Statistical emulation of high-resolution SAR wind fields from low-resolution model predictions

AbstractThe influence of a high horizontal resolution (5–15 km) on the general circulation and hydrography in the North Atlantic is investigated using the Finite Element Sea Ice–Ocean Model (FESOM). We find a stronger shift of the upper-ocean circulation and water mass properties during the model spinup in the high-resolution model version compared to the low-resolution (~1°) control run. In quasi equilibrium, the high-resolution model is able to reduce typical low-resolution model biases. Especially, it exhibits a weaker salinification of the North Atlantic subpolar gyre and a reduced mixed layer depth in the Labrador Sea. However, during the spinup adjustment, we see that initially improved high-resolution features partially reduce over time: the strength of the Atlantic overturning and the path of the North Atlantic Current are not maintained, and hence hydrographic biases known from low-resolution ocean models return in the high-resolution quasi-equilibrium state. We identify long baroclinic Rossby waves as a potential cause for the strong upper-ocean adjustment of the high-resolution model and conclude that a high horizontal resolution improves the state of the modeled ocean but the model integration length should be chosen carefully.

Download Full-text

Evaluating seasonal hydrological extremes in mesoscale (pre-)Alpine basins at coarse 0.5° and fine hyperresolution

Hydrology and Earth System Sciences ◽

10.5194/hess-23-1593-2019 ◽

2019 ◽

Vol 23 (3) ◽

pp. 1593-1609 ◽

Cited By ~ 1

Author(s):

Joost Buitink ◽

Remko Uijlenhoet ◽

Adriaan J. Teuling

Keyword(s):

High Resolution ◽

Land Cover ◽

Swiss Alps ◽

Hydrological Response ◽

Low Resolution ◽

Continental Scale ◽

Wide Range ◽

Resolution Model ◽

High Resolution Model ◽

Basin Response

Abstract. Hydrological models are being applied for impact assessment across a wide range of resolutions. In this study, we quantify the effect of model resolution on the simulated hydrological response in five mesoscale basins in the Swiss Alps using the distributed hydrological model Spatial Processes in Hydrology (SPHY). We introduce a new metric to compare a range of values resulting from a distributed model with a single value: the density-weighted distance (DWD). Model simulations are performed at two different spatial resolutions, matching common practices in hydrology: 500 m × 500 m matching regional-scale models, and 40 km × 40 km matching global-scale modeling. We investigate both the intra-basin response in seasonal streamflow and evapotranspiration from the high-resolution model and the difference induced by the two different spatial resolutions, with a focus on four seasonal extremes, selected based on temperature and precipitation. Results from the high-resolution model show that the intra-basin response covers a surprisingly large range of anomalies and show that it is not uncommon to have both extreme positive and negative flux anomalies occurring simultaneously within a catchment. The intra-basin response was grouped by land cover, where different dominant runoff-generating processes are driving the differences between these groups. The low-resolution model failed to capture the diverse and contrasting response from the high-resolution model, since neither the complex topography nor land cover classes were properly represented. DWD values show that, locally, the hydrological response simulated with a high-resolution model can be a lot more extreme than a low-resolution model might indicate, which has important implications for global or continental scale assessments carried out at coarse grids of 0.5∘×0.5∘ or 0.25∘×0.25∘ resolution.

Download Full-text

Resolution Dependency of Future Caribbean Sea Level Response

10.5194/egusphere-egu2020-3381 ◽

2020 ◽

Author(s):

René van Westen ◽

Henk Dijkstra

Keyword(s):

