Evaluation of Statistical Downscaling Methods for Simulating Daily Precipitation Distribution, Frequency, and Temporal Sequence

2021 ◽  
Vol 64 (3) ◽  
pp. 771-784
Author(s):  
Xunchang Zhang ◽  
Mingxi Shen ◽  
Jie Chen ◽  
Joel W. Homan ◽  
Phillip R. Busteed
Keyword(s):  
Climate Change ◽  
Statistical Downscaling ◽  
Climate Changes ◽  
Daily Precipitation ◽  
Weather Generator ◽  
Temporal Sequence ◽  
Precipitation Distribution ◽  
Climate Conditions ◽  
Weather Stations ◽  
Distribution Frequency

HighlightsNine statistical downscaling methods from three downscaling categories were evaluated.Weather generator-based methods had advantages in simulating non-stationary precipitation.Differences in downscaling performance were smaller within each category than between categories.The performance of each downscaling method varied with climate conditions.Abstract. Spatial discrepancy between global climate model (GCM) projections and the climate data input required by hydrological models is a major limitation for assessing the impact of climate change on soil erosion and crop production at local scales. Statistical downscaling techniques are widely used to correct biases of GCM projections. The objective of this study was to evaluate the ability of nine statistical downscaling methods from three available statistical downscaling categories to simulate daily precipitation distribution, frequency, and temporal sequence at four Oklahoma weather stations representing arid to humid climate regions. The three downscaling categories included perfect prognosis (PP), model output statistics (MOS), and stochastic weather generator (SWG). To minimize the effect of GCM projection error on downscaling quality, the National Centers for Environmental Prediction (NCEP) Reanalysis 1 data at a 2.5° grid spacing (treated as observed grid data) were downscaled to the four weather stations (representing arid, semi-arid, sub humid, and humid regions) using the nine downscaling methods. The station observations were divided into calibration and validation periods in a way that maximized the differences in annual precipitation means between the two periods for assessing the ability of each method in downscaling non-stationary climate changes. All methods were ranked with three metrics (Euclidean distance, sum of absolute relative error, and absolute error) for their ability in simulating precipitation amounts at daily, monthly, yearly, and annual maximum scales. After eliminating the poorest two performers in simulating precipitation mean, distribution, frequency, and temporal sequence, the top four remaining methods in ascending order were Distribution-based Bias Correction (DBC), Generator for Point Climate Change (GPCC), SYNthetic weather generaTOR (SYNTOR), and LOCal Intensity scaling (LOCI). DBC and LOCI are bias-correction methods, and GPCC and SYNTOR are generator-based methods. The differences in performances among the downscaling methods were smaller within each downscaling category than between the categories. The performance of each method varied with the climate conditions of each station. Overall results indicated that the SWG methods had certain advantages in simulating daily precipitation distribution, frequency, and temporal sequence for non-stationary climate changes. Keywords: Climate change, Climate downscaling, Downscaling method evaluation, Statistical downscaling.

scholarly journals Configuration and intercomparison of deep learning neural models for statistical downscaling

2020 ◽  
Vol 13 (4) ◽  
pp. 2109-2124 ◽  
Author(s):  
Jorge Baño-Medina ◽  
Rodrigo Manzanas ◽  
José Manuel Gutiérrez
Keyword(s):  
Climate Change ◽  
Deep Learning ◽  
Statistical Downscaling ◽  
Linear Models ◽  
Regional Climate ◽  
Added Value ◽  
Climate Conditions ◽  
Continental Scale ◽  
Spatial Features ◽  
Learning Techniques

Abstract. Deep learning techniques (in particular convolutional neural networks, CNNs) have recently emerged as a promising approach for statistical downscaling due to their ability to learn spatial features from huge spatiotemporal datasets. However, existing studies are based on complex models, applied to particular case studies and using simple validation frameworks, which makes a proper assessment of the (possible) added value offered by these techniques difficult. As a result, these models are usually seen as black boxes, generating distrust among the climate community, particularly in climate change applications. In this paper we undertake a comprehensive assessment of deep learning techniques for continental-scale statistical downscaling, building on the VALUE validation framework. In particular, different CNN models of increasing complexity are applied to downscale temperature and precipitation over Europe, comparing them with a few standard benchmark methods from VALUE (linear and generalized linear models) which have been traditionally used for this purpose. Besides analyzing the adequacy of different components and topologies, we also focus on their extrapolation capability, a critical point for their potential application in climate change studies. To do this, we use a warm test period as a surrogate for possible future climate conditions. Our results show that, while the added value of CNNs is mostly limited to the reproduction of extremes for temperature, these techniques do outperform the classic ones in the case of precipitation for most aspects considered. This overall good performance, together with the fact that they can be suitably applied to large regions (e.g., continents) without worrying about the spatial features being considered as predictors, can foster the use of statistical approaches in international initiatives such as Coordinated Regional Climate Downscaling Experiment (CORDEX).

