Tag: Climate Change Downscaling Techniques

Abstract

In the age of big data, scientific progress is fundamentally limited by our capacity to extract critical information. Here, we map fine-grained spatiotemporal distributions for thousands of species, using deep neural networks (DNNs) and ubiquitous citizen science data. Based on 6.7 M observations, we jointly model the distributions of 2477 plant species and species aggregates across Switzerland with an ensemble of DNNs built with different cost functions. We find that, compared to commonly-used approaches, multispecies DNNs predict species distributions and especially community composition more accurately. Moreover, their design allows investigation of understudied aspects of ecology. Including seasonal variations of observation probability explicitly allows approximating flowering phenology; reweighting predictions to mirror cover-abundance allows mapping potentially canopy-dominant tree species nationwide; and projecting DNNs into the future allows assessing how distributions, phenology, and dominance may change. Given their skill and their versatility, multispecies DNNs can refine our understanding of the distribution of plants and well-sampled taxa in general.

Abstract

We describe and demonstrate a two-step approach for screening global climate models (GCMs) and produce robust annual and seasonal climate projections for Chile. First, we assess climate model simulations through a Past Performance Index (PPI) inspired by the Kling-Gupta Efficiency, which accounts for climatological averages, interannual variability, seasonal cycles, monthly probabilistic distribution, spatial patterns of climatological means, and the capability of the GCMs to reproduce teleconnection responses to El Niño Southern Oscillation (ENSO) and the Southern Annular Mode (SAM). The PPI formulation is flexible enough to include additional variables and evaluation metrics and weight them differently. Secondly, we use a recently proposed GCM classification based on model code genealogy to obtain a subset of independent model structures from the top 60% GCMs in terms of PPI values. We use this approach to evaluate 27 models from the sixth phase of the Coupled Model Intercomparison Project (CMIP6) and generate projections in five regions with very different climates across continental Chile. The results show that the GCM evaluation framework is able to identify pools of poor-performing and well-behaved models at each macrozone. Because of its flexibility, the model features that may be improved through bias correction can be excluded from the model evaluation process to avoid culling GCMs that can replicate other climate features and observed teleconnections. More generally, the results presented here can be used as a reference for regional studies and GCM selection for dynamical downscaling, while highlighting the difficulty in constraining precipitation and temperature projections.

Abstract

Wind erosion climatic erosivity is a measure of climatic conditions that affect wind erosion. Projecting wind erosion climatic erosivity is curcial for predicting future wind erosion risk. In this study, we employed dynamic downscaling outputs from the MPI-ESM1-2-HR model to project changes in wind erosion climatic erosivity over High Mountain Asia (HMA) from 2041 to 2060 under a middle-emission scenario (an additional radiative forcing of 4.5 W/m² by 2100). From 1995 to 2014, wind erosion climatic erosivity in HMA was high in the southwest, on the Qiangtang Plateau, and in the Qaidam Basin, exceeding 1 kg·m⁻¹ s⁻¹. Compared to the period 1995–2014, wind erosion climatic erosivity is projected to decrease by 0.5 kg·m⁻¹ s⁻¹ over the east of the Qiangtang Plateau and increase by approximately 1 kg·m⁻¹ s⁻¹ in the southwest of the HMA during 2041–2060 under the middle emission scenario. This increase in wind erosion climatic erosivity in the southwest of HMA is attributed to a projected rise in high-wind frequency for 2041–2060 compared to 1995–2014. Conversely, the decrease in wind erosion climatic erosivity in the east of the Qiangtang Plateau results from increased precipitation during 2041–2060, which mitigates the effects of increased high-wind frequencies. Given the growing risk of wind erosion in the southwest of the HMA, it’s essential to implement appropriate mitigation policies for the future.

