scholarly journals Applying machine learning for drought prediction in a perfect model framework using data from a large ensemble of climate simulations

2021 ◽  
Vol 21 (12) ◽  
pp. 3679-3691
Author(s):  
Elizaveta Felsche ◽  
Ralf Ludwig
Keyword(s):  
Machine Learning ◽  
Lead Time ◽  
Climate Model ◽  
Regional Climate ◽  
Standardized Precipitation Index ◽  
Machine Learning Algorithms ◽  
Large Ensemble ◽  
Model Framework ◽  
The North ◽  
Drought Prediction

Abstract. There is a strong scientific and social interest in understanding the factors leading to extreme events in order to improve the management of risks associated with hazards like droughts. In this study, artificial neural networks are applied to predict the occurrence of a drought in two contrasting European domains, Munich and Lisbon, with a lead time of 1 month. The approach takes into account a list of 28 atmospheric and soil variables as input parameters from a single-model initial-condition large ensemble (CRCM5-LE). The data were produced in the context of the ClimEx project by Ouranos, with the Canadian Regional Climate Model (CRCM5) driven by 50 members of the Canadian Earth System Model (CanESM2). Drought occurrence is defined using the standardized precipitation index. The best-performing machine learning algorithms manage to obtain a correct classification of drought or no drought for a lead time of 1 month for around 55 %–57 % of the events of each class for both domains. Explainable AI methods like SHapley Additive exPlanations (SHAP) are applied to understand the trained algorithms better. Variables like the North Atlantic Oscillation index and air pressure 1 month before the event prove essential for the prediction. The study shows that seasonality strongly influences the performance of drought prediction, especially for the Lisbon domain.

scholarly journals Applying machine learning for drought prediction using data from a large ensemble of climate simulations

2021 ◽  
Author(s):  
Elizaveta Felsche ◽  
Ralf Ludwig
Keyword(s):  
Machine Learning ◽  
Lead Time ◽  
Climate Model ◽  
Regional Climate ◽  
Standardized Precipitation Index ◽  
Machine Learning Algorithms ◽  
Large Ensemble ◽  
The North ◽  
Soil Variables ◽  
Drought Prediction

Abstract. There is strong scientific and social interest to understand the factors leading to extreme events in order to improve the management of risks associated with hazards like droughts. In this study, artificial neural networks are applied to predict the occurrence of a drought in two contrasting European domains, Munich and Lisbon, with a lead time of one month. The approach takes into account a list of 30 atmospheric and soil variables as input parameters from a single-model initial condition large ensemble (CRCM5-LE). The data was produced the context of the ClimEx project by Ouranos with the Canadian Regional Climate Model (CRCM5) driven by 50 members of the Canadian Earth System Model (CanESM2). Drought occurrence was defined using the Standardized Precipitation Index. The best performing machine learning algorithms managed to obtain a correct classification of drought or no drought for a lead time of one month for around 55–60 % of the events of each class for both domains. Explainable AI methods like SHapley Additive exPlanations (SHAP) were applied to gain a better understanding of the trained algorithms. Variables like the North Atlantic Oscillation Index and air pressure one month before the event proved to be of high importance for the prediction. The study showed that seasonality has a high influence on goodness of drought prediction, especially for the Lisbon domain.

Using a nested single-model large ensemble to assess the internal variability of the North Atlantic Oscillation and its climatic implications for Central Europe

2020 ◽  
Author(s):  
Andrea Böhnisch ◽  
Ralf Ludwig ◽  
Martin Leduc
Keyword(s):  
North Atlantic ◽  
Climate Model ◽  
Regional Climate ◽  
Internal Variability ◽  
Single Model ◽  
Large Ensemble ◽  
Central European ◽  
The North ◽  
Temperature And Precipitation ◽  
The North Atlantic Oscillation