Downscaling and Projection of Winter Extreme Daily Precipitation over North America

2008 ◽  
Vol 21 (5) ◽  
pp. 923-937 ◽  
Author(s):  
Jiafeng Wang ◽  
Xuebin Zhang
Keyword(s):  
Climate Change ◽  
United States ◽  
North America ◽  
Statistical Downscaling ◽  
Daily Precipitation ◽  
Generalized Extreme Value Distribution ◽  
Specific Humidity ◽  
Climate Modelling ◽  
Central United States ◽  
Distribution Parameters

Abstract Large-scale atmospheric variables have been statistically downscaled to derive winter (December–March) maximum daily precipitation at stations over North America using the generalized extreme value distribution (GEV). Here, the leading principal components of the sea level pressure field and local specific humidity are covariates of the distribution parameters. The GEV parameters are estimated using data from 1949 to 1999 and the r-largest method. This statistical downscaling procedure is found to yield skill over the southern and northern West Coast, central United States, and areas of western and eastern Canada when tested with independent data. The projected changes in covariates or predictors are obtained from transient climate change simulations conducted with the Canadian Centre for Climate Modelling and Analysis (CCCma) Coupled General Circulation Model, version 3.1 (CGCM3.1) forced by the Intergovernmental Panel on Climate Change (IPCC) A2 forcing scenario. They are then used to derive the GEV distribution parameters for the period 2050–99. The projected frequency of the current 20-yr return maximum daily precipitation for that period suggests that extreme precipitation risk will increase heavily over the south and central United States but decrease over the Canadian prairies. The difference between the statistical downscaling results and those estimated using GCM simulation is also discussed.

scholarly journals Bridging the scale gap: Obtaining high-resolution stochastic simulations of gridded daily precipitation in a future climate

2021 ◽  
Author(s):  
Qifen Yuan ◽  
Thordis L. Thorarinsdottir ◽  
Stein Beldring ◽  
Wai Kwok Wong ◽  
Chong-Yu Xu
Keyword(s):  
Climate Change ◽  
High Resolution ◽  
Climate Model ◽  
Daily Precipitation ◽  
Regional Climate ◽  
Weather Generator ◽  
Future Climate ◽  
Stochastic Simulations ◽  
Climate Change Information ◽  
Historical Observation

Abstract. Climate change impact assessment related to floods, infrastructure networks and water resources management applications requires realistic simulations of high-resolution gridded precipitation series under a future climate. This paper proposes to produce such simulations by combining a weather generator for high-resolution gridded daily precipitation, trained on historical observation-based gridded data product, with coarser scale climate change information obtained using a regional climate model. The climate change information can be added to various components of the weather generator, related to both the probability of precipitation as well as the amount of precipitation on wet days. The information is added in a transparent manner, allowing for an assessment of the plausibility of the added information. In a case study of nine hydrological catchments in central Norway with the study areas covering 1000–5500 km2, daily simulations are obtained on a 1 km grid for a period of 19 years. The method yields simulations with realistic temporal and spatial structures and outperforms empirical quantile delta mapping in terms of marginal performance.

scholarly journals Coupling a Markov Chain and Support Vector Machine for At-Site Downscaling of Daily Precipitation

2017 ◽  
Vol 18 (9) ◽  
pp. 2385-2406 ◽  
Author(s):  
Yu-Kun Hou ◽  
Hua Chen ◽  
Chong-Yu Xu ◽  
Jie Chen ◽  
Sheng-Lian Guo
Keyword(s):  
Climate Change ◽  
Markov Chain ◽  
Statistical Downscaling ◽  
Daily Precipitation ◽  
Global Climate Models ◽  
Support Vector ◽  
Statistical Downscaling Method ◽  
Calibration And Validation ◽  
Downscaling Method ◽  
Calibration Methods