Abstract

Aerosols influence weather and climate by interacting with radiation through absorption and scattering. These effects heavily rely on the optical properties of aerosols, which are mainly governed by attributes such as morphology, size distribution, and chemical composition. These attributes undergo continuous changes due to chemical reactions and aerosol micro-physics, resulting in significant spatio-temporal variations. Most atmospheric models struggle to incorporate this variability because they use pre-calculated tables to handle aerosol optics. This offline approach often leads to substantial errors in estimating the radiative impacts of aerosols along with posing significant computational burdens. To address this challenge, we introduce a computationally efficient and robust machine learning approach called MieAI. It allows for relatively inexpensive calculation of the optical properties of internally mixed aerosols with a log-normal size distribution. Importantly, MieAI fully incorporates the variability in aerosol chemistry and microphysics. Our evaluation of MieAI against traditional Mie calculations, using number concentrations from the ICOsahedral Nonhydrostatic model with Aerosol and Reactive Trace gases (ICON-ART) simulations, demonstrates that MieAI exhibits excellent predictive accuracy for aerosol optical properties. MieAI achieves this with errors well within 10%, and it operates more than 1000 times faster than the benchmark approach of Mie calculations. Due to its generalized nature, the MieAI approach can be implemented in any chemistry transport model which represents aerosol size distribution in the form of log-normally distributed internally mixed modes. This advancement has the potential to replace frequently employed look-up tables and plays a substantial role in the ongoing attempts to reduce uncertainties in estimating aerosol radiative forcing.

Abstract

We present unprecedented datasets of current and future projected weather files for building simulations in 15 major cities distributed across 10 climate zones worldwide. The datasets include ambient air temperature, relative humidity, atmospheric pressure, direct and diffuse solar irradiance, and wind speed at hourly resolution, which are essential climate elements needed to undertake building simulations. The datasets contain typical and extreme weather years in the EnergyPlus weather file (EPW) format and multiyear projections in comma-separated value (CSV) format for three periods: historical (2001–2020), future mid-term (2041–2060), and future long-term (2081–2100). The datasets were generated from projections of one regional climate model, which were bias-corrected using multiyear observational data for each city. The methodology used makes the datasets among the first to incorporate complex changes in the future climate for the frequency, duration, and magnitude of extreme temperatures. These datasets, created within the IEA EBC Annex 80 “Resilient Cooling for Buildings”, are ready to be used for different types of building adaptation and resilience studies to climate change and heatwaves.

Abstract

Land surface Temperature (LST) at high spatial resolution and at sub-daily scale is highly useful for monitoring evaporative stress in plants, heatwave events, and droughts. Spatial downscaling methods are often used to improve the spatial resolution of LST and Diurnal Temperature Cycle (DTC) models are available to estimate the diurnal variation in LST using limited multi-temporal satellite observations. In this paper, we propose a simple approach to estimate DTC at field scale combining spatial downscaling and DTC modelling. For downscaling the LST from medium-resolution sensors, we have compared three spatial downscaling techniques: Principal Component Regression based disaggregation, DisTrad disaggregation model and a Spatio Temporal Integrated Temperature Fusion Model (STITFM). The PCR-based disaggregation technique uses multiple fine-resolution auxiliary datasets such as vegetation indices, radar backscattering coefficient, etc. The downscaled LSTs from PCR and DisTrad were compared with the original fine-resolution LST from ECOSTRESS and Landsat. The spatially downscaled LST observations from all the three models were then used in the GOT01‑t_s DTC model to estimate the corresponding diurnal temperature cycle at fine resolution. The DTC estimated from the downscaled LSTs from all the three methods were compared with in situ DTC obtained from ground observations over four sites. The PCR technique using multiple indices captured the spatial and diurnal patterns of LST across four different sites, yielding a combined Root Mean Square Error (RMSE) of 2.48 K and 0.95 coefficient of determination (R²). The proposed approach can be potentially used to model the diurnal variability of land surface fluxes over different landscapes with finer spatial resolution.

Abstract

Quantum computing has recently emerged as a transformative technology. Yet, its promised advantages rely on efficiently translating quantum operations into viable physical realizations. Here we use generative machine learning models, specifically denoising diffusion models (DMs), to facilitate this transformation. Leveraging text conditioning, we steer the model to produce desired quantum operations within gate-based quantum circuits. Notably, DMs allow to sidestep during training the exponential overhead inherent in the classical simulation of quantum dynamics—a consistent bottleneck in preceding machine learning techniques. We demonstrate the model’s capabilities across two tasks: entanglement generation and unitary compilation. The model excels at generating new circuits and supports typical DM extensions such as masking and editing to, for instance, align the circuit generation to the constraints of the targeted quantum device. Given their flexibility and generalization abilities, we envision DMs as pivotal in quantum circuit synthesis, both enhancing practical applications and providing insights into theoretical quantum computation.