High Resolution ◽

Sea Level Rise ◽

Sea Level ◽

Climate Models ◽

Caribbean Sea ◽

Model Output ◽

Low Resolution ◽

Ocean Eddies ◽

Resolution Model ◽

Mesoscale Processes

<div> <div> <div> The current global climate models, which are often used in inter-comparison projects, have a large variety in their spatial resolution. For most climate models, the resolution of the ocean grid does not allow to resolve mesoscale processes such as ocean eddies. Current sea level projections are based on these coarse climate models, but might have biases (either positive or negative) in these projections since mesoscale processes are parameterised. Here we investigate the differences in future Caribbean sea level rise using a centennial simulation of a high- and low-resolution version of the Community Earth System Model under the same anthropogenic forcing. In the high-resolution version of the model mesoscale processes are resolved. Locally, we find a decrease of 7.2 cm in sea level extremes over a 100-year period in the high-resolution version; this decrease is almost absent in the low-resolution version. This local decrease in sea level extremes is related to ocean eddies, which are not resolved in the low-resolution version, hence explaining the different sea level response between the models. When comparing modelled sea level trends to observed sea level trends over the past 25 years, we find a reasonable agreement between observations and the high-resolution model. However, for the low-resolution model and some of the preliminary CMIP6 model output, there is a substantial mismatch between the observed- and modelled sea level trends. By analysing model output from two different resolutions of the same climate model, we find that the sea level response in the Caribbean Sea is resolution-dependent. As a result, not resolving mesoscale processes in climate models can locally result in overestimations of future sea level rise projections. </div> </div> </div>

Download Full-text

Potential Underestimation of Future Mei-Yu Rainfall with Coarse-Resolution Climate Models

Journal of Climate ◽

10.1175/jcli-d-17-0741.1 ◽

2018 ◽

Vol 31 (17) ◽

pp. 6711-6727 ◽

Cited By ~ 5

Author(s):

Xiaolong Chen ◽

Peili Wu ◽

Malcolm J. Roberts ◽

Tianjun Zhou

Keyword(s):

High Resolution ◽

Large Scale ◽

Climate Models ◽

Climate Model ◽

Summer Rainfall ◽

Climate Projections ◽

Model Resolution ◽

Low Resolution ◽

Resolution Model ◽

High Resolution Model

The amount of rainfall during June and July along the mei-yu front contributes about 45% to the total summer precipitation over the Yangtze River valley. How it will change under global warming is of great concern to the people of China because of its particular socioeconomic importance, but climate model projections from phase 5 of the Coupled Model Intercomparison Project (CMIP5) show large uncertainties. This paper examines model resolution sensitivity and reports large differences in projected future summer rainfall along the mei-yu front between a low-resolution (Gaussian N96 grid, ~1.5° latitude–longitude) and a high-resolution (N216, ~0.7°) version of the Hadley Centre’s latest climate model, the HadGEM3 Global Coupled Configuration 2.0 (HadGEM3-GC2). The high-resolution model projects large increases of summer rainfall under two representative concentration pathway scenarios (RCP8.5 and RCP4.5) whereas the low-resolution model shows a decrease. A larger increase of projected mei-yu rainfall in higher-resolution models is also observed across the CMIP5 ensemble. These differences can be explained in terms of enhanced moist static energy advection and moisture convergence by stationary eddies in the high-resolution model. A large-scale manifestation of the anomalous stationary eddies is the contrasting response to the same warming scenario by the western North Pacific subtropical high, which is almost unchanged in N216 but retreats evidently eastward in N96, reducing the southwesterly flow and consequently moisture supply to the mei-yu front. Further increases in model resolution to resolve parameterized processes and detailed orographic features will hopefully reduce the spread in future climate projections.

Download Full-text

Hydrological response to climate extremes in mesoscale (pre-)Alpine basins at 0.5° and hyperresolution

10.5194/hess-2017-629 ◽

2017 ◽

Author(s):

Joost Buitink ◽

Remko Uijlenhoet ◽

Adriaan J. Teuling

Keyword(s):

