seasons with significant and robust signal of RX1day during JJA and SON.

Impacts of land surface model and land use data on WRF model simulations of rainfall and temperature over Lake Tana Basin, Ethiopia

The Weather Research and Forecasting (WRF) model is one of the regional climate models for dynamically downscaling climate variables at finer spatial and temporal scales. The objective of this study was to evaluate the performance of WRF model for simulating temperature and rainfall over Lake Tana basin in Ethiopia. The WRF model was configured for six experimental setups using three land surface models (LSMs): Noah, RUC and TD; and two land use datasets: USGS and updated New Land Use (NLU). The performances of WRF configurations were assessed by comparing simulated and observed data from March to August 2015. The result showed that temperature and rainfall simulations were sensitive to LSM and land use data choice. The combination of NLU with RUC and TD produced very small cold bias (0.27 °C) and warm bias (0.20 °C) for 2m maximum temperature (Tmax) and 2m minimum temperature (Tmin), respectively. WRF model with RUC and NLU captured well the observed spatial and temporal variability of Tmax, while TD and NLU for Tmin. Moreover, rainfall simulation was better with NLU; especially NLU and Noah configuration produced the smallest mean bias (2.39 mm/day) and root mean square error (6.6 mm/day). All the WRF experiments overestimated light and heavy rainfall events. Overall, findings showed that the application of updated land use data substantially improved the WRF model performance in simulating temperature and rainfall. The study would provide valuable support for identifying suitable LSM and land use data that can accurately predict the climate variables in the Blue Nile basin.

Impact of increased atmospheric moisture on the precipitation depth caused by Hurricane Ivan (2004) over a target area

In this article, numerical experiments are performed to investigate the effects of increasing atmospheric moisture on the precipitation depth (PD) produced by Hurricane Ivan (2004) over a target area, chosen as the drainage basin of the city of Asheville, NC. Atmospheric moisture was increased indirectly by increasing the sea surface temperature (SST) in the simulation initial conditions, and by letting the regional atmospheric model adjust the atmospheric fields to the SST perturbation. The SST was increased in two ways: 1) using spatially constant increments and 2) using a climate change perturbation field obtained from a climate projection. For each SST scenario, the PD over the target area was maximized by using a physically based storm transposition method. Although the mean PD, that was obtained by averaging over all shifting increments, increased with SST, the maximum PD was obtained for the case of no SST increase. It was found that, in the case of no SST increase, the worst-case tropical cyclone track was significantly different than in the SST increase scenarios. In particular, in this case, the storm spent a longer time in the simulation inner domain, thus spawning a larger PD over the target area.

Reconstruction and evaluation of changes in hydrologic conditions over a transboundary region by a regional climate model coupled with a physically-based hydrology model: Application to Thao river watershed

The differences among countries in terms of physical features, governmental policies, priorities in short- and long-term water resources management may lead to conflicts in managing and sharing of water resources over the transboundary regions. Due to no formal data sharing agreement between countries, there has been usually no data availability at transboundary regions. In this study, a methodology, in which a physically-based hydrology model was coupled with a regional climate model, is proposed to reconstruct and evaluate hydrologic conditions over transboundary regions. For the case study, Thao river watershed (TRW), within Vietnam and China, was selected. The Watershed Environmental Hydrology (WEHY) model was implemented based on topography, soil, and land use/cover information which was retrieved from global satellite data resources. The watershed model-WEHY was first validated over the TRW, and then was used to reconstruct historical hydrologic conditions during 1950-2007. The results of this study suggest no significant trend in the annual streamflow over the target watershed. In addition, there is a time shift in the wet season between the upstream sector in China and the downstream sector in Vietnam over the TRW. The annual flow contribution from the upstream sector in China to the outlet of TRW is estimated to be around 66%, and the remaining 34% contribution comes from the downstream sector in Vietnam territory. Last but not the least, the annual flow as a function of return period varies not only with the return period but also as a function of the time window, reflecting the effect of the changing regime on the streamflows at the TRW. The evolution of the flow frequency through time is an evidence of the ongoing non-stationarity in the hydrologic conditions over TRW.

