Search

In this paper, we show a new clustering technique (k-gaps) aiming to generate a robust regionalization using sparse climate datasets with incomplete information in space and time. Hence, this method provides a new approach to cluster time series of different temporal lengths, using most of the information contained in heterogeneous sets of climate records that, otherwise, would be eliminated during data homogenization procedures. The robustness of the method has been validated with different synthetic datasets, demonstrating that k-gaps performs well with sample-starved datasets and missing climate information for at least 55% of the study period. We show that the algorithm is able to generate a climatically consistent regionalization based on temperature observations similar to those obtained with complete time series, outperforming other clustering methodologies developed to work with fragmentary information. k-Gaps clusters can therefore provide a useful framework for the study of long-term climate trends and the detection of past extreme events at regional scales. [ABSTRACT FROM AUTHOR]

The present paper reports a study, where growing hierarchical self-organising map (GHSOM) has been applied to achieve a visual cluster analysis to the Indian rainfall dataset consisting of 142 years of Indian rainfall data so that the yearly rainfall can be segregated into small groups to visualise the pattern of clustering behaviour of yearly rainfall due to changes in monthly rainfall for each year. Also, through support vector machine (SVM), it has been observed that generation of clusters impacts positively on the prediction of the Indian summer monsoon rainfall. Results have been presented through statistical and graphical analyses. [ABSTRACT FROM AUTHOR]

This paper presents a hybrid model to better predict Indian summer monsoon rainfall. The algorithm considers suitable techniques for processing dense datasets. The proposed three-step algorithm comprises closed itemset generation-based association rule mining for feature selection, cluster membership for dimensionality reduction, and simple logistic function for prediction. The application of predicting rainfall into flood, excess, normal, deficit, and drought based on 36 predictors consisting of land and ocean variables is presented. Results show good accuracy in the considered study period of 37years (1969-2005). [ABSTRACT FROM AUTHOR]

This paper aimed to investigate the potential of using spectral and textural features in the Meteosat Second Generation-Spinning Enhanced Visible and Infrared Imager (MSG-SEVIRI) data for developing techniques capable of classifying convective and stratiform rain areas in the satellite image. Two different classification methods were introduced that use the brightness temperature (BT) T and brightness temperature differences T- T T- T T- T T- T T- T T- T, and T- T as spectral parameters, along with textural parameters derived from the thermal infrared MSG-SEVIRI channel to discriminate between convective and stratiform rainy clouds. The first is an algorithm based on the probability of convective rainfall (PCR) for each pixel of the MSG satellite data and the second is an artificial neural network multilayer perceptron (MLP) model that relies on the correlation of satellite data with convective and stratiform rain. Both schemes were trained using as reference convective/stratiform classification data from 88 stations in Greece for 20 rainy days with high convective activity and evaluated against an independent dataset of gauge data for six rainy days. Two PCR and two MLP algorithms were constructed based on the previous results: the PCR1 and MLP1 algorithms that use only spectral measures and the PCR2 and MLP2 algorithms based on both spectral and textural measures. It was found that the introduction of textural parameters as additional information to a technique (PCR2 and MLP2 ) works in improving the delineation between convective and stratiform rainy pixels compared to the techniques based only on spectral information (PCR1 and MLP1). The comparison of the two schemes revealed a clear outperformance of the MLP techniques over the PCR techniques. Best skill is provided by MLP2 (POD = 74.5 %, FAR = 44.3 %, POFD = 22.5 %, CSI = 0.47, ETS = 0.31, bias = 1.34) followed by MLP1 and the two PCR techniques. All algorithms overestimate the convective rain occurrences observed by the rain gauge network. These findings showed that the combined use of spectral and textural features in the MSG-SEVIRI imagery can be beneficial for the classification of convective and stratiform precipitating clouds. [ABSTRACT FROM AUTHOR]