Land Use ◽

High Resolution ◽

Climate Extremes ◽

Swiss Alps ◽

Distributed Model ◽

Distributed Hydrological Model ◽

Hydrological Response ◽

Low Resolution ◽

Resolution Model ◽

High Resolution Model

Abstract. The response of key hydrological variables to climate extremes within five meso-scale basins in the Swiss Alps is investigated at two different resolutions using the distributed hydrological model Spatial Processes in Hydrology (SPHY). Based on elevation and presence of glaciers, three catchments are identified as Alpine and two as pre-Alpine. We run SPHY both at hyperresolution and at 0.5 × 0.5 degree, and aggregate simulated runoff and evapotranspiration per season. For four seasonal extremes representing flood and drought/heatwave conditions we investigate the simulated response at both model resolutions. Results from the high resolution model show that the within-basin response gets more complex with more extreme events. The response within each basin can be grouped per land use type, due to different dominant runoff generating processes. A comparison with the coarse resolution model results shows that there is a large discrepancy between the two simulated responses. The low resolution model is not able to correctly simulate the complex hydrological response as simulated with the distributed model, since both the complex topography and land use classes are not properly represented. We show that hydrological response simulated with a high resolution model can be a lot more extreme than a low resolution model might indicate, which has important implications for global assessments carried out at course resolution.

Download Full-text

Estimate hydrodynamic connectivity and probability of contamination through Lagrangian experiments in a high resolution shelf sea model

10.5194/egusphere-egu2020-20673 ◽

2020 ◽

Author(s):

Michele Bendoni ◽

Carlo Brandini ◽

Maria Fattorini ◽

Chiara Lapucci ◽

Carlo Pretti

Keyword(s):

High Resolution ◽

Initial Conditions ◽

Low Resolution ◽

Reanalysis Dataset ◽

North West ◽

The North ◽

Port Area ◽

Shelf Sea ◽

Resolution Model ◽

One Year

Coastal areas are experiencing an increasing anthropic pressure worldwide, especially due to port activities. In addition, valuable ecosystems such as Marine Protected Areas (MPA) might be located close to ports and be potentially subject to pollutant driven by the local current pattern. It is then fundamental to develop tools to analyze and quantify the tendency of a MPA to be affected by generic pollutant released from a port. Present work is based on a series of Lagrangian experiments carried out on a domain containing the port of Livorno and the Meloria Sholas MPA, located in the Tuscany Archipelago (Italy). The flow field employed to force the experiments is obtained from a downscaling modelling chain implemented with the 3D ROMS software. The top level is a 1.2 km low-resolution model covering the North-West portion of the Mediterranean basin which feeds with a one-way nesting algorithm a 400 m mid-resolution model for the Tuscany Archipelago, extending West of Corsica Island and up to the Gulf of Genova. The inner level of the modelling chain is a 50 m high-resolution coastal model (offline nesting) which covers the area of Meloria Shoals, the port, and their surroundings. Hydrodynamic simulations are carried out for one year. Initial conditions are provided by the CMEMS (1/24&#176; res) model Analysis, as well as boundary conditions for the low-resolution model. Atmospheric forcing comes from the downscaling of the ERA-5 reanalysis dataset, consisting on the BOLAM model implemented on a 7 km grid of the Med-CORDEX domain, in which the MOLOCH model is nested on a 2.5 km spaced grid. Lagrangian numerical experiments are carried out considering the consecutive release of passive particles in the port area, at finite intervals for one year, following the trajectories for ten days. To estimate the degree of hydrodynamic connectivity between the port and the MPA and give a measure of the probability of contamination, the &#8220;oceanographic distance&#8221; is computed in several ways from the calculated trajectories. Preliminary results show the main transport pattern is mostly distributed alongshore, making the MPA less connected to the port compared to areas placed at the same distance.

