An Inter-Comparison of Arctic Synoptic Scale Storms between Four Global Reanalysis Datasets

Quantifying the impacts of human interventions on relative mean sea level change in the Pearl River Delta, China

ABSTRACTThis study compares observations of 22 climate variables [rainfall, minimum and maximum temperature, mean sea level pressure (MSLP), and air temperature, geopotential heights, relative and specific humidity, and u‐wind and v‐winds at 500, 700 and 850 hPa] over Australia with NCEP‐NCAR (National Centers for Environmental Prediction‐National Center for Atmospheric Research) and ERA‐Interim, two most often used reanalysis products. The results indicate that both NCEP‐NCAR and ERA‐Interim generally reproduce the observed spatial patterns of long‐term mean annual rainfall, daily maximum/minimum temperature and MSLP, the monthly distribution (annual cycle) of rainfall and temperature, although temperature is generally better simulated than rainfall and ERA‐Interim shows an overall better performance than NCEP‐NCAR in term of continental scale. In term of linear trends, both NCEP‐NCAR and ERA‐Interim simulate observed trend signs in some regions, but not in others, and the spatial distributions of temperature trends are generally not as well simulated as that of annual rainfall for both NCEP‐NCAR and ERA‐Interim. In addition, NCEP‐NCAR and ERA‐Interim show similar spatial patterns of annual mean air temperature, geopotential heights, humidity and winds. This implies that studies using NCEP‐NCAR data could also make use of ERA‐Interim to explore the range of uncertainties for the regions and climate variables where NCEP‐NCAR and ERA‐Interim show differences. This conclusion may also apply to regions other than Australia. For example, the different trend signs of MSLP between NCEP‐NCAR and ERA‐Interim could present a scientific challenging for climatic change studies using MSLP trends, and the split between large‐scale and convective rainfall would also have merits globally to understand if the differences between the two reanalysis data sets are global or if we see convergence on key areas, such as tropices and frontal systems.

The authors examine five recent reanalysis products [NCEP Climate Forecast System Reanalysis (CFSR), Modern-Era Retrospective Analysis for Research and Applications (MERRA), Japanese 25-year Reanalysis Project (JRA-25), Interim ECMWF Re-Analysis (ERA-Interim), and Arctic System Reanalysis (ASR)] for 1) trends in near-surface radiation fluxes, air temperature, and humidity, which are important indicators of changes within the Arctic Ocean and also influence sea ice and ocean conditions, and 2) fidelity of these atmospheric fields and effects for an extreme event: namely, the 2007 ice retreat. An analysis of trends over the Arctic for the past decade (2000–09) shows that reanalysis solutions have large spreads, particularly for downwelling shortwave radiation. In many cases, the differences in significant trends between the five reanalysis products are comparable to the estimated trend within a particular product. These discrepancies make it difficult to establish a consensus on likely changes occurring in the Arctic solely based on results from reanalyses fields. Regarding the 2007 ice retreat event, comparisons with remotely sensed estimates of downwelling radiation observations against these reanalysis products present an ambiguity. Remotely sensed observations from a study cited herewith suggest a large increase in downwelling summertime shortwave radiation and decrease in downwelling summertime longwave radiation from 2006 and 2007. On the contrary, the reanalysis products show only small gains in summertime shortwave radiation, if any; however, all the products show increases in downwelling longwave radiation. Thus, agreement within reanalysis fields needs to be further checked against observations to assess possible biases common to all products.

Rainfall forecasting, especially extreme rainfall forecasting, is one of crucial tasks in weather forecasting since it has direct impact on accompanying devastating events such as flash floods and fast-moving landslides. However, obtaining rainfall forecasts with high accuracy, especially for extreme rainfall occurrences, is a challenging task. This study focuses on developing a forecasting model which is capable of forecasting rainfall, including extreme rainfall values. The rainfall forecasting was achieved through sequence learning capability of the Long Short-Term Memory (LSTM) method. The identification of the optimal set of features for the LSTM model was conducted using Random Forest and Granger Causality tests. Then, that best set of features was fed into Stacked LSTM, Bidirectional LSTM, and Encoder-Decoder LSTM models to obtain three days-ahead forecasts of rainfall with the input of the past fourteen days-values of selected features. Out of the three models, the best model was taken through post hoc residual analysis and extra validation approaches. This entire approach was illustrated utilizing rainfall and weather-related measurements obtained from the gauging station located in the city of Ratnapura, Sri Lanka. Originally, twenty-three features were collected including relative humidity, ssunshine hours, and mean sea level pressure. The performances of the three models were compared using RMSE. The Bidirectional LSTM model outperformed the other methods (RMSE < 5 mm and MAE < 3 mm) and this model has the capability to forecast extreme rainfall values with high accuracy.