<p>The ClimEx-project ("Climate change and hydrological extreme events"; www.climex-project.org) provides a single-model initial-condition ensemble that is unprecedented in terms of size, resolution and domain coverage: 50 members of the Canadian Earth System Model version 2 (CanESM2 Large Ensemble, 2.8° spatial resolution) are downscaled using the Canadian Regional Climate Model version 5 (CRCM5 Large Ensemble, 0.11° spatial and up to hourly temporal resolution) over two domains, Europe and northeastern North America. The high-resolution climate information serves as input for hydrological simulations to investigate the impact of internal variability and climate change on hydrometeorological extremes.</p><p>This study evaluates the downscaling of a teleconnection which affects northern hemisphere climate variability, the North Atlantic Oscillation (NAO), within the nested single-model large ensemble of the ClimEx project. The overall goal of this study is to assess whether the range of NAO internal variability is represented consistently between the driving global climate model (GCM, i.e., the CanESM2) and the nested regional climate model (RCM, i.e., the CRCM5).</p><p>The NAO pressure dipole is quantified in the CanESM2-LE; responses of mean surface air temperature and total precipitation sum to changes in the NAO index are evaluated within a Central European domain in both the CanESM2-LE and the CRCM5-LE. NAO–response relationships are expressed via Pearson correlation coefficients and the change per unit index change for historical (1981–2010) and future (2070–2099) winters.</p><p>Results show that statistically robust NAO patterns are found in the CanESM2-LE under current forcing conditions, and reproductions of the NAO flow pattern present in the CanESM2-LE produce plausible temperature and precipitation responses in the high-resolution CRCM5-LE. The NAO–response relationship is more strongly evolved in the CRCM5-LE than in the CanESM2-LE, but the inter-member spread shows no significant differences: thus internal variability expressed as inter-member spread can be seen as being represented consistently between the GCM and RCM. NAO–response relationships weaken in the future period in both the CanESM2-LE and CRCM5-LE, suggesting that the NAO influence on Central European temperature and precipitation decreases.</p><p>The results stress the advantages of a single-model ensemble regarding the evaluation of internal variability. They also strengthen the validity of the nested ensemble for further impact modelling using RCM data only, since important large-scale teleconnections present in the driving GCM propagate properly to the fine scale dynamics in the RCM.</p>

Applying machine learning for drought prediction using a large ensemble of climate simulations

2021 ◽  
Author(s):  
Elizaveta Felsche ◽  
Ralf Ludwig
Keyword(s):  
Machine Learning ◽  
Regional Climate Model ◽  
North Atlantic Oscillation ◽  
North Atlantic ◽  
North Atlantic Oscillation Index ◽  
Climate Model ◽  
Regional Climate ◽  
The North ◽  
The North Atlantic ◽  
The North Atlantic Oscillation

<p>There is strong scientific and social interest to understand the factors leading to extreme events in order to improve the management of risks associated with hazards like droughts. Recent events like the summer 2018 drought in Germany already had severe und unexpected impacts, e.g. forest fires and crop failures; in order to increase preparedness robust prediction tools are  urgently required. In this study, machine learning methods are applied to predict the occurrence of a drought with lead times of one to three months. The approach takes into account a list of thirty atmospheric and soil variables<strong> </strong>as predictor input parameters from a single regional climate model initial condition large ensemble (CRCM5-LE). The data was produced the context of the ClimEx project by Ouranos with the Canadian Regional Climate Model (CRCM5) driven by 50 members of the Canadian Earth System Model (CanESM2) for the Bavarian and Quebec domains.</p><p>Drought occurrence was defined using the Standardized Precipitation Index. The training and test datasets were chosen from the current climatology (1955-2005) for the Munich and Lisbon subdomain within the CRCM5-LE. The best performing machine learning algorithms managed to obtain a correct classification of drought or no drought for a lead time of one month for around 60 % of the events of each class for the both domains. Explainable AI methods like feature importance and shapley values were applied to gain a better understanding of the trained algorithms. Physical variables like the North Atlantic Oscillation Index and air pressure one month before the event proved to be of high importance for the prediction. The study showed that better accuracies can be obtained for the Lisbon domain, due to the stronger influence of the North Atlantic Oscillation Index on Portugal’s climate.</p>

scholarly journals Temperature seasonality in the North American continental interior during the early Eocene climatic optimum

