Correction to: Streamflow-based evaluation of climate model sub-selection methods

Various methods are available for assessing uncertainties in climate impact studies. Among such methods, model weighting by expert elicitation is a practical way to provide a weighted ensemble of models for specific real-world impacts. The aim is to decrease the influence of improbable models in the results and easing the decision-making process. In this study both climate and hydrological models are analyzed and the result of a research experiment is presented using model weighting with the participation of 6 climate model experts and 6 hydrological model experts. For the experiment, seven climate models are a-priori selected from a larger Euro-CORDEX ensemble of climate models and three different hydrological models are chosen for each of the three European river basins. The model weighting is based on qualitative evaluation by the experts for each of the selected models based on a training material that describes the overall model structure and literature about climate models and the performance of hydrological models for the present period. The expert elicitation process follows a three-stage approach, with two individual elicitations of probabilities and a final group consensus, where the experts are separated into two different community groups: a climate and a hydrological modeller group. The dialogue reveals that under the conditions of the study, most climate modellers prefer the equal weighting of ensemble members, whereas hydrological impact modellers in general are more open for assigning weights to different models in a multi model ensemble, based on model performance and model structure. Climate experts are more open to exclude models, if obviously flawed, than to put weights on selected models in a relatively small ensemble. The study shows that expert elicitation can be an efficient way to assign weights to different hydrological models, and thereby reduce the uncertainty in climate impact. However, for the climate model ensemble, comprising seven models, the elicitation in the format of this study could only reestablish a uniform weight between climate models.

Climate modeling is a complex process, requiring accurate and complete metadata in order to identify, assess and use climate data stored in digital repositories. The preservation of such data is increasingly important given the development of ever-increasingly complex models to predict the effects of global climate change. The EU METAFOR project has developed a Common Information Model (CIM) to describe climate data and the models and modelling environments that produce this There is a wide degree of variability between different climate models and modelling groups. To accommodate this, the CIM has been designed to be highly generic and flexible, with extensibility built in. METAFOR describes the climate modelling process simply as activity undertaken using software on computers to produce data. This process has been described as separate UML packages (and, ultimately, XML schemas). This fairly generic structure can be paired with more specific controlled vocabularies in order to restrict the range of valid CIM instances. The CIM will aid digital preservation of climate models as it will provide an accepted standard structure for the model metadata. Tools to write and manage CIM instances, and to allow convenient and powerful searches of CIM databases,. Are also under development. Community buy-in of the CIM has been achieved through a continual process of consultation with the climate modelling community, and through the METAFOR team's development of a questionnaire that will be used to collect the metadata for the Intergovernmental Panel on Climate Change's (IPCC) Coupled Model Intercomparison Project Phase 5 (CMIP5) model runs.

GCMeval – An interactive tool for evaluation and selection of climate model ensembles

Assessment of hydrological parameter uncertainty versus climate projection spread on urban streamflow and floods

Regional climate modeling is a dynamical downscaling technique applied to the results of global climate models (GCMs) in order to acquire more information on climate simulations and climate change projections. GCMs and regional climate models (RCMs) have undergone considerable development over the past few decades, and both have increased in resolution. The higher‐resolution edge of RCMs compared to GCMs still remains, however. This has been demonstrated in a number of specific studies. As GCMs operate on relatively coarse resolutions, they do not resolve more variable land forms and similar features that shape regional‐scale climates. RCMs operate on higher resolutions than GCMs, by a factor of 2–10. Some RCMs now explore resolutions down to 1–5 km. This adds value in regions with variable orography, land–sea and other contrasts, as well as in capturing sharp, short‐duration and extreme events. In contrast, large‐scale and time‐averaged fields, not least over smooth terrain and on scales that have been already skillfully resolved in GCMs, are not much affected. RCMs also generate additional detail compared to GCMs when in climate projection mode. Compared to the present‐day climate for which observations exist, here the added value aspect is more complex to evaluate. Nevertheless, added value is meaningfully underlined when there is a clear physical context for it to appear in. In addition to climate modeling and model evaluation‐related added value considerations, a significant relevant aspect of added value is the provision of regional scale information, including climate change projections, for climate impact, adaptation, and vulnerability research. WIREs Clim Change 2016, 7:145–159. doi: 10.1002/wcc.378This article is categorized under: Climate Models and Modeling > Knowledge Generation with Models