Download Full-text

Evaluating seasonal hydrological extremes in mesoscale (pre-)Alpine basins at coarse 0.5° and fine hyperresolution

10.5194/hess-2018-407 ◽

2018 ◽

Author(s):

Joost Buitink ◽

Remko Uijlenhoet ◽

Adriaan J. Teuling

Keyword(s):

High Resolution ◽

Land Cover ◽

Regional Scale ◽

Swiss Alps ◽

Distributed Hydrological Model ◽

Hydrological Response ◽

Low Resolution ◽

Resolution Model ◽

High Resolution Model ◽

Basin Response

Abstract. In this study, we investigate the effect of model resolution on the simulated hydrological response in five mesoscale basins in the Swiss Alps using the distributed hydrological model Spatial Processes in Hydrology (SPHY). Model simulations are performed at resolutions matching regional scale (500 × 500 m, also matching hyperresolution) and global scale modeling (40 × 40 km, matching a 0.5 × 0.5° pixel). The simulated response is investigated for four seasonal extremes, selected based on temperature and precipitation anomalies. Results from the high resolution model show that the intra-basin response covers a large range of anomalies, often with contrasting anomaly signs. The intra-basin response was grouped by land cover, where different dominant runoff generating processes are driving the differences between these groups. The low resolution model failed to capture the diverse and contrasting response from the high resolution model, since both the complex topography and land cover classes were not properly represented. We show that the hydrological response simulated with a high resolution model can be a lot more extreme than a low resolution model might indicate, which has important implications for global assessments carried out at their typical 0.5 × 0.5° resolution.

Download Full-text

Controls of Global Snow under a Changed Climate

Journal of Climate ◽

10.1175/jcli-d-12-00528.1 ◽

2013 ◽

Vol 26 (15) ◽

pp. 5537-5562 ◽

Cited By ~ 69

Author(s):

Sarah B. Kapnick ◽

Thomas L. Delworth

Keyword(s):

High Resolution ◽

Regional Scale ◽

Geophysical Fluid Dynamics Laboratory ◽

Positive Contribution ◽

Total Precipitation ◽

High Latitudes ◽

Present Climate ◽

Low Resolution ◽

Resolution Model ◽

High Resolution Model

Abstract This study assesses the ability of a newly developed high-resolution coupled model from the Geophysical Fluid Dynamics Laboratory to simulate the cold-season hydroclimate in the present climate and examines its response to climate change forcing. Output is assessed from a 280-yr control simulation that is based on 1990 atmospheric composition and an idealized 140-yr future simulation in which atmospheric carbon dioxide increases at 1% yr−1 until doubling in year 70 and then remains constant. When compared with a low-resolution model, the high-resolution model is found to better represent the geographic distribution of snow variables in the present climate. In response to idealized radiative forcing changes, both models produce similar global-scale responses in which global-mean temperature and total precipitation increase while snowfall decreases. Zonally, snowfall tends to decrease in the low to midlatitudes and increase in the mid- to high latitudes. At the regional scale, the high- and low-resolution models sometimes diverge in the sign of projected snowfall changes; the high-resolution model exhibits future increases in a few select high-altitude regions, notably the northwestern Himalaya region and small regions in the Andes and southwestern Yukon, Canada. Despite such local signals, there is an almost universal reduction in snowfall as a percent of total precipitation in both models. By using a simple multivariate model, temperature is shown to drive these trends by decreasing snowfall almost everywhere while precipitation increases snowfall in the high altitudes and mid- to high latitudes. Mountainous regions of snowfall increases in the high-resolution model exhibit a unique dominance of the positive contribution from precipitation over temperature.

Download Full-text

High-resolution Ensemble Kalman Fiter with a low-resolution model using a machine learning super-resolution approach.

10.5194/egusphere-egu21-2585 ◽

2021 ◽

Author(s):

Sébastien Barthélémy ◽

Julien Brajard ◽

Laurent Bertino

Keyword(s):