2018 ◽  
Author(s):  
Ethan G. Hyland ◽  
Katharine W. Huntington ◽  
Nathan D. Sheldon ◽  
Tammo Reichgelt
Keyword(s):  
Climate Model ◽  
Regional Climate ◽  
Early Eocene ◽  
Early Eocene Climatic Optimum ◽  
The North ◽  
Annual Range ◽  
Climatic Optimum ◽  
Temperature Records ◽  
Model Ensembles ◽  
Continental Interior

Abstract. Paleogene greenhouse climate equability has long been a paradox in paleoclimate research. However, recent developments in proxy and modeling methods have suggested that strong seasonality may be a feature of at least some greenhouse periods. Here we present the first multi-proxy record of seasonal temperatures during the Paleogene from paleofloras, paleosol geochemistry, and carbonate clumped isotope thermometry in the Green River Basin (Wyoming, USA). These combined temperature records allow for the reconstruction of past seasonality in the continental interior, which shows that temperatures were warmer in all seasons during the peak early Eocene climatic optimum and that the mean annual range of temperature was high, similar to the modern value (~ 26 °C). Proxy data and downscaled Eocene regional climate model results suggest amplified seasonality during greenhouse events. Increased seasonality reconstructed for the early Eocene is similar in scope to the higher seasonal range predicted by downscaled climate model ensembles for future high-CO2 emissions scenarios. Overall, these data and model comparisons have substantial implications for understanding greenhouse climates in general, and may be important for predicting future seasonal climate regimes and their impacts in continental regions.

scholarly journals The Community Firn Model (CFM) v1.0

2020 ◽  
Author(s):  
C. Max Stevens ◽  
Vincent Verjans ◽  
Jessica M.D. Lundin ◽  
Emma C. Kahle ◽  
Annika N. Horlings ◽  
...  
Keyword(s):  
Climate Model ◽  
Noble Gas ◽  
Regional Climate ◽  
Ice Core ◽  
Ice Cores ◽  
Elevation Change ◽  
Model Framework ◽  
Modular Model ◽  
Surface Elevation Change ◽  
Air Diffusion

Abstract. Models that simulate evolution of polar firn are important for several applications in glaciology, including converting ice-sheet elevation-change measurements to mass change and interpreting climate records in ice cores. We have developed the Community Firn Model (CFM), an open-source, modular model framework designed to simulate numerous physical processes in firn. The modules include firn densification, heat transport, meltwater percolation and refreezing, water-isotope diffusion, and firn-air diffusion. The CFM is designed so that new modules can be added with ease. In this paper, we first describe the CFM and its modules. We then demonstrate the CFM's usefulness in two model applications that utilize two of its novel aspects. The CFM currently has the ability to run any of 13 previously published firn-densification models, and in the first application we compare those models' results when they are forced with regional climate model outputs for Summit, Greenland. The results show that the models do not agree well (spread greater than 10 %) when predicting depth-integrated porosity, firn age, or trend in surface-elevation change trend. In the second application, we show that the CFM's coupled firn-air and firn-densification models can simulate noble-gas records from an ice core better than a firn-air model alone.

Cereal yield forecasting combining satellite drought-based indices, regional climate and weather data using machine learning approaches in Morocco

2021 ◽  
Author(s):  
El houssaine Bouras ◽  
Lionel Jarlan ◽  
Salah Er-Raki ◽  
Riad Balaghi ◽  
Abdelhakim Amazirh ◽  
...  
Keyword(s):  
Machine Learning ◽  
Regional Climate ◽  
Model Development ◽  
Machine Learning Algorithms ◽  
Weather Data ◽  
Drought Indices ◽  
Support Vector ◽  
Learning Approaches ◽  
Climate Data ◽  
Yield Forecasting

