Downscaling of the flood discharge in a probabilistic framework

Many modeled and observed data are in coarse resolution, which are required to be downscaled. This study develops a probabilistic method to downscale 3-hourly runoff to hourly resolution. Hourly data recorded at the Poldokhtar Stream gauge (Karkheh River basin, Iran) during flood events (2009–2019) are divided into two groups including calibration and validation. Statistical tests including Chi-Square and Kolmogorov–Smirnov test indicate that the Burr distribution is proper distribution functions for rising and falling limbs of the floods’ hydrograph in calibration (2009–2013). A conditional ascending/descending random sampling from the constructed distributions on rising/falling limb is applied to produce hourly runoff. The hourly-downscaled runoff is rescaled based on observation to adjust mean three-hourly data. To evaluate the efficiency of the developed method, statistical measures including root mean square error, Nash–Sutcliffe efficiency, Kolmogorov-Smirnov, and correlation are used to assess the performance of the downscaling method not only in calibration but also in validation (2014–2019). Results show that the hourly downscaled runoff is in close agreement with observations in both calibration and validation periods. In addition, cumulative distribution functions of the downscaled runoff closely follow the observed ones in rising and falling limb in two periods. Although the performance of many statistical downscaling methods decreases in extreme values, the developed model performs well at different quantiles (less and more frequent values). This developed method that can properly downscale other hydroclimatological variables at any time and location is useful to provide high-resolution inputs to drive other models. Furthermore, high-resolution data are required for valid and reliable analysis, risk assessment, and management plans.     

New Failure Mode and Effects Analysis: An Evidential Downscaling Method

Failure mode and effects analysis (FMEA) is an engineering and management technique, which is widely used to define, identify, and eliminate known or potential failures, problems, errors, and risk from the design, process, service, and so on. In a typical FMEA, the risk evaluation is determined by using the risk priority number (RPN), which is obtained by multiplying the scores of the occurrence, severity, and detection. However, because of the uncertainty in FMEA, the traditional RPN has been criticized because of several shortcomings. In this paper, an evidential downscaling method for risk evaluation in FMEA is proposed. In FMEA model, we utilize evidential reasoning approach to express the assessment from different experts. Multi‐expert assessments are transformed to a crisp value with weighted average method. Then, Euclidean distance from multi‐scale is applied to construct the basic belief assignments in Dempster–Shafer evidence theory application. According to the proposed method, the number of ratings is decreased from 10 to 3, and the frame of discernment is decreased from 2¹⁰ to 2³, which greatly decreases the computational complexity. Dempster's combination rule is utilized to aggregate the assessment of risk factors. We illustrate a numerical example and use the proposed method to deal with the risk priority evaluation in FMEA. The results and comparison show that the proposed method is more flexible and reasonable for real applications. Copyright © 2014 John Wiley & Sons, Ltd.     

基于统计降尺度方法的夏季降水精细化预报

利用1999—2009年安徽省淮河以南地区60个县市站夏季逐日降水资料和安庆市探空站逐日资料,研究了中低层不同风向配置下局地降水与大尺度降水场之间的关系,以3种不同预报对象及相应的预报因子分别采用神经网络和线性回归方法设计6种预报模型对观测资料进行逼近和优化,从而实现空间降尺度.分析对比6种预报模型46站逐日降水量的拟合和预报效果,结果表明:采取相同的预报对象及预报因子的BP神经网络模型在拟合和预报效果上均好于线性回归模型,可见夏季降水场之间以非线性相关为主;神经网络模型预报结果同常用的Cressman插值预报相比,能很好地反映出降水的基本分布及局地特征;预报对象为单站降水序列的神经网络模型在以平原、河流为主要地形的区域预报效果较好,预报对象为REOF主成分的神经网络模型则在山地和丘陵地形区域预报效果较好.     

Brazil's current and future land balances: Is there residual land for bioenergy production?

This study presents a new database of land use categories in Brazil at a spatial resolution of 30 arc-second (about 1 km²). The spatial representation of current major land uses formally combines agricultural statistics from Brazil's latest census of the year 2006 at micro-region level and the Food and Agriculture Organization 2010 forest statistics with spatial land cover data sets. Spatial allocation (“downscaling”) algorithms were applied to obtain a spatial distribution of seven major land use categories. Remaining shares in each grid-cell were termed residual land, and were categorized according to legal protection status, biodiversity value, and whether they belong to the territory of the Amazon biome. We found a total of 84 Mha residual land of which 37 Mha occurred outside the territory of the Amazon biome and was neither legally protected nor categorized as highly biodiverse land. The 37 Mha “available residual land” equates to 4.4% of Brazil's geographical area and to 50% of its current cultivated land area. We assessed land quality using the Agro-ecological Zones modelling framework provides land suitability and productivity estimates of the available residual land. Nearly one-third of land emerged of prime quality and is therefore promising for biofuel feedstock production. Analysis of potential food-fuel competition suggests that until 2030 productivity improvements on current pastures could accommodate land demand for Brazil's increasing cattle herd and expanding croplands. If these productivity increases could be achieved on current agricultural land, residual land could provide areas for the sustainable expansion of biofuel feedstock production.     