The quality controlled (RAW) and homogenized (ADJ) radiosonde temperatures within 850–30 hPa collected at 118 stations in China are compared, on a monthly mean basis, with the temperatures extracted from 8 reanalysis datasets (REA) including NCEP-1, NCEP-2, ERA-40 (ECMWF 45-yr Reanalysis), ERA-Interim, JRA-55 (Japanese 55-yr Reanalysis), 20CR (20th Century Reanalysis), MERRA (Modern Era Retrospective-Analysis), and CFSR (Climate Forecast System Reanalysis). Average differences, correlations, standard deviations, and linear trends among RAW, ADJ, and REA for the period 1981–2010 are analyzed. The results reveal significant inhomogeneity in the time series of RAW radiosonde temperature in China; an overall negative adjustment was thus employed to obtain the ADJ temperatures, and the effect of the negative adjustment is the most significant within 200–100 hPa. Such a homogenization process has removed the system errors in RAW, possibly caused by radiosonde instrument changes and observation system upgrades. Hence, the correlation is higher between ADJ and REA than that between RAW and REA. The mean difference between ADJ and REA is about 1℃ during 1981–2010, while REA are mostly cooler in the troposphere and warmer in the stratosphere than ADJ; nonetheless, they have a significant high and positive correlation and their annual variability is notably consistent. Furthermore, the linear trends in REA and ADJ both demonstrate warming in the lower–mid troposphere and cooling in the mid stratosphere, with large uncertainties found in the upper troposphere and lower stratosphere. In general, ERA-Interim, JRA-55, and MERRA are more consistent with ADJ than other reanalysis datasets.

Inter-comparison of extra-tropical cyclone activity in nine reanalysis datasets

The wind resource assessment is a decisive process for the development of any wind power generation project. The wind data obtained from any single means may have errors which may lead to the wrong estimation of wind speed and hence the power density. Therefore, the data obtained from any source or method has to be verified and validated with another data source in order to make the resource estimation least erroneous. In the present work, the wind data measured by means of LiDAR and wind mast has been compared with the multiple reanalysis datasets (ERA5, EMD-ERA, MERRA2 and CFSR2) for offshore location in Gujarat, India. From the consequences, the conclusion has been drawn that the reanalysis data provides underestimated wind speeds as compared to LiDAR measurements. Further, among the studied reanalysis datasets, the ERA5 data exhibits highest correlation (0.9329) with the LiDAR measured data, which proves it to be most reliable data source among the other considered reanalysis datasets. The LiDAR measured the mean wind speed of 7.41 m/s at 100 m height, which implies to the mean wind power density of 328.24 W/m2. Thus, the study location is highly suitable for the establishment of an offshore wind power project.

A single-model 16-member ensemble is used to investigate how external model factors can affect model performance. Ensemble members are constructed with the land surface model (LSM) Joint UK Land Environment Simulator (JULES), with different choices of meteorological forcing [in situ, NCEP Climate Forecast System Reanalysis (CFSR)/CFSv2, or Water and Global Change (WATCH) Forcing Data ERA-Interim (WFDEI)] and ancillary datasets (in situ or remotely sensed), and with four time step modes. Effects of temporal averaging are investigated by comparing the hourly, daily, monthly, and seasonal ensemble performance against snow depth and water equivalent, soil temperature and moisture, and latent and sensible heat fluxes from one forest site and one clearing in the boreal ecozone of Finnish Lapland. Results show that meteorological data are the largest source of uncertainty; differences in ancillary data have little effect on model results. Although generally informative and representative, aggregated performance metrics fail to identify “right results for the wrong reasons”; to do so, scrutinizing of time series and of interactions between variables is necessary. Temporal averaging over longer intervals improves metrics—with the notable exception of bias, which increases—by reducing the effects of internal data and model variability on model response. Model evaluation during shoulder seasons (fall minus spring) identifies weaknesses in the reanalyses datasets that conventional seasonal performance (winter minus summer) neglects. In view of the importance of snow on the range of results obtained with the same model, let alone identical simulations using different temporal averaging, it is recommended that systematic evaluation, quantification of errors, and uncertainties in snow-covered regions be incorporated in future efforts to standardize evaluation methods of LSMs.