Water vapour plays a key role in the Earth's energy balance. Almost 50% of the absorbed solar radiation at the surface is used to cool the surface, through evaporation, and warm the atmosphere, through release of latent heat. Latent heat is the single largest factor in warming the atmosphere and in transporting heat from low to high latitudes. Water vapour is also the dominant greenhouse gas and contributes to a warming of the climate system by some 24°C (Kondratev 1972). However, water vapour is a passive component in the troposphere as it is uniquely determined by temperature and should therefore be seen as a part of the climate feedback system. In this short overview, we will first describe the water on planet Earth and the role of the hydrological cycle: the way water vapour is transported between oceans and continents and the return of water via rivers to the oceans. Generally water vapour is well observed and analysed; however, there are considerable obstacles to observing precipitation, in particular over the oceans. The response of the hydrological cycle to global warming is far reaching. Because different physical processes control the change in water vapour and evaporation/precipitation, this leads to a more extreme distribution of precipitation making, in general, wet areas wetter and dry areas dryer. Another consequence is a transition towards more intense precipitation. It is to be expected that the changes in the hydrological cycle as a consequence of climate warming may be more severe that the temperature changes.

Global and regional climate model (GCM and RCM) outputs are often used as climate forcing for ecological impact models, and this potentially results in large cumulative errors because information and error are passed sequentially along the modeling chain from GCM to RCM to impact model. There are also a growing number of Earth system modeling platforms in which climate and ecological models are dynamically coupled, and in this case error amplification due to feedbacks can lead to even more serious problems. It is essential in both cases to rethink the organization of evaluation which typically relies on independent validation at each successive step, and to rely more heavily on analyses that cover the full modeling chain and thus require stronger interactions between climate and impact modelers. In this paper, we illustrate the benefits of using impact models as an additional source of information for evaluating climate models. Four RCMs that are part of the HyMeX (Hydrological cycle in Mediterranean EXperiment) and Mediterranean CORDEX projects (MED-CORDEX) were tested with observed climatology and a process-based model of European beech (Fagus sylvatica L.) tree growth and forest ecosystem functioning that has been rigorously validated. This two part analysis i) indicates that evaluation of RCMs on climate variables alone may be insufficient to determine the suitability of RCMs for studies of climate-forest interactions and ii) points to areas of improvement in these RCMs that would improve impact studies or behavior in coupled climate-ecosystem models over the spatial domain studied.

<p>The calculation of the near-surface turbulent fluxes of energy and momentum in climate and weather prediction models requires transfer coefficients. Currently used parametrizations of these coefficients are based on stability functions derived from measurements over land and not over sea ice. However, recently, a non-iterative parametrization has been proposed by Gryanik and Lüpkes (2018), which can be applied to climate and weather prediction models as well but uses stability functions of Grachev et al. (2007). These functions had been obtained from measurements during the Surface Heat Budget over the Arctic Ocean campaign (SHEBA) and thus from measurements over sea ice. A drawback of the scheme of Gryanik and Lüpkes (2018) is that there is still some complexity due to the complexity of the SHEBA based functions.</p><p>Thus new stability functions are proposed for the stable boundary layer, which are also based on the SHEBA measurements but avoid the complexity. It is shown that the new functions are superior to the former ones with respect to the representation of the measured relationship between the Obukhov length and the bulk Richardson number. Moreover, the resulting transfer coefficients agree slightly better with the SHEBA observations in the very stable range. Nevertheless, the functions fulfill the same criteria of applicability as the earlier functions and contain furthermore as an extension a dependence on the neutral Prandtl number. Applying the new functions, an efficient non-iterative parametrization of the near-surface turbulent fluxes of momentum and heat is developed where transfer coefficients result as a function of the bulk Richardson number (Ri<sub>b</sub>) and roughness parameters. The new transfer coefficients, which are recommended for weather and climate models, agree well with the SHEBA data in a large range of stability (0< Ri<sub>b</sub><0.5) and with those based on the Dyer-Businger functions in the range Ri<sub>b</sub> <0.08.</p><p><strong>References</strong></p><p><span>Grachev A.A., Andreas E.L, Fairall C.W., Guest P.S., Persson POG (2007) Boundary-Layer Meteorol., </span><span><strong>124</strong></span><span>, 315–333</span><span>.</span></p><p><span>Gryanik, V.M. and Lüpkes C. (2018) An Efficient Non-iterative Bulk Parametrization of Surface Fluxes for Stable Atmospheric Conditions Over Polar Sea-Ice,Boundary-Layer Meteorol 166:301-325</span></p>