Abstract Statistical downscaling is useful for managing scale and resolution problems in outputs from global climate models (GCMs) for climate change impact studies. To improve downscaling of precipitation occurrence, this study proposes a revised regression-based statistical downscaling method that couples a support vector classifier (SVC) and first-order two-state Markov chain to generate the occurrence and a support vector regression (SVR) to simulate the amount. The proposed method is compared to the Statistical Downscaling Model (SDSM) for reproducing the temporal and quantitative distribution of observed precipitation using 10 meteorological indicators. Two types of calibration and validation methods were compared. The first method used sequential split sampling of calibration and validation periods, while the second used odd years for calibration and even years for validation. The proposed coupled approach outperformed the other methods in downscaling daily precipitation in all study periods using both calibration methods. Using odd years for calibration and even years for validation can reduce the influence of possible climate change–induced nonstationary data series. The study shows that it is necessary to combine different types of precipitation state classifiers with a method of regression or distribution to improve the performance of traditional statistical downscaling. These methods were applied to simulate future precipitation change from 2031 to 2100 with the CMIP5 climate variables. The results indicated increasing tendencies in both mean and maximum future precipitation predicted using all the downscaling methods evaluated. However, the proposed method is an at-site statistical downscaling method, and therefore this method will need to be modified for extension into a multisite domain.

scholarly journals Reassessing Statistical Downscaling Techniques for Their Robust Application under Climate Change Conditions

2013 ◽  
Vol 26 (1) ◽  
pp. 171-188 ◽  
Author(s):  
J. M. Gutiérrez ◽  
D. San-Martín ◽  
S. Brands ◽  
R. Manzanas ◽  
S. Herrera
Keyword(s):  
Climate Change ◽  
Statistical Downscaling ◽  
Climate Model ◽  
Global Climate ◽  
Global Climate Model ◽  
Lower Troposphere ◽  
Temperature Inversion ◽  
Max Planck ◽  
Near Surface ◽  
Climate Conditions

Abstract The performance of statistical downscaling (SD) techniques is critically reassessed with respect to their robust applicability in climate change studies. To this end, in addition to standard accuracy measures and distributional similarity scores, the authors estimate the robustness of the methods under warming climate conditions working with anomalous warm historical periods. This validation framework is applied to intercompare the performances of 12 different SD methods (from the analog, weather typing, and regression families) for downscaling minimum and maximum temperatures in Spain. First, a calibration of these methods is performed in terms of both geographical domains and predictor sets; the results are highly dependent on the latter, with optimum predictor sets including near-surface temperature data (in particular 2-m temperature), which appropriately discriminate cold episodes related to temperature inversion in the lower troposphere. Although regression methods perform best in terms of correlation, analog and weather generator approaches are more appropriate for reproducing the observed distributions, especially in case of wintertime minimum temperature. However, the latter two families significantly underestimate the temperature anomalies of the warm periods considered in this work. This underestimation is found to be critical when considering the warming signal in the late twenty-first century as given by a global climate model [the ECHAM5–Max Planck Institute (MPI) model]. In this case, the different downscaling methods provide warming values with differences in the range of 1°C, in agreement with the robustness significance values. Therefore, the proposed test is a promising technique for detecting lack of robustness in statistical downscaling methods applied in climate change studies.

scholarly journals Why Do Precipitation Intensities Tend to Follow Gamma Distributions?

2019 ◽  
Vol 76 (11) ◽  
pp. 3611-3631 ◽  
Author(s):  
Cristian Martinez-Villalobos ◽  
J. David Neelin
Keyword(s):  
Power Law ◽  
Daily Precipitation ◽  
Present Theory ◽  
Physical Parameters ◽  
Precipitation Distribution ◽  
Climate Conditions ◽  
Gamma Distributions ◽  
Wide Range ◽  
Distribution Shifts ◽  
Averaging Time