<p>Cereals are the main crop in Morocco. Its production exhibits a high inter-annual due to uncertain rainfall and recurrent drought periods. Considering the importance of this resource to the country's economy, it is thus important for decision makers to have reliable forecasts of the annual cereal production in order to pre-empt importation needs. In this study, we assessed the joint use of satellite-based drought indices, weather (precipitation and temperature) and climate data (pseudo-oscillation indices including NAO and the leading modes of sea surface temperature -SST- in the mid-latitude and in the tropical area) to predict cereal yields at the level of the agricultural province using machine learning algorithms (Support Vector Machine -SVM-, Random forest -FR- and eXtreme Gradient Boost -XGBoost-) in addition to Multiple Linear Regression (MLR). Also, we evaluate the models for different lead times along the growing season from January (about 5 months before harvest) to March (2 months before harvest). The results show the combination of data from the different sources outperformed the use of a single dataset; the highest accuracy being obtained when the three data sources were all considered in the model development. In addition, the results show that the models can accurately predict yields in January (5 months before harvesting) with an R² = 0.90 and RMSE about 3.4 Qt.ha<sup>-1</sup>.  When comparing the model’s performance, XGBoost represents the best one for predicting yields. Also, considering specific models for each province separately improves the statistical metrics by approximately 10-50% depending on the province with regards to one global model applied to all the provinces. The results of this study pointed out that machine learning is a promising tool for cereal yield forecasting. Also, the proposed methodology can be extended to different crops and different regions for crop yield forecasting.</p>

Evaluating the ability of NA-CORDEX to simulate the seasonal modes of precipitation variability across the Western United States: Does resolution matter?

2021 ◽  
Author(s):  
Jonathan Meyer ◽  
Shih-Yu (Simon) Wang ◽  
Robert Gillies ◽  
Jin-Ho Yoon
Keyword(s):  
Performance Metrics ◽  
Climate Model ◽  
Regional Climate ◽  
Storm Track ◽  
Model Performance ◽  
Precipitation Variability ◽  
Future Trend ◽  
Spring Precipitation ◽  
Precipitation Patterns ◽  
The North

<p>The western U.S. precipitation climatology simulated by the NA-CORDEX regional climate model ensembles are examined to evaluate the capability of the 0.44<sup>° </sup>and 0.22<sup>° </sup>resolution<sup></sup>ensembles to reproduce 1) the annual and semi-annual precipitation cycle of several hydrologically important western U.S. regions and 2) localized seasonality in the amount and timing of precipitation. Collectively, when compared against observation-based gridded precipitation, NA-CORDEX RCMs driven by ERA-Interim reanalysis at the higher resolution 0.22<sup>° </sup>domain resolution dramatically outperformed the 0.44<sup>°</sup> ensemble over the 1950-2005 historical periods. Furthermore, the ability to capture the annual and semi-annual modes of variability was starkly improved in the higher resolution 0.22° ensemble. The higher resolution members reproduced more consistent spatial patterns of variance featuring lower errors in magnitude—especially with respect to the winter-summer and spring-fall seasonality. A great deal of spread in model performance was found for the semi-annual cycles, although the higher-resolution ensemble exhibited a more coherent clustering of performance metrics. In general, model performance was a function of which RCM was used, while future trend scenarios seem to cluster around which GCM was downscaled.</p><p><br>Future projections of precipitation patterns from the 0.22° NA-CORDEX RCMs driven by the RCP4.5 “stabilization scenario” and the RCP8.5 “high emission” scenario were analyzed to examine trends to the “end of century” (i.e. 2050-2099) precipitation patterns. Except for the Desert Southwest’s spring season, the RCP4.5 and RCP8.5 scenarios show a consensus change towards an increase in winter and spring precipitation throughout all regions of interest with the RCP8.5 scenario containing a greater number of ensemble members simulating greater wetting trends. The future winter-summer mode of variability exhibited a general consensus towards increasing variability with greatest change found over the region’s terrain suggesting a greater year-to-year variability of the region’s orographic response to the strength and location of the mid-latitude jet streams and storm track. Increasing spring-fall precipitation variability suggests an expanding influence of tropical moisture advection associated with the North American Monsoon, although we note that like many future monsoon projections, a spring “convective barrier” was also apparent in the NA-CORDEX ensembles.</p>