Machine learning approaches, such as random forests (RF), have been used to effectively emulate various aspects of climate and weather models in recent years. The limitations to these approaches are not yet known, particularly with regards to varying complexity of the underlying physical parameterization scheme within the climate model. Utilizing a hierarchy of model configurations, we explore the limits of random forest emulator skill using simplified model frameworks within NCAR's Community Atmosphere Model, version 6 (CAM6). These include a dry CAM6 configuration, a moist extension of the dry model, and an extension of the moist case that includes an additional convection scheme. Each model configuration is run with identical resolution and over the same time period. With unique RF being optimized for each tendency or precipitation rate across the hierarchy, we create a variety of “best case” emulators. The random forest emulators are then evaluated against the CAM6 output as well as a baseline neural network emulator for completeness. All emulators show significant skill when compared to the “truth” (CAM6), often in line with or exceeding similar approaches within the literature. In addition, as the CAM6 complexity is increased, the random forest skill noticeably decreases, regardless of the extensive tuning and training process each random forest goes through. This indicates a limit on the feasibility of RF to act as physics emulators in climate models and encourages further exploration in order to identify ideal uses in the context of state‐of‐the‐art climate model configurations.

<p>Atmospheric radiative transfer, which describes the evolution of radiation emitted by the Sun, the Earth's surface, clouds, and greenhouse gases, is an essential component of climate and weather modeling. In climate models, the transfer of radiation is approximated by parameterizations. Theoretically, however, with sufficient computing power, the electromagnetic radiation equations could be solved, but in practice this is out of reach.  The current operational radiative transfer solver in the Icosahedral Nonhydrostatic Weather and Climate Model (ICON) is ecRad, which, developed at ECMWF, is one of the most advanced available radiative transfer parameterizations.  It considers surface optics, gas optics, aerosol optics and cloud optics [1]. It is an accurate radiation parametrization but remains computationally expensive. Therefore, the radiation solver is usually not invoked at every time step and only runs on a reduced spatial grid, which can affect prediction accuracy, or only in a 1D setting without 3D transfer.</p><p>In this project, we are trying to develop an ecRad solver improved by machine learning to speed up the computation without loss of accuracy. Machine learning-based parametrizations would in general allow to fully replace existing sub-grid scale parameterizations, once trained from data. However, such parametrizations do not necessarily preserve essential physical quantities, which can lead to instabilities, model drifts or unphysical behavior as observed in [2] and [3].</p><p>We present here an emulation strategy, composed of three steps. First, we continue to call ecRad on a significantly coarser grid to predict the clear-sky radiation. We thereby use ecRad as a regularizer while reducing computation costs. Then, we interpolate the data on the full spatial grid using Gaussian processes. Finally, we predict the effect of the clouds on the radiation with random forests. The underlying idea is to avoid unphysical climate drifts and to support the generalization capabilities of the ML method.</p><p>Our first numerical experiments on an aqua planet simulation are promising. We hope to obtain a valuable outcome when considering more complex datasets with seasonality and realistic topography. Our final goal is to run a full ICON simulation with a machine learning enhanced ecRad parametrization, though the online performance remains open. There the clear-sky low resolution radiation field, computed with ecRad in the first part of our strategy, is expected to play a central role in model stability.</p><p>[1] <em>Hogan, R. J., & Bozzo, A. (2018). A flexible and efficient radiation scheme for the ECMWF model. Journal of Advances in Modeling Earth Systems, 10, 1990-2008. </em></p><p>[2] <em>Brenowitz, N. D., & Bretherton, C. S. (2018). Prognostic validation of a neural network unified physics parametrization. Geophysical Research Letters, 17, 6289–6298</em></p><p>[3] <em>Brenowitz, N. D., and Bretherton, C. S. (2019). Spatially Extended Tests of a Neural Network Parametrization Trained by Coarse‐Graining, J. Adv. Model. Earth Syst., 11, 2728–2744</em></p>