Abstract The probability distribution of daily precipitation intensities, especially the probability of extremes, impacts a wide range of applications. In most regions this distribution decays slowly with size at first, approximately as a power law with an exponent between 0 and −1, and then more sharply, for values larger than a characteristic cutoff scale. This cutoff is important because it limits the probability of extreme daily precipitation occurrences in current climate. There is a long history of representing daily precipitation using a gamma distribution—here we present theory for how daily precipitation distributions get their shape. Processes shaping daily precipitation distributions can be separated into nonprecipitating and precipitating regime effects, the former partially controlling how many times in a day it rains, and the latter set by single-storm accumulations. Using previously developed theory for precipitation accumulation distributions—which follow a sharper power-law range (exponent < −1) with a physically derived cutoff for large sizes—analytical expressions for daily precipitation distribution power-law exponent and cutoff are calculated as a function of key physical parameters. Precipitating and nonprecipitating regime processes both contribute to reducing the power-law range exponent for the daily precipitation distribution relative to the fundamental exponent set by accumulations. The daily precipitation distribution cutoff is set by the precipitating regime and scales with moisture availability, with important consequences for future distribution shifts under global warming. Similar results extend to different averaging periods, providing insight into how the precipitation intensity distribution evolves as a function of both underlying physical climate conditions and averaging time.

scholarly journals Configuration and Intercomparison of Deep Learning Neural Models for Statistical Downscaling

2019 ◽  
Author(s):  
Jorge Baño-Medina ◽  
Rodrigo Manzanas ◽  
José Manuel Gutiérrez
Keyword(s):  
Climate Change ◽  
Deep Learning ◽  
Statistical Downscaling ◽  
Linear Models ◽  
Added Value ◽  
Climate Conditions ◽  
Continental Scale ◽  
Temperature And Precipitation ◽  
Spatial Features ◽  
Learning Techniques

Abstract. Deep learning techniques (in particular convolutional neural networks, CNNs) have recently emerged as a promising approach for statistical downscaling due to their ability to learn spatial features from huge spatio-temporal datasets. However, existing studies are based on complex models, applied to particular case studies and using simple validation frameworks, which makes difficult a proper assessment of the (possible) added value offered by these techniques. As a result, these models are usually seen as black-boxes generating distrust among the climate community, particularly in climate change problems. In this paper we undertake a comprehensive assessment of deep learning techniques for continental-scale statistical downscaling, building on the VALUE validation framework. In particular, different CNN models of increasing complexity are applied for downscaling temperature and precipitation over Europe, comparing them with a few standard benchmark methods from VALUE (linear and generalized linear models) which have been traditionally used for this purpose. Besides analyzing the adequacy of different components and topologies, we also focus on their extrapolation capability, a critical point for their possible application in climate change studies. To do this, we use a warm test period as surrogate of possible future climate conditions. Our results show that, whilst the added value of CNNs is mostly limited to the reproduction of extremes for temperature, these techniques do outperform the classic ones for the case of precipitation for most aspects considered. This overall good performance, together with the fact that they can be suitably applied to large regions (e.g. continents) without worrying about the spatial features being considered as predictors, can foster the use of statistical approaches in international initiatives such as CORDEX.

scholarly journals On the manifestations of climate change in the wind regime in the territory of Syunik marz (Armenia)

2021 ◽  
Vol 21 (1) ◽  
pp. 22-26
Author(s):  
Vardui G. Margaryan ◽  
◽  
Gorik D. Avetisyan ◽  
Gor A. Khachatryan ◽  
Pargev N. Margaryan ◽  
...  
Keyword(s):  
Climate Change ◽  
Wind Speed ◽  
Wind Energy ◽  
National Economy ◽  
Climate Changes ◽  
Source Material ◽  
Weather Station ◽  
Wind Regime ◽  
Weather Stations ◽  
Building Standards

The manifestations of climate change in the regularities of the wind regime in the territory of Syunik marz of Armenia are considered. Monthly wind data for the period 1966–2018 of six weather stations were used as a source material. It was found that at all meteostations, except for Kapan, currently operating in the territory of Syunik marz there is a tendency towards a decrease in wind speed for 1966–2018. The number of cases of wind directions and calmness also fluctuates, due to climate changes, which are presented in the work on the example of the Goris weather station. The results obtained can be used to monitor the climate in the territory of Syunik marz in the climatic service of the national economy branches, in the development of wind energy cadastres of territories, in the adjustment of building standards.

Sign in / Sign up

Export Citation Format

Share Document