Observations are required to understand monsoon changes over time. Reanalyses are useful supplements to observations; however, their ability to reveal long-term monsoon changes remains unknown. Here, we evaluate against observations the performance of two reanalysis datasets covering the whole twentieth century, the ECMWF reanalysis of the twentieth century (ERA20C) and the Twentieth Century Reanalysis Project (20CR), in terms of reproducing the climatological averages, the centennial co-variation, and long-term trends of Northern Hemisphere land monsoon rainfall (NHLMR). Their performance is compared with three other widely used reanalyses, the NCEP–NCAR reanalysis (NCEP1), the ECMWF 40-year reanalysis (ERA40), and the Japanese 55-year Reanalysis Project (JRA55), for the period after 1955. Results show that the five reanalysis datasets reasonably reproduce the climatological NHLMR and NHLM domains. The leading co-variation mode of NHLMR, with an overall increasing tendency before 1955 and a decreasing tendency afterwards, is captured by the corresponding principal components of the two long-term datasets, with some spatial biases. For the long-term precipitation trends, both the ERA20C and 20CR reproduce the increasing trend during 1901–1955, except for parts of the North African (NAF) monsoon region. However, neither dataset captures the decreasing trend of NHLMR after 1955 because of limitations in the NAF and North American (NAM) monsoon region. The NCEP1, ERA40, and JRA55 datasets also have shortcomings in the NAM region. An overall decreasing NHLMR is only shown in the NCEP1 dataset. A moisture budget analysis reveals that the long-term trends of NHLMR are dominated by the dynamic component of moisture convergence, suggesting a prominent role for atmospheric circulation changes. The influence of residual terms related to observations assimilated in the reanalyses is also discussed.

Several statistical downscaling models have been developed in the past couple of decades to assess the hydrologic impacts of climate change by projecting the station-scale hydrological variables from large-scale atmospheric variables simulated by general circulation models (GCMs). This paper presents and compares different statistical downscaling models that use multiple linear regression (MLR), positive coefficient regression (PCR), stepwise regression (SR), and support vector machine (SVM) techniques for estimating monthly rainfall amounts in the state of Florida. Mean sea level pressure, air temperature, geopotential height, specific humidity, U wind, and V wind are used as the explanatory variables/predictors in the downscaling models. Data for these variables are obtained from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis dataset and the Canadian Centre for Climate Modelling and Analysis (CCCma) Coupled Global Climate Model, version 3 (CGCM3) GCM simulations. The principal component analysis (PCA) and fuzzy c-means clustering method (FCM) are used as part of downscaling model to reduce the dimensionality of the dataset and identify the clusters in the data, respectively. Evaluation of the performances of the models using different error and statistical measures indicates that the SVM-based model performed better than all the other models in reproducing most monthly rainfall statistics at 18 sites. Output from the third-generation CGCM3 GCM for the A1B scenario was used for future projections. For the projection period 2001–10, MLR was used to relate variables at the GCM and NCEP grid scales. Use of MLR in linking the predictor variables at the GCM and NCEP grid scales yielded better reproduction of monthly rainfall statistics at most of the stations (12 out of 18) compared to those by spatial interpolation technique used in earlier studies.

In this paper, we present a comprehensive review of the data sources and estimation methods of 30 currently available global precipitation data sets, including gauge‐based, satellite‐related, and reanalysis data sets. We analyzed the discrepancies between the data sets from daily to annual timescales and found large differences in both the magnitude and the variability of precipitation estimates. The magnitude of annual precipitation estimates over global land deviated by as much as 300 mm/yr among the products. Reanalysis data sets had a larger degree of variability than the other types of data sets. The degree of variability in precipitation estimates also varied by region. Large differences in annual and seasonal estimates were found in tropical oceans, complex mountain areas, northern Africa, and some high‐latitude regions. Overall, the variability associated with extreme precipitation estimates was slightly greater at lower latitudes than at higher latitudes. The reliability of precipitation data sets is mainly limited by the number and spatial coverage of surface stations, the satellite algorithms, and the data assimilation models. The inconsistencies described limit the capability of the products for climate monitoring, attribution, and model validation.

The climatological storm-track activity simulated by 17 Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4)/phase 3 of the Coupled Model Intercomparison Project (CMIP3) models is compared to that in the interim ECMWF Re-Analysis (ERA-Interim). Nearly half of the models show significant biases in storm-track amplitude: four models simulate storm tracks that are either significantly (>20%) too strong or too weak in both hemispheres, while four other models have interhemispheric storm-track ratios that are biased by over 10%. Consistent with previous studies, storm-track amplitude is found to be negatively correlated with grid spacing. The interhemispheric ratio of storm-track activity is highly correlated with the interhemispheric ratio of mean available potential energy, and this ratio is biased in some model simulations due to biases in the midlatitude temperature gradients. In terms of geographical pattern, the storm tracks in most CMIP3 models exhibit an equatorward bias in both hemispheres. For the seasonal cycle, most models can capture the equatorward migration and strengthening of the storm tracks during the cool season, but some models exhibit biases in the amplitude of the seasonal cycle. Possible implications of model biases in storm-track climatology have been investigated. For both hemispheres, models with weak storm tracks tend to have larger percentage changes in storm-track amplitudes over the seasonal cycle. Under global warming, for the NH, models with weak storm tracks tend to project larger percentage changes in storm-track amplitude whereas, for the SH, models with large equatorward biases in storm-track latitude tend to project larger poleward shifts. Preliminary results suggest that CMIP5 model projections also share these behaviors.