Physically based storm transposition of four Atlantic tropical cyclones

In this article, a method for the storm transposition of tropical cyclones is presented. This method is physically based as it uses a regional atmospheric model to reconstruct the precipitation depth field from a tropical cyclone, thus crucially conserving the mass, momentum and energy in the system. In this physically based storm transposition method, the tropical cyclone vortex in the simulation initial conditions is first shifted spatially. More precisely, the tropical cyclone at the simulation start date is first separated from its background environment, then shifted, and finally recombined with the background environment. Afterwards, the regional atmospheric model is run as usual to simulate the shifted tropical cyclone and its precipitation depth field. The storm transposition method was then applied to four hurricanes which spawned torrential precipitation in the United States: Hurricanes Floyd (1999), Frances (2004), Ivan (2004), and Isaac (2012), in order to maximize the 72-h precipitation depth over the drainage basin of the city of Asheville, NC. It was observed that the precipitation depth fields changed in both structure and intensity after the physically based storm transposition. Besides, the tropical cyclone tracks were generally very sensitive to changes in the initial conditions, which is expected for a storm system whose dynamics is strongly nonlinear. In particular, it was found that a small change in the location of the initial tropical cyclone vortex may result in a very different track, allowing the tropical cyclone's precipitation depth field to move over the target area.

Numerical reconstruction of the intense precipitation and moisture transport fields for six tropical cyclones affecting the eastern United States

Tropical cyclones (TCs) are intense atmospheric vortices that form over the warm tropical oceans. They are recognized for their ability to generate intense precipitation that may in turn create disastrous floods. This article first assesses the suitability of a regional atmospheric model, the Weather Research and Forecasting (WRF) model, to simulate the intense precipitation depth (PD) fields of six North Atlantic TCs that affected the eastern United States during 2002-2016. Due to the strong nonlinearity involved in tropical cyclones' dynamics and thermodynamics, which causes tropical cyclones' tracks to be very sensitive to the different modeling choices, placing the PD fields in the observed locations was challenging. This involved trying several simulation start dates and combinations of the WRF model's parameterization schemes for each storm simulated. Model performance was evaluated by comparing the simulated PD fields with the observed PD fields obtained from the NCEP Stage IV precipitation dataset. In addition to qualitative comparisons, three quantitative metrics were used to quantify the WRF model performance in simulating a PD field's location, structure and intensity. The sensitivity of the simulation results to the choice of the parameterization schemes was then illustrated using Hurricane Gustav (2008). Eventually, the most satisfactory simulations were used to investigate the mechanisms responsible for the generation of intense precipitation in these TCs. More specifically, the vertically integrated vapor transport field and its divergence were calculated using the model outputs, and it was found that horizontal moisture convergence played a central role in the generation of intense precipitation in these TCs.

Analysis of future climate change impacts on snow distribution over mountainous watersheds in Northern California by means of a physically-based snow distribution model

The impacts of climate change on snow distribution through the 21st century were investigated over three mountainous watersheds in Northern California by means of a physically-based snow distribution model. The future climate conditions during a 90-year future period from water year 2010 to 2100 were obtained from 13 future climate projection realizations from two GCMs (ECHAM5 and CCSM3) based on four SRES scenarios (A1B, A1FI, A2, and B1). The 13 future climate projection realizations were dynamically downscaled at 9 km resolution by a regional climate model. Using the downscaled variables based on the 13 future climate projection realizations, snow distribution over the Feather, Yuba, and American River watersheds (FRW, YRW, and ARW) was projected by means of the physically-based snow model. FRW and YRW watersheds cover the main source areas of the California State Water Project (SWP), and ARW is one of the key watersheds in the California Central Valley Project (CVP). SWP and CVP are of great importance as they provide and regulate much of the California's water for drinking, irrigation, flood control, environmental, and hydro-power generation purposes. Ensemble average snow distribution over the study watersheds was calculated over the 13 realizations and for each scenario, revealing differences among the scenarios. While the snow reduction through the 21st century was similar between A1B and A2, the snow reduction was milder for B1, and more severe for A1FI. A significant downward trend was detected in the snowpack over nearly the entire watershed areas for all the ensemble average results.