Motivated by recent advances in operational weather forecasting, we study the efficacy of low-precision arithmetic for climate simulations. We develop a framework to measure rounding error in a climate model, which provides a stress test for a low-precision version of the model, and we apply our method to a variety of models including the Lorenz system, a shallow water approximation for flow over a ridge, and a coarse-resolution spectral global atmospheric model with simplified parameterizations (SPEEDY). Although double precision [52 significant bits (sbits)] is standard across operational climate models, in our experiments we find that single precision (23 sbits) is more than enough and that as low as half precision (10 sbits) is often sufficient. For example, SPEEDY can be run with 12 sbits across the code with negligible rounding error, and with 10 sbits if minor errors are accepted, amounting to less than 0.1 mm (6 h)−1 for average gridpoint precipitation, for example. Our test is based on the Wasserstein metric and this provides stringent nonparametric bounds on rounding error accounting for annual means as well as extreme weather events. In addition, by testing models using both round-to-nearest (RN) and stochastic rounding (SR) we find that SR can mitigate rounding error across a range of applications, and thus our results also provide some evidence that SR could be relevant to next-generation climate models. Further research is needed to test if our results can be generalized to higher resolutions and alternative numerical schemes. However, the results open a promising avenue toward the use of low-precision hardware for improved climate modeling. Significance Statement Weather and climate models provide vital information for decision-making, and will become ever more important in the future with a changed climate and more extreme weather. A central limitation to improved models are computational resources, which is why some weather forecasters have recently shifted from conventional 64-bit to more efficient 32-bit computations, which can provide equally accurate forecasts. Climate models, however, still compute in 64 bits, and adapting to lower precision requires a detailed analysis of rounding errors. We develop methods to quantify rounding error in a climate model, and find similar precision acceptable across weather and climate models, with even 16 bits often sufficient for an accurate climate. This opens a promising avenue for computational efficiency gains in climate modeling.

Transitioning to renewable energy will be instrumental in mitigating the devastating effects of climate change. Because of the many unknowns in the design and dispatch of future energy systems, quantifying climate risk in the energy sector is challenging: in particular, renewable energy production and heating demand is highly reliant on meteorological conditions, which are variable in nature and shifting due to climate change.It is therefore important to use large samples of renewable generation and demand, for current and future climates, in energy system modeling. However, lacking standardized ways to translate between the climate and energy model world, most existing studies rely on different assumptions and draw from a limited sample of available climate variables.To this end, we created a modular climate-to-energy pipeline: starting with hourly output from the climate model CESM2, it bias corrects, translates, and scales to the various inputs of energy system models. We base the conversion on open-source tools: GSEE for solar power generation, windpowerlib for wind from climate model levels and demand.ninja for heating and cooling demand. The resulting pipeline ensures consistency of variables, with inputs and outputs tailorable to specific needs.We use the pipeline to analyze seasonal cycles of energy generation and demand under different weather conditions, for current and future climates deploying the AnyMOD.jl framework for energy system modeling. Because of the modular approach, the pipeline could easily be adapted for other climate models and time-series, providing better evidence for climate-informed energy system planning.