Neural Network ◽

Machine Learning ◽

High Resolution ◽

Data Assimilation ◽

Ensemble Kalman Filter ◽

Super Resolution ◽

Low Resolution ◽

Resolution Model ◽

The Cost ◽

Resolution Data

Going from low- to high-resolution models is an efficient way to improve the data assimilation process in three ways: it makes better use of high-resolution observations, it represents more accurately the small scale features of the dynamics and it provides a high-resolution field that can further be used as an initial condition of a forecast. Of course, the pitfall of such an approach is the cost of computing a forecast with a high-resolution numerical model. This drawback is even more acute when using an ensemble data assimilation approach, such as the ensemble Kalman filter, for which an ensemble of forecasts is to be issued by the numerical model.In our approach, we propose to use a cheap low-resolution model to provide the forecast while still performing the assimilation step in a high-resolution space. The principle of the algorithm is based on a machine learning approach: from a low-resolution forecast, a neural network (NN) emulates a high-resolution field that can then be used to assimilate high-resolution observations. This NN super-resolution operator is trained on one high-resolution simulation. This new data assimilation approach denoted "Super-resolution data assimilation" (SRDA), is built on an ensemble Kalman filter (EnKF) algorithm.We applied SRDA to a quasi-geostrophic model representing simplified ocean dynamics of the surface layer, with a low-resolution up to four times smaller than the reference high-resolution (so the cost of the model is divided by 64). We show that this approach outperforms the standard low-resolution data assimilation approach and the SRDA method using standard interpolation instead of a neural network as a super-resolution operator. For the reduced cost of a low-resolution model, SRDA provides a high-resolution field with an error close to that of the field that would be obtained using a high-resolution model.

Download Full-text

Towards Operational Downscaling of Low Resolution Wind Fields using Neural Networks

10.5194/egusphere-egu2020-5447 ◽

2020 ◽

Author(s):

Michael Kern ◽

Kevin Höhlein ◽

Timothy Hewson ◽

Rüdiger Westermann

Keyword(s):

Neural Networks ◽

High Resolution ◽

Wind Field ◽

Input Data ◽

Vector Fields ◽

Wind Vector ◽

Proof Of Concept ◽

Low Resolution ◽

Wind Fields ◽

Resolution Simulation

Numerical weather prediction models with high resolution (of order kms or less) can deliver very accurate low-level winds. The problem is that one cannot afford to run simulations at very high resolution over global or other large domains for long periods because the computational power needed is prohibitive.Instead, we propose using neural networks to downscale low-resolution wind-field simulations (input) to high-resolution fields (targets) to try to match a high-resolution simulation. Based on short-range forecasts of wind fields (at the 100m level) from the ECMWF ERA5 reanalysis, at 31km resolution, and the HRES (deterministic) model version, at 9km resolution, we explore two complementary approaches, in an initial &#8220;proof-of-concept&#8221; study.In a first step, we evaluate the ability of U-Net-type convolutional neural networks to learn a one-to-one mapping of low-resolution input data to high-resolution simulation results. By creating a compressed feature-space representation of the data, networks of this kind manage to encode important flow characteristics of the input fields and assimilate information from additional data sources. Next to wind vector fields, we use topographical information to inform the network, at low and high resolution, and include additional parameters that strongly influence wind-field prediction in simulations, such as vertical stability (via the simple, compact metric of boundary layer height) and the land-sea mask. We thus infer weather-situation and location-dependent wind structures that could not be retrieved otherwise.In some situations, however, it will be inappropriate to deliver only a single estimate for the high-resolution wind field. Especially in regions where topographic complexity fosters the emergence of complex wind patterns, a variety of different high-resolution estimates may be equally compatible with the low-resolution input, and with physical reasoning. In a second step, we therefore extend the learning task from optimizing deterministic one-to-one mappings to modelling the distribution of physically reasonable high-resolution wind-vector fields, conditioned on the given low-resolution input. Using the framework of conditional variational autoencoders, we realize a generative model, based on convolutional neural networks, which is able to learn the conditional distributions from data. Sampling multiple estimates of the high-resolution wind vector fields from the model enables us to explore multimodalities in the data and to infer uncertainties in the predictand.In a future customer-oriented extension of this proof-of-concept work, we would envisage using a target resolution higher than 9km - say in the 1-4km range - to deliver much better representivity for users. Ensembles of low resolution input data could also be used, to deliver as output an &#8220;ensemble of ensembles&#8221;, to condense into a meaningful probabilistic format for users. The many exciting applications of this work (e.g. for wind power management) will be highlighted.

Download Full-text