Impacts of transportation sector emissions on future U.S. air quality in a changing climate. Part I: Projected emissions, simulation design, and model evaluation

Emissions from the transportation sector are rapidly changing worldwide; however, the interplay of such emission changes in the face of climate change are not as well understood. This two-part study examines the impact of projected emissions from the U.S. transportation sector (Part I) on ambient air quality in the face of climate change (Part II). In Part I of this study, we describe the methodology and results of a novel Technology Driver Model (see graphical abstract) that includes 1) transportation emission projections (including on-road vehicles, non-road engines, aircraft, rail, and ship) derived from a dynamic technology model that accounts for various technology and policy options under an IPCC emission scenario, and 2) the configuration/evaluation of a dynamically downscaled Weather Research and Forecasting/Community Multiscale Air Quality modeling system. By 2046-2050, the annual domain-average transportation emissions of carbon monoxide (CO), nitrogen oxides (NO_x), volatile organic compounds (VOCs), ammonia (NH₃), and sulfur dioxide (SO₂) are projected to decrease over the continental U.S. The decreases in gaseous emissions are mainly due to reduced emissions from on-road vehicles and non-road engines, which exhibit spatial and seasonal variations across the U.S. Although particulate matter (PM) emissions widely decrease, some areas in the U.S. experience relatively large increases due to increases in ship emissions. The on-road vehicle emissions dominate the emission changes for CO, NO_x, VOC, and NH₃, while emissions from both the on-road and non-road modes have strong contributions to PM and SO₂ emission changes. The evaluation of the baseline 2005 WRF simulation indicates that annual biases are close to or within the acceptable criteria for meteorological performance in the literature, and there is an overall good agreement in the 2005 CMAQ simulations of chemical variables against both surface and satellite observations.

Impacts of transportation sector emissions on future U.S. air quality in a changing climate. Part II: Air quality projections and the interplay between emissions and climate change

In Part II of this work we present the results of the downscaled offline Weather Research and Forecasting/Community Multiscale Air Quality (WRF/CMAQ) model, included in the "Technology Driver Model" (TDM) approach to future U.S. air quality projections (2046-2050) compared to a current-year period (2001-2005), and the interplay between future emission and climate changes. By 2046-2050, there are widespread decreases in future concentrations of carbon monoxide (CO), nitrogen oxides (NO_x = NO + NO₂), volatile organic compounds (VOCs), ammonia (NH₃), sulfur dioxide (SO₂), and particulate matter with an aerodynamic diameter ≤ 2.5 μm (PM_2.5) due mainly to decreasing on-road vehicle (ORV) emissions near urban centers as well as decreases in other transportation modes that include non-road engines (NRE). However, there are widespread increases in daily maximum 8-hr ozone (O₃) across the U.S., which are due to enhanced greenhouse gases (GHG) including methane (CH₄) and carbon dioxide (CO₂) under the Intergovernmental Panel on Climate Change (IPCC) A1B scenario, and isolated areas of larger reduction in transportation emissions of NO_x compared to that of VOCs over regions with VOC-limited O₃ chemistry. Other notable future changes are reduced haze and improved visibility, increased primary organic to elemental carbon ratio, decreases in PM_2.5 and its species, decreases and increases in dry deposition of SO₂ and O₃, respectively, and decreases in total nitrogen (TN) deposition. There is a tendency for transportation emission and CH₄ changes to dominate the increases in O₃, while climate change may either enhance or mitigate these increases in the west or east U.S., respectively. Climate change also decreases PM_2.5 in the future. Other variable changes exhibit stronger susceptibility to either emission (e.g., CO, NO_x, and TN deposition) or climate changes (e.g., VOC, NH₃, SO₂, and total sulfate deposition), which also have a strong dependence on season and specific U.S. regions.