The future bloom and risk of blossom frosts for Malus domestica were projected using regional climate realizations and phenological ( = impact) models. As climate impact projections are susceptible to uncertainties of climate and impact models and model concatenation, the significant horizon of the climate impact signal was analyzed by applying 7 impact models, including two new developments, on 13 climate realizations of the IPCC emission scenario A1B. Advancement of phenophases and a decrease in blossom frost risk for Lower Saxony (Germany) for early and late ripeners was determined by six out of seven phenological models. Single model/single grid point time series of bloom showed significant trends by 2021–2050 compared to 1971–2000, whereas the joint signal of all climate and impact models did not stabilize until 2043. Regarding blossom frost risk, joint projection variability exceeded the projected signal. Thus, blossom frost risk cannot be stated to be lower by the end of the 21st century despite a negative trend. As a consequence it is however unlikely to increase. Uncertainty of temperature, blooming date and blossom frost risk projection reached a minimum at 2078–2087. The projected phenophases advanced by 5.5 d K−1, showing partial compensation of delayed fulfillment of the winter chill requirement and faster completion of the following forcing phase in spring. Finally, phenological model performance was improved by considering the length of day.

It is well known that the results of climate model simulations differ from observations, and accordingly such projections can lead to different future temperature characteristics for the same region of interest. While the relative changes described by different climate models are of similar magnitude, the absolute temperature characteristics (based on their relative performance compared to observations) can be so different. At different warming levels (WLs, where a given temperature threshold can be calculated from temperature projections relative to a reference period or observations) global climate models (GCMs) can provide valuable information on climate change at the global scale. The main uncertainty factor for different WLs is time relative to a reference period (e.g. 1976-2005). Moving from global to regional scales, from coarser (~150 km) to finer (~12 km) resolution, regional climate models (RCMs) are expected to provide more detailed information than GCMs. For example, RCMs can better capture precipitation extremes at finer resolutions than the driving GCMs, especially over regions with complex topography, illustrating the benefits of high-resolution modelling. Thus, the timing of reaching a given WL can be assessed at regional and local scales based on high-resolution RCM simulations and high-resolution observations available for the region of interest. The REtuning Climate Model Outputs (RECMO) method helps to reduce differences between different RCM simulations by reducing uncertainties arising from the different climates described by different climate models under different WLs. It should be noted that the reference is based on observations and not on the model outputs from which the WLs are determined. It can be argued that the RECMO method may be too selective depending on the performance of the climate models over the region (as RCMs with a relatively large cool bias may never reach that WL). Accordingly, the high-resolution raw and bias-corrected EURO-CORDEX and Med-CORDEX outputs are also used in the present work following the RECMO method for the Carpathian Region. Overall, the new concept helps to reduce differences between climate model simulations and leads to more reliable results with respect to possible future climatic temperature conditions for the region of interest. It should be emphasized that the concept is generally applicable to climate models over any region of the globe, the only limitation being the availability of data.

A crucial branch of science called climate modeling uses mathematical equations and computer simulations to study and forecast the Earth's climate sys- tem. The main elements of climate modeling, such as general circulation models (GCMs), data assimilation methods, and numerical formulations, are outlined in this paper. GCMs, which include grid point and spectral models, are effective instruments for examining the behavior of the climate. Four-Dimensional Data Assimilation (4D-Var) is one example of a data assimilation technique that in- corporates observational data into models to improve their correctness. Numeri cal methods, ocean dynamics, heat transport, radiative transfer, and atmospheric dynamics are all included in numerical formulations. The simulation of different climate processes is possible because to these mathematical representations. Fur thermore, the detection of precipitation patterns within climate modeling is using machine learning techniques like Random Forest more frequently. This paper highlights the importance of high-performance computing (HPC) in climate modeling, boosting efficiency and simulations, in the context of research technique. Advanced data assimilation and validation techniques are also examined, as well as the influence of high-resolution modeling on small-scale climatic processes. On HPC platforms, accessibility to climate modeling is addressed. It is shown how climate modeling crosses physics, mathematics, computer science, and engineering to be interdisciplinary. A comprehensive understanding of the Earth's intricate climate system gains from the integration of all its parts, from atmospheric dynamics to data assimilation. We explore the consequences of these research approaches, their contribution to enhancing climate prediction models, and the influence of various factors on climatic variables in the debate. Climate modeling becomes an essential tool for studying precipitation patterns and climate change, ultimately improving our comprehension of the complex cli- mate system on Earth.