We compare and examine diurnal temperature tides including their migrating component (DW1) from the troposphere to the lower mesosphere, using data from Thermosphere‐Ionosphere‐Mesosphere‐Energetics and Dynamics/Sounding of the Atmosphere using Broadband Emission Radiometry (TIMED/SABER) and from six different reanalysis data sets: (1) the Modern Era Retrospective analysis for Research and Applications (MERRA), (2) the European Centre for Medium‐range Weather Forecasts (ECMWF) reanalysis (ERA‐Interim) (3) the National Centers for Environmental Prediction (NCEP) Climate Forecast System Reanalysis (CFSR), (4) the Japanese 25‐year reanalysis by Japanese Meteorological Agency (JMA) and the Central Research Institute of Electric Power Industry (CRIEPI) (JRA25), (5) the NCEP/National Center for Atmospheric Research reanalysis (NCEP1), and (6) the NCEP and Department of Energy (DOE) Atmospheric Model Intercomparison Project (AMIP‐II) reanalysis data (NCEP2). The horizontal and vertical structures of the diurnal tides in SABER and reanalyses reasonably agree, although the amplitudes are up to 30–50% smaller in the reanalyses than in the SABER in the upper stratosphere to lower mesosphere. Of all tidal components, the DW1 is dominant while a clear eastward propagating zonal wave number 3 component (DE3) is observed at midlatitudes of the Southern Hemisphere in winter. Among the six reanalyses, MERRA, ERA‐Interim and CFSR are better at reproducing realistic diurnal tides. It is found that the diurnal tides extracted from SABER data in the winter‐hemisphere stratosphere suffer from sampling issues that are caused by short‐term variations of the background temperature. In addition, the GSWM underestimates the amplitude in the midlatitude upper stratosphere by about 50%.

Diurnal variations in the troposphere are a source of the diurnal tides which are the prevalent dynamical phenomenon in the mesosphere and lower thermosphere. They are also discussed as contributing to the excitation of Rossby waves. Here we study diurnal variations of upper tropospheric and lower stratospheric winds (up to 22 km) over Japan from 1986 to 2008 mainly using data from the middle and upper atmosphere (MU) radar (34.85°N, 136.10°E) and JRA25/JCDAS data, as well as other four global reanalysis data sets (ERA40, ERA‐Interim, NCEP1, and NCEP2) and output data from Global Scale Wave Model (GSWM). The diurnal and semidiurnal components are extracted and analyzed. For the diurnal wind component, the MU radar data are used to validate the reanalysis data, and the reanalysis data are chosen for the analysis. The diurnal amplitude monotonically increases with height above 15–20 km. The diurnal phase shows an upward progression up to 15 to 20 km, while above 15 to 20 km, it shows a downward progression in most months. It is found that the diurnal tide, defined as the diurnal component with absolute zonal wave numbers of ≤6, is dominant in the upper troposphere (explaining 60 to 80% of the variance) and in the stratosphere (explaining 80 to 90% of the variance). It is also observed that medium‐scale waves contributed to the diurnal wind component in the upper troposphere from winter to spring (∼20% of the variance). For the semidiurnal wind component, only MU radar data are used for the analysis, which confirmed that the semidiurnal migrating tide is dominant through the troposphere and the lower stratosphere. The semidiurnal tidal amplitude shows a marked seasonal variation in the troposphere; the amplitude is largest in winter (∼0.4 m s−1) and smallest in summer (∼0.2 m s−1).

Salt marsh ecosystems are maintained by the dominant macrophytes that regulate the elevation of their habitat within a narrow portion of the intertidal zone by accumulating organic matter and trapping inorganic sediment. The long-term stability of these ecosystems is explained by interactions among sea level, land elevation, primary production, and sediment accretion that regulate the elevation of the sediment surface toward an equilibrium with mean sea level. We show here in a salt marsh that this equilibrium is adjusted upward by increased production of the salt marsh macrophyte Spartina alterniflora and downward by an increasing rate of relative sea-level rise (RSLR). Adjustments in marsh surface elevation are slow in comparison to interannual anomalies and long-period cycles of sea level, and this lag in sediment elevation results in significant variation in annual primary productivity. We describe a theoretical model that predicts that the system will be stable against changes in relative mean sea level when surface elevation is greater than what is optimal for primary production. When surface elevation is less than optimal, the system will be unstable. The model predicts that there is an optimal rate of RSLR at which the equilibrium elevation and depth of tidal flooding will be optimal for plant growth. However, the optimal rate of RSLR also represents an upper limit because at higher rates of RSLR the plant community cannot sustain an elevation that is within its range of tolerance. For estuaries with high sediment loading, such as those on the southeast coast of the United States, the limiting rate of RSLR was predicted to be at most 1.2 cm/yr, which is 3.5 times greater than the current, long-term rate of RSLR.