Long-term trend analysis on total and extreme precipitation over Shasta Dam watershed

California's interconnected water system is one of the most advanced water management systems in the world, and understanding of long-term trends in atmospheric and hydrologic behavior has increasingly being seen as vital to its future well-being. Knowledge of such trends is hampered by the lack of long-period observation data and the uncertainty surrounding future projections of atmospheric models. This study examines historical precipitation trends over the Shasta Dam watershed (SDW), which lies upstream of one of the most important components of California's water system, Shasta Dam, using a dynamical downscaling methodology that can produce atmospheric data at fine time-space scales. The Weather Research and Forecasting (WRF) model is employed to reconstruct 159years of long-term hourly precipitation data at 3km spatial resolution over SDW using the 20th Century Reanalysis Version 2c dataset. Trend analysis on this data indicates a significant increase in total precipitation as well as a growing intensity of extreme events such as 1, 6, 12, 24, 48, and 72-hour storms over the period of 1851 to 2010. The turning point of the increasing trend and no significant trend periods is found to be 1940 for annual precipitation and the period of 1950 to 1960 for extreme precipitation using the sequential Mann-Kendall test. Based on these analysis, we find the trends at the regional scale do not necessarily apply to the watershed-scale. The sharp increase in the variability of annual precipitation since 1970s is also detected, which implies an increase in the occurrence of extreme wet and dry conditions. These results inform long-term planning decisions regarding the future of Shasta Dam and California's water system.

Trend analysis of watershed-scale precipitation over Northern California by means of dynamically-downscaled CMIP5 future climate projections

The impacts of climate change on watershed-scale precipitation through the 21st century were investigated over eight study watersheds in Northern California based on dynamically downscaled CMIP5 future climate projections from three GCMs (CCSM4, HadGEM2-ES, and MIROC5) under the RCP4.5 and RCP8.5 future climate scenarios. After evaluating the modeling capability of the WRF model, the six future climate projections were dynamically downscaled by means of the WRF model over Northern California at 9km grid resolution and hourly temporal resolution during a 94-year period (2006-2100). The biases in the model simulations were corrected, and basin-average precipitation over the eight study watersheds was calculated from the dynamically downscaled precipitation data. Based on the dynamically downscaled basin-average precipitation, trends in annual depth and annual peaks of basin-average precipitation during the 21st century were analyzed over the eight study watersheds. The analyses in this study indicate that there may be differences between trends of annual depths and annual peaks of watershed-scale precipitation during the 21st century. Furthermore, trends in watershed-scale precipitation under future climate conditions may be different for different watersheds depending on their location and topography even if they are in the same region.

Dynamically downscaling predictions for deciduous tree leaf emergence in California under current and future climate

Models that predict the timing of deciduous tree leaf emergence are typically very sensitive to temperature. However, many temperature data products, including those from climate models, have been developed at a very coarse spatial resolution. Such coarse-resolution temperature products can lead to highly biased predictions of leaf emergence. This study investigates how dynamical downscaling of climate models impacts simulations of deciduous tree leaf emergence in California. Models for leaf emergence are forced with temperatures simulated by a general circulation model (GCM) at ~200-km resolution for 1981-2000 and 2031-2050 conditions. GCM simulations are then dynamically downscaled to 32- and 8-km resolution, and leaf emergence is again simulated. For 1981-2000, the regional average leaf emergence date is 30.8 days earlier in 32-km simulations than in ~200-km simulations. Differences between the 32 and 8 km simulations are small and mostly local. The impact of downscaling from 200 to 8 km is ~15 % smaller in 2031-2050 than in 1981-2000, indicating that the impacts of downscaling are unlikely to be stationary.