Typical urban and rural temperature records are essential for the estimation and comparison of urban heat island effects in different regions, and one of the key issues is how to identify the typical urban and rural stations. This study tried to analyze the similarity of air temperature sequences by using dynamic time warping algorithm (DTW) to improve the selection of typical stations. We examined the similarity of temperature sequences of 20 stations in Beijing, and the results indicated that DTW algorithm could identify the difference of temperature sequence, and clearly divide them into different groups according to their probability distributions. DTW algorithm could create more rigorous and reasonable classification for typical urban and rural stations in Beijing by comparing with remote sensing and cluster analysis method. We also found that some traditional rural stations did not represent temperature variation in rural areas because their surrounding environments have been highly modified by urbanization process in last decades, and this might lead to an underestimate of the urban climate effect by 1.24℃. The study suggests that DTW algorithm is a simple and efficient method for selecting appropriate temperature records and analyzing temperature sequences with a good potential in improving urban heat island estimation. [ABSTRACT FROM AUTHOR]

Delineation of homogeneous regions has found its way into many hydrological applications as it helps in addressing the challenges in understanding the behavior of rainfall distribution and its variability at a local scale. In the present study, rainfall data recoded by 83 tea gardens in the upper Brahmaputra valley region of Assam have been used to identify homogeneous rainfall regions by using fuzzy clustering analysis. Furthermore, seven different cluster validity indices (CVs) were utilized to find out the optimum clustering in the fuzzy c-means (FCM) algorithm. The clusters thus formed were assessed for statistical homogeneity by performing homogeneity tests based on L-moment. Three different combinations of feature vectors were employed in FCM algorithm and the outputs were compared for attaining best solutions to regionalization. The results were further compared with previous regionalization studies. The analysis and comparison conclude that if regionalization needs to be done at a local scale, further sub-clustering of a larger clustered region to smaller regions may be required. Local rainfall data can be used for the purpose provided a good dataset with large number of station points are available within the region. Along with rainfall data, geographical location parameters (latitude, longitude, and elevation) need to be taken into account for getting a definite conclusion. [ABSTRACT FROM AUTHOR]

High-precision areal rainfall is crucial for hydrometeorological coupled forecasts. The accuracy of quantitative precipitation estimates (QPE) is improved by merging radar-rain gauge data with an integration approach based on a statistical weight matrix in the Yishu River catchment, China. First, a local Z-R relationship (Z = 85R1.82) is reconstructed using a genetic optimization algorithm to minimize the error from different precipitation patterns and climate zones. Next, based on the local Z-R relationship, six methods of merging radar-rain gauge data are respectively adapted to improve the accuracy of QPE, as follows: mean field bias (MFB), Kalman filter (KLM), optimum interpolation (OPT), variation method (VAR), two-step calibration of KLM and OPT (KOP), and two-step calibration of KLM and VAR (KVR). The results indicate that QPE accuracy is clearly improved, and is in good agreement with rain gauge observations, after the six merging methods are applied. Among these methods, KOP performs the best, reducing the mean relative error from 55.2 to 15.1%. An innovative aspect of this work is the inclusion of an integrated ideology based on a statistical weight matrix, which further improves the accuracy of QPE by incorporating the advantages of each estimation mode. The results further show that the accuracy of QPE derived from the integration approach is higher than that obtained by any individual method; QPE values are similar to those obtained the automatic rain gauge network in both the spatial distribution and location of the intense precipitation centers, and better reflects the precipitation status over the ground surface. This approach could serve as a promising conventional method for QPE in the study region. [ABSTRACT FROM AUTHOR]

The new 1° × 1° resolution global Full Data Daily Analysis Version 2018 published by the Global Precipitation Climatology Centre (GPCC) of Deutscher Wetterdienst was compared with an analysis of the measurements from the national dataset over the mainland of China with regard to four of the 27 ETCCDI indices (http://etccdi.pacificclimate.org/list%5f27%5findices.shtml) commonly used to determine extreme precipitation (Rx5day, R10mm, CDD and SDII). After extreme value check, integrity check, and homogeneity check, the observations from 2327 surface stations covering the years from 1982 to 2016 fulfilled the criteria for the evaluation. The in situ daily precipitation data were interpolated onto a 1° × 1° grid over the mainland of China by employing Shepard's angular and distance weighting algorithm. The four aforementioned indices were then calculated on the national station–based analysis being referred to as STA. Moreover, the aforementioned gridded GPCC Full Data Daily product was directly utilized to calculate the same indices (FDDA). The China national means of Rx5day, R10mm, CDD and SDII calculated from FDDA and STA had similar variations and trends with high correlation coefficients, and the mean biases between FDDA and STA were 2.5 mm, 1.2 days, 0.0 day and 0.3 mm respectively. The trends of Rx5day, R10mm and SDII are increasing, whereas the trend of CDD is negative. The distributions of the grid mean and the grid trends of indices over China from FDDA and STA show similar patterns too, indicating that the FDDA shows a surprisingly high fidelity in reproducing almost the same patterns in the four ETCCDI indices chosen compared with the STA-based analysis. [ABSTRACT FROM AUTHOR]