Drought dynamics and variability over Bundelkhand region of central India: Past, Present and Future

2021 ◽  
Author(s):  
Md Saquib Saharwardi ◽  
Aditya Kumar Dubey ◽  
Pankaj Kumar ◽  
Dmitry V. Sein
Keyword(s):  
Climate Model ◽  
Monsoon Season ◽  
Science And Technology ◽  
Regional Climate ◽  
Standardized Precipitation Index ◽  
Central India ◽  
Severe Drought ◽  
Future Projection ◽  
Russian Science ◽  
Regional Drought

<p>In the present study, an evaluation of the past, present, and future variability of droughts in the Bundelkhand region of Central India are analyzed. Bundelkhand is a severe drought-prone region with intense water stress, where in the last five years four were drought. Therefore, understanding the drivers of drought over the region and its future projection is quite crucial for regional water management. The assessment has been made by analyzing the observational dataset from 1951-2018 to understand the regional drought dynamics. The future projection is made using a multi-model ensemble from a regional climate model over the CORDEX South-Asia domain under the highest emission scenario. The Standardized Precipitation Index (SPI) and the Standardized Precipitation Evapotranspiration Index (SPEI) indices are used to understand present drought and its future projection. In addition to this, drought driving parameters like precipitation, temperature, sea-surface temperature wind circulation has been assessed to understand the regional drought dynamics. The composite analysis of drought indicates that the moisture-laden low-level jet from the Arabian Sea branch generally weakened compared to Bay of Bengal branch for monsoon season. Teleconnections of drought over Bundelkhand region shows that nearly half of the droughts are linked to El-Nino events that have become stronger in recent past. The model result reveals that regional climate variability is reasonably captured over the region. In addition, we found increasing drought frequency since the beginning of the 21<sup>st</sup> century. The detailed results from the analysis will be shown briefly in the general assembly.</p><p><strong>Acknowledgement: </strong>This work is jointly supported by the Department of Science and Technology (DST), Govt. of India, grant number DST/INT/RUS/RSF/P-33/G and the Russian Science Foundation (Project No.: 19-47-02015). The first author is also thankful to the Department of Science and Technology (DST), Govt. of India for providing DST INSPIRE fellowship (Grant No. IF160281).</p>

scholarly journals Accelerating changes in ice mass within Greenland, and the ice sheet’s sensitivity to atmospheric forcing

2019 ◽  
Vol 116 (6) ◽  
pp. 1934-1939 ◽  
Author(s):  
Michael Bevis ◽  
Christopher Harig ◽  
Shfaqat A. Khan ◽  
Abel Brown ◽  
Frederik J. Simons ◽  
...  
Keyword(s):  
Global Positioning System ◽  
Climate Model ◽  
Regional Climate ◽  
Atmospheric Forcing ◽  
Surface Mass ◽  
The North ◽  
Global Positioning ◽  
The North Atlantic Oscillation ◽  
Gravity Recovery ◽  
Gps Observations

From early 2003 to mid-2013, the total mass of ice in Greenland declined at a progressively increasing rate. In mid-2013, an abrupt reversal occurred, and very little net ice loss occurred in the next 12–18 months. Gravity Recovery and Climate Experiment (GRACE) and global positioning system (GPS) observations reveal that the spatial patterns of the sustained acceleration and the abrupt deceleration in mass loss are similar. The strongest accelerations tracked the phase of the North Atlantic Oscillation (NAO). The negative phase of the NAO enhances summertime warming and insolation while reducing snowfall, especially in west Greenland, driving surface mass balance (SMB) more negative, as illustrated using the regional climate model MAR. The spatial pattern of accelerating mass changes reflects the geography of NAO-driven shifts in atmospheric forcing and the ice sheet’s sensitivity to that forcing. We infer that southwest Greenland will become a major future contributor to sea level rise.

Sign in / Sign up

Export Citation Format

Share Document