An analytical comparison of four approaches to modelling the daily variability of solar irradiance using meteorological records

Temporal solar variability significantly affects the integration of solar power systems into the grid. It is thus essential to predict temporal solar variability, particularly given the increasing popularity of solar power generation globally. In this paper, the daily variability of solar irradiance at four sites across Australia is quantified using observed time series of global horizontal irradiance for 2003–2012. It is shown that the daily variability strongly depends on sky clearness with generally low values under a clear or overcast condition and high values under an intermittent cloudiness condition. Various statistical techniques are adopted to model the daily variability using meteorological variables selected from the ERA-Interim reanalysis as predictors. The nonlinear regression technique (i.e. random forest) is demonstrated to perform the best while the performance of the simple analog method is only slightly worse. Among the four sites, Alice Springs has the lowest daily variability index on average and Rockhampton has the highest daily variability index on average. The modelling results of the four sites produced by random forest have a correlation coefficient of above 0.7 and a median relative error around 40%. While the approach of statistical downscaling from a large spatial domain has been applied for other problems, it is shown in this study that it generally suffices to use only the predictors at a single near point for the problem of solar variability. The relative importance of the involved meteorological variables and the effects of clearness on the modelling of the daily variability are also explored.     

Investigation of the impact of global warming on precipitation pattern of Saudi Arabia

This study has been carried out to forecast the impact of global warming on the precipitation pattern of Saudi Arabia by the end of year 2100. Simulation has been done using EdGCM model software (with available 8×10 resolution) developed at Columbia University on which there have been produced global precipitation maps for the seasonal and annual averages for the last 5 years (2096–2100). For each map, EdGCM grid values surrounding Saudi Arabia are used as input to one of the tools of Eagle point software called surface modelling (SM). SM is a new approach for downscaling global climate model results. SM software modelled out isohyets at 0.2 mm/day interval. The results indicate that the present pattern of precipitation (more in winter and less in summer) is going to change by almost equal occurrence of precipitation in all seasons for double_CO₂ (2CO₂) experiment. The 2CO₂ experiment indicates an increase of about 16.05% over the annual average precipitation across the country.     

基于人工神经网络的统计降尺度模型研究

基于BP神经网络模型对黄河源区的降水、温度进行了统计降尺度研究,探讨了统计降尺度模式中考虑预报量的敏感大气环流因子随季节变化时对降水的降尺度效果的影响。结果表明,人工神经网络降尺度模型能成功地捕捉黄河源区的日平均温度及气温极值的年际变化趋势,纳什效率系数均达0.95以上;比较CON模型及PIE模型对降水指标的模拟能力,发现两种模型对1961~2000年不同降水指标时间序列的模拟能力相当;从季节尺度看,在冬季PIE模型显示了更好的模拟能力,但在夏秋季节PIE模型对多数降水指标的模拟能力略不及CON模型。总之,CON模型对降水指标的模拟效果更好。     

Can Artificial Climate Trends in Global Reanalysis be Reduced by Dynamical Downscaling:A Case Study of China

In this study, the ability of dynamical downscaling for reduction of artificial climate trends in global reanalysis is tested in China. Dynamical downscaling is performed using a 60-km horizontal resolution Regional Integrated Environmental Model System (RIEMS) forced by the NCEP-Department of Energy (DOE) reanalysis Ⅱ(NCEP-2). The results show that this regional climate model (RCM) can not only produce dynamically consistent fine scale fields of atmosphere and land surface in the regional domain, but it also has the ability to minimize artificial climate trends existing in the global reanalysis to a certain extent. As compared to the observed 2-meter temperature anomaly averaged across China, our model can simulate the observed inter-annual variation and variability as well as reduce artificial climate trends in the reanalysis by approximately 0.10℃ decade-1 from 1980to 2007. The RIEMS can effectively reduce artificial trends in global reanalysis for areas in western China,especially for regions with high altitude mountains and deserts, as well as introduce some new spurious changes in other local regions. The model simulations overestimated observed winter trends for most areas in eastern China with the exception of the Tibetan Plateau, and it greatly overestimated observed summer trends in the Sichuan Basin located in southwest China. This implies that the dynamical downscaling of RCM for long-term trends has certain seasonal and regional dependencies due to imperfect physical processes and parameterizations.     

Statistical downscaling of rainfall data using sparse variable selection methods

In many statistical downscaling methods, atmospheric variables are chosen by using a combination of expert knowledge with empirical measures such as correlations and partial correlations. In this short communication, we describe the use of a fast, sparse variable selection method, known as RaVE, for selecting atmospheric predictors, and illustrate its use on rainfall occurrence at stations in South Australia. We show that RaVE generates parsimonious models that are both sensible and interpretable, and whose results compare favourably to those obtained by a non-homogeneous hidden Markov model (Hughes et al., 1999).     

Impacts of small scale rainfall variability in urban areas: a case study with 1D and 1D/2D hydrological models in a multifractal framework

In this paper the sensitivity to small scale unmeasured rainfall variability (i.e. at scales smaller than 1 km by 1 km by 5 min in time, which are usually available with C-band radars) of a 1D/2D model with a 10 m resolution and a semi-distributed 1D model of the same 1.47 km² urban area is analyzed. The 1D/2D model is the open source numerical platform Multi-Hydro, which couples (open source) distributed models of involved hydrological/hydraulic processes. The methodology implemented to evaluate the uncertainties consists of generating an ensemble of realistic rainfall fields downscaled to a resolution of 12.3 m in space and 18.75 s in time with the help of a stochastic universal multifractal model. The corresponding ensemble of hydrographs is then simulated. It appears that the uncertainty is significant and that Multi-Hydro unveils much more uncertainty than the simpler 1D model. This points out a need to develop high resolution distributed modelling in urban areas.