We describe a new two-step modeling framework for investigating the impact of climate change on human comfort in outdoor urban environments. In the first step, climate change scenarios for air temperature and solar radiation (global, diffuse, direct components) are created using a change-factor algorithm. The change factors are calculated by comparing ranked daily regional climate model outputs for a future-period and a present-day period, and then changes consistent with these daily change factors are applied to historical hourly climate observations. In the second step, the mean-radiant-temperature ( T) is calculated using the SOLWEIG (SOlar and LongWave Environmental Irradiance Geometry) model. T which describes the radiant heat exchange between a person and their surroundings, is one of the most important meteorologically derived parameters governing human energy balance and outdoor thermal comfort, especially during warm and sunny days. We demonstrate that change factors can be applied independently to maximum air temperature and daily global solar radiation, and show that the outputs from the algorithm, when aggregated to daily values, are consistent with the driving regional climate model. Finally, we demonstrate how to obtain quantitative information from the scenarios regarding the potential impact of climate change on outdoor thermal comfort, by calculating changes in the distribution of hourly summer day-time T and changes in the number of hours with T >55 °C. [ABSTRACT FROM AUTHOR]

Tracer selection is the fundamental step in the retrieval of atmospheric motion vectors (AMVs). In this study, a new technique for tracer selection based on extracting the corner points in an infrared (IR) image of a geostationary satellite for the retrieval of AMVs is developed. Corner points are frequently used in computer vision to identify the important features of an image. These points are usually characterized by high gradient values of the image intensity in all directions and lie at the junctions of different brightness regions in the image. Corner points find application in computer vision for motion tracking, stereo vision, mosaics, etc., but this is the first time that the information from corners is used for tracer selection in AMV retrieval. In the present study, a commonly used Harris corner (HC) detection algorithm is followed to extract corners from the image intensity of an IR image. The tracers selected using the HC method are then passed on to the other steps of the retrieval algorithm, viz., tracking, height assignment, and quality control procedures for the retrieval of AMVs. For the initial development of the HC, Meteosat-7 IR images are used to derive AMVs for July and December 2010. The AMVs retrieved using HC are validated against collocated radiosonde observations, and the results are compared with the local anomaly (LA) method as reference. LA is used for tracer selection in operational AMV retrieval algorithm from the Indian geostationary satellite Kalpana-1. AMVs retrieved using HC have shown considerable improvement in the AMV accuracy over the AMVs derived using LA. [ABSTRACT FROM AUTHOR]

Aroma types of flue-cured tobacco (FCT) are classified into light, medium, and heavy in China. However, the spatial distribution of FCT aroma types and the relationships among aroma types, chemical parameters, and climatic variables were still unknown at national scale. In the current study, multi-year averaged chemical parameters (total sugars, reducing sugars, nicotine, total nitrogen, chloride, and KO) of FCT samples with grade of C3F and climatic variables (mean, minimum and maximum temperatures, rainfall, relative humidity, and sunshine hours) during the growth periods were collected from main planting areas across China. Significant relationships were found between chemical parameters and climatic variables ( p < 0.05). A spatial distribution map of FCT aroma types were produced using support vector machine algorithms and chemical parameters. Significant differences in chemical parameters and climatic variables were observed among the three aroma types based on one-way analysis of variance ( p < 0.05). Areas with light aroma type had significantly lower values of mean, maximum, and minimum temperatures than regions with medium and heavy aroma types ( p < 0.05). Areas with heavy aroma type had significantly lower values of rainfall and relative humidity and higher values of sunshine hours than regions with light and medium aroma types ( p < 0.05). The output produced by classification and regression trees showed that sunshine hours, rainfall, and maximum temperature were the most important factors affecting FCT aroma types at national scale. [ABSTRACT FROM AUTHOR]