Abstract

Morocco is encountering record daily maximum temperatures, severe rainfall deficits, intense thunderstorms, droughts, and powerful wind gusts, causing significant harm to people and property. Therefore, it is crucial to understand the course of these occurrences and to determine to what extent the global climate models (GCMs) used to project climate can replicate rainfall before they can be used in downscaling or impact assessment studies. GCMs are essential tools for climate studies, but selecting the best-performing ones remains challenging. This study aims to assess the extent to which certain climate models from the Coupled Model Intercomparison Project’s 6th phase (CMIP6) reproduce the spatial and temporal variability of precipitation across Morocco between 1981 and 2014. Total monthly precipitation from the Climate Hazards Group Infrared Precipitation with Station data (CHIRPS) were used as observational references. We used six robust statistical metrics on monthly and annual scales, including relative bias, correlation coefficient, root means square error, relative error, Taylor diagram, and Kling–Gupta efficiency. The outcomes demonstrated that the ability of GCMs to simulate precipitation varied over space and time. The spatio-temporal properties of precipitation were well reproduced by all GCMs, with correlation values ranging from 0.78 to 0.87. The research also revealed that only a few models accurately captured the spatial patterns of the detected trends. According to the KGE metric, the GCM INM_CM5_0 is ranked first among the models with the highest KGE value (0.45), followed by GCM FGOALS_f3_L with a value of around 0.41. The study results can be applied to climate projections using CMIP6 under different IPCC scenarios.

Abstract

In the present study, the Soil and Water Assessment Tool was applied to determine the impacts of changing Land Use and Land Cover (LULC), Geophysical Fluid Dynamics Laboratory – Earth System Model Version 2, and Representative Concentration Pathways (RCP) 8.5 climate scenario on the monthly streamflow in the Subarnarekha basin of India. The results showed increased flow due to a reduction in agricultural area, a rise in built-up area, and a reduction in water bodies due to LULC change. In addition, lower annual precipitation and increased projected temperature were observed under RCP8.5. Although annual precipitation is decreasing, some components of the water balance are slightly increasing. From 2013 to 2020, surface flow increased by 98.85 mm and water yield decreased by 13.33 mm. However, in the climate change scenario, surface flow increased by 142.85 mm. Water yield decreased by 21.88 mm, lateral flow slightly decreased by 7.06 mm, and a further significant decrease 68.37% was noted in groundwater flow. The downward trend in groundwater flow is a serious concern, and therefore, more surface water storage structures must be planned to increase groundwater recharge and capture the increased surface flow. The model performance was statistically tested for NSE (Nash–Sutcliffe efficiency), R², and PBIAS (percent bias). During the calibration period and validation stages, NSE, R², and PBIAS were found to be 0.72, 0.83, and − 15.20%, and 0.85, 0.82, and − 27%, respectively, with the 2013 LULC map. The decreased monthly water availability and declining trend of winter rainfall need to be taken care of while planning the cropping pattern of the basin.

Abstract

A 29-year variable resolution climate simulation is conducted from January 1988 to December 2016 using the Model for Prediction Across Scale-Atmosphere (MPAS-A), with prescribed sea surface temperatures obtained from ERA-Interim reanalysis. The global variable resolution configuration employs a mesh refinement of 92–25 km centered over East Asia. Model validations against combined observed datasets highlight that MPAS-A demonstrated advantages over three selected Regional Climate Models (RCMs) in terms of the spatial distribution of precipitation and spatial variability of the near-surface air temperature but struggled with accurately depicting temporal precipitation patterns. MPAS-A’s anomalies in mid-latitude circulation and wave activity fluxes explained the weaker cold air activities during winter in eastern China and the northward shift of the Meiyu rain belt. Common issues with reference RCMs exist in MPAS-A, such as excessive zonal moisture transport over the ocean and unrealistic interannual variability over the northwest Pacific Ocean. The wet biases over the ocean are associated with systematically higher Convective Available Potential Energy (CAPE) for MPAS-A. However, the extreme rainfall indices such as R95pTOT and R99pTOT are not completely dominated by these wet biases and still exhibit reasonable results. This finding underscores the robustness and potential of the variable resolution (VR) approach in obtaining regional information within a single model framework.