The ensemble method has long been used to reduce the errors that are caused by initial conditions and/or parameterizations of models in forecasting problems. In this study, neural network (NN) simulations are applied to ensemble weather forecasting. Temperature forecasts averaged over 2 weeks from four different forecasts are used to develop the NN model. Additionally, an ensemble mean of bias-corrected data is used as the control experiment. Overall, ensemble forecasts weighted by NN with feed forward backpropagation algorithm gave better root mean square error, mean absolute error, and same sign percent skills compared to those of the control experiment in most stations and produced more accurate weather forecasts. [ABSTRACT FROM AUTHOR]

In the framework of the COST733 Action 'Harmonisation and Applications of Weather Types Classifications for European Regions' a new circulation type classification software (hereafter, referred to as cost733class software) is developed. The cost733class software contains a variety of (European) classification methods and is flexible towards choice of domain of interest, input variables, time step, number of circulation types, sequencing and (weighted) target variables. This work introduces the capabilities of the cost733class software in which the resulting circulation types (CTs) from various circulation type classifications (CTCs) are applied on observed summer surface ozone concentrations in Central Europe. Firstly, the main characteristics of the CTCs in terms of circulation pattern frequencies are addressed using the baseline COST733 catalogue (cat 2.0), at present the latest product of the new cost733class software. In a second step, the probabilistic Brier skill score is used to quantify the explanatory power of all classifications in terms of the maximum 8 hourly mean ozone concentrations exceeding the 120-μg/m threshold; this was based on ozone concentrations from 130 Central European measurement stations. Averaged evaluation results over all stations indicate generally higher performance of CTCs with a higher number of types. Within the subset of methodologies with a similar number of types, the results suggest that the use of CTCs based on optimisation algorithms are performing slightly better than those which are based on other algorithms (predefined thresholds, principal component analysis and leader algorithms). The results are further elaborated by exploring additional capabilities of the cost733class software. Sensitivity experiments are performed using different domain sizes, input variables, seasonally based classifications and multiple-day sequencing. As an illustration, CTCs which are also conditioned towards temperature with various weights are derived and tested similarly. All results exploit a physical interpretation by adapting the environment-to-circulation approach, providing more detailed information on specific synoptic conditions prevailing on days with high surface ozone concentrations. This research does not intend to bring forward a favourite classification methodology or construct a statistical ozone forecasting tool but should be seen as an introduction to the possibilities of the cost733class software. It this respect, the results presented here can provide a basic user support for the cost733class software and the development of a more user- or application-specific CTC approach. [ABSTRACT FROM AUTHOR]

This study reveals homogeneous sub-regions over the poorly studied area of western equatorial Africa (10S-7N and 7E-30E). Monthly totals of 141 stations covering the period 1955-1984 are used. The stations are grouped based on the similarity of their interannual rainfall variability. In addition to annual totals, four different seasons are examined separately for regionalization, an approach that has lacked in previous studies. The four 3-month seasons are defined as follows: January-February-March (JFM), April-May-June (AMJ), July-August-September (JAS), and October-November-December (OND). Two different algorithms are applied and compared: the rotated principal component analysis (RPCA) in conjunction with Ward's method, and the RPCA in conjunction with k-means method. The principal components that explain about 65% of total variance are retained and then varimax rotated. The corresponding scores are utilized as input for cluster analysis. Using Ward's method, five sub-regions are recognized for AMJ, JAS and OND and 4 sub-regions for JFM and annual data. The regions are geographically well distributed over the area and consist of roughly the same number of stations. The F-test is used to evaluate the homogeneity of each sub-region. The results show that all sub-regions are strongly homogeneous. Assuming the same number of clusters, the k-means method provides comparable spatial patterns with those of Ward's method. However, there are some differences, which are more evident in JAS and OND. Like Ward's method, the values of F-ratio for the k-means algorithm also confirm the homogeneity of all seasons/sub-regions. The interannual variability of rainfall for each season/sub-region is also provided and compared. [ABSTRACT FROM AUTHOR]

Climate change scenarios generated by general circulation models have too coarse a spatial resolution to be useful in planning disaster risk reduction and climate change adaptation strategies at regional to river basin scales. This study presents a new non-parametric statistical K-nearest neighbor algorithm for downscaling climate change scenarios for the Rohini River Basin in Nepal. The study is an introduction to the methodology and discusses its strengths and limitations within the context of hindcasting basin precipitation for the period of 1976-2006. The actual downscaled climate change projections are not presented here. In general, we find that this method is quite robust and well suited to the data-poor situations common in developing countries. The method is able to replicate historical rainfall values in most months, except for January, September, and October. As with any downscaling technique, whether numerical or statistical, data limitations significantly constrain model ability. The method was able to confirm that the dataset available for the Rohini Basin does not capture long-term climatology. Yet, we do find that the hindcasts generated with this methodology do have enough skill to warrant pursuit of downscaling climate change scenarios for this particularly poor and vulnerable region of the world. [ABSTRACT FROM AUTHOR]

Existing simple but theoretically based clear-sky models for longwave down-welling radiation (LDR) and cloud impact algorithms transforming them to all-sky radiation (LDR) are checked against locally calibrated empirical algorithms. They are evaluated for daylight hours based on measurements in regionally differing climates of Germany. The Prata clear-sky scheme is additionally tested with adjusted coefficients so that LDR converges against a realistic emissivity for a completely dry atmosphere. This version is characterised by an improved modelled variance. Compared with locally calibrated schemes, bias and root mean square error (RMSE) of the more theoretical clear-sky schemes do not differ significantly and yield even better results at a mountain site. In contrast, the locally calibrated algorithms yield biases up to 9% and an increase in RMSE between 6% and 67%, if applied for other sites. For daylight hours, the cloud impact on LDR can be calculated via the ratio of observed to clear-sky global irradiation (CMF). With CMF, the Crawford and Duchon scheme reveals the lowest bias and a decrease in RMSE by 22% against the next best performing algorithms. Compared with synoptic cloud observations as input, the bias is reduced by 9 to 28 W m and the scattering of the residuals decreases by 20% to 30%. Based on published results for also non-European sites, it is inferred that the more theoretically based LDR schemes and cloud impact evaluated via CMF are universally applicable and perform at least in the order of magnitude of locally calibrated empirical algorithms. [ABSTRACT FROM AUTHOR]

Extreme normalised residuals, defined as departures from the average values, of 65 daily maximum, Tmax, and minimum, Tmin, temperature series recorded in Catalonia (NE Spain) during 1950–2004 are analysed. Similarly to the sampling strategies applied to long dry spells, the partial duration series (PDS) offer some advantages in comparison with the annual extreme series. Instead of using a common percentile threshold for all temperature series, PDS are chosen according to the mean excess plot procedure. Series of extreme residuals are modelled, in terms of the L-moments formulation, by the generalised Pareto distribution. Extreme residuals of Tmax and Tmin are estimated for return periods ranging from 2 to 50 years and their spatial distribution is represented for selected return periods of 2, 5, 10, 25 and 50 years. Two daily extreme temperatures events, a hot episode (in August) and a cold episode (in February), are simulated taking into account the average Tmax ( Tmin) for a day in August (February), their standard deviations and the extremes for a 50-year return period. Both simulations are compared with outstanding real episodes recorded on August 13th 2003 and February 11th 1956. Additionally, a spatial regionalisation of Catalonia in several clusters, in terms of the extreme residuals for return periods from 2 to 50 years, is done. A principal component analysis is applied to the extreme residual curves characterising every temperature series and, using as variables the principal components, the regionalisation is obtained by applying the average linkage clustering algorithm. Finally, each cluster is characterised by its average extreme residual curve for return periods ranging from 2 to 50 years at 1-year interval. [ABSTRACT FROM AUTHOR]

Reflectances measured by the ‘visible’ channel of Meteosat are converted to global radiation at the ground by the Heliosat-algorithm. This algorithm is based on the inverse relationship between reflectance and transmittance. This relationship is, however, complicated by two factors: 1) the absorptance of the atmosphere is varying, and 2) the reflectivity is different in different directions. These two factors are again depending on the state of the atmosphere (mainly clouds) and the sun-ground-satellite geometry. The performance of the Heliosat-method is here tested against ground data from Bergen (Norway), Geneva (Switzerland) and Lyon (France), and analysed in light of cloud properties as observed from ground, and different sun-ground-satellite geometries. Situations where modelled and measured radiation differ are identified, and possible causes are suggested. Some of the suggestions are supported by calculations with a radiative transfer model. [ABSTRACT FROM AUTHOR]