Downscaling Of Temperature Research Articles

Statistical Downscaling of Temperature with the Random Forest Model

The issues with downscaling the outputs of a global climate model (GCM) to a regional scale that are appropriate to hydrological impact studies are investigated using the random forest (RF) model, which has been shown to be superior for large dataset analysis and variable importance evaluation. The RF is proposed for downscaling daily mean temperature in the Pearl River basin in southern China. Four downscaling models were developed and validated by using the observed temperature series from 61 national stations and large-scale predictor variables derived from the National Center for Environmental Prediction–National Center for Atmospheric Research reanalysis dataset. The proposed RF downscaling model was compared to multiple linear regression, artificial neural network, and support vector machine models. Principal component analysis (PCA) and partial correlation analysis (PAR) were used in the predictor selection for the other models for a comprehensive study. It was shown that the model efficiency of the RF model was higher than that of the other models according to five selected criteria. By evaluating the predictor importance, the RF could choose the best predictor combination without using PCA and PAR. The results indicate that the RF is a feasible tool for the statistical downscaling of temperature.

MLR Based Statistical Downscaling of Temperature and Precipitation in Lidder Basin Region of India

In the present study, the multiple linear regression technique was employed to relate the monthly predictors obtained from the Global Climate Model namely CGCM3 (Canadian Centre for Climate Modelling and Analysis) with the monthly Predictands such as the locally observed precipitation and temperature at Pahalgam meteorological station which is located in the Lidder River Basin. Appropriate predictand-predictor relationships were found out for the region by carrying out sensitivity analysis .Regression equations were developed and subsequently future monthly and annual projections for precipitation, maximum temperature and minimum temperature for the entire 21st century were made. It was observed that at the end of the 21st century the CGCM3 model predicted an increase of 18.07% annual precipitation, whereas the average maximum and minimum annual temperatures were predicted to increase by 0.62°C and 0.76°C, respectively.

Improving the Downscaling of Diurnal Land Surface Temperatures Using the Annual Cycle Parameters as Disaggregation Kernels

The downscaling of geostationary diurnal thermal data can ease the lack of land surface temperature (LST) datasets that combine high spatial and temporal resolution. However, the downscaling of diurnal LST data is more demanding than single scenes. This is because the spatiotemporal interrelationships of the original LST data have to be preserved and accurately reproduced by the downscaled LST (DLST) data. To that end, LST disaggregation kernels/predictors that provide information about the spatial distribution of LST during different times of a day can prove especially useful. Such LST predictors are the LST Annual Cycle Parameters (ACPs). In this work, multitemporal ACPs are employed for downscaling daytime and nighttime ~4 km geostationary LST from SEVIRI (Spinning Enhanced Visible and Infrared Imager) down to 1 km. The overall goal is to assess if the use of the ACPs can improve the estimation of the diurnal range of DLST (daytime DLST minus nighttime DLST). The evaluation is performed by comparing the DLST diurnal range maps with reference data from MODIS (Moderate Imaging Spectroradiometer) and also with data retrieved from a modified version of the TsHARP (Thermal Sharpening) algorithm. The results suggest that the ACPs increase the downscaling performance, improve the estimation of diurnal DLST range and produce more accurate spatial patterns.

Spring and summer monthly MODIS LST is inherently biased compared to air temperature in snow covered sub-Arctic mountains

Satellite-derived land surface temperature (skin temperature) provides invaluable information for data-sparse high elevation and Arctic regions. However, the relationship between satellite-derived clear-sky skin temperature and various downscaled air temperature products for snow covered sub-Arctic alpine regions remain poorly understood, such that trend analysis or air temperature product integration is difficult. We compared monthly average air temperatures from two independent downscaled temperature products to MODIS Land Surface Temperature (LST) and air temperature at nine meteorological stations situated above tree-line in the southwest Yukon, Canada, between May and August 2008 for a full range of snow cover fractions. We found that both downscaled products generally agreed with LST for the low elevation, snow-free, vegetation classes. However, a systematic cold bias in Average LST emerged for snow fractions greater than approximately 40%, and this bias increased in magnitude as snow cover increased. In these situations the downscaled air temperatures were 5–7°C warmer than Average LST for snow fractions of >90%, and this pattern was largely independent of the number of measurements of LST within a month. Maximum LST was similar to average air temperatures for high snow fractions, but Minimum LST was colder by 10°C or more for all snow fractions. Consequently, the average of Maximum and Minimum LST produces the cold bias, compared to air temperature, for high snow cover fractions. Air temperature measured at nine meteorological monitoring stations located between elevations of 1408–2690m, on land cover classes Barren, Sparsely Vegetated or Permanent Snow and Ice, confirmed the cold bias results when incorporating Minimum LST in monthly averages. For snow fractions of <40% the RMSE for all of the temperature products was <2.5°C when compared to station air temperature and all biases were positive and <2.0°C. For snow fractions of >40%, the average LST bias became strongly negative at −4.5°C, and the RMSE increased to 6.1°C, whereas the downscaled products bias and RMSE were similar to those from snow fractions of <40%. A weak warm bias for all the temperature products occurred for small snow fractions over non-forested land cover classes. Downscaled air temperature fields show physically real differences from Average LST in spring and summer, caused by snow cover and the interplay of Maximum and Minimum LST. These findings indicate that the integration of MODIS LST with downscaled air temperature products or local air temperature requires the incorporation of snow cover.

An extended linear scaling method for downscaling temperature and its implication in the Jhelum River basin, Pakistan, and India, using CMIP5 GCMs

In this study, the linear scaling method used for the downscaling of temperature was extended from monthly scaling factors to daily scaling factors (SFs) to improve the daily variations in the corrected temperature. In the original linear scaling (OLS), mean monthly SFs are used to correct the future data, but mean daily SFs are used to correct the future data in the extended linear scaling (ELS) method. The proposed method was evaluated in the Jhelum River basin for the period 1986–2000, using the observed maximum temperature (Tmax) and minimum temperature (Tmin) of 18 climate stations and the simulated Tmax and Tmin of five global climate models (GCMs) (GFDL-ESM2G, NorESM1-ME, HadGEM2-ES, MIROC5, and CanESM2), and the method was also compared with OLS to observe the improvement. Before the evaluation of ELS, these GCMs were also evaluated using their raw data against the observed data for the same period (1986–2000). Four statistical indicators, i.e., error in mean, error in standard deviation, root mean square error, and correlation coefficient, were used for the evaluation process. The evaluation results with GCMs’ raw data showed that GFDL-ESM2G and MIROC5 performed better than other GCMs according to all the indicators but with unsatisfactory results that confine their direct application in the basin. Nevertheless, after the correction with ELS, a noticeable improvement was observed in all the indicators except correlation coefficient because this method only adjusts (corrects) the magnitude. It was also noticed that the daily variations of the observed data were better captured by the corrected data with ELS than OLS. Finally, the ELS method was applied for the downscaling of five GCMs’ Tmax and Tmin for the period of 2041–2070 under RCP8.5 in the Jhelum basin. The results showed that the basin would face hotter climate in the future relative to the present climate, which may result in increasing water requirements in public, industrial, and agriculture sectors; change in the hydrological cycle and monsoon pattern; and lack of glaciers in the basin.

A Dual-Grid Method for the Upscaling of Solid-Based Thermal Reactive Flow, With Application to the In-Situ Conversion Process

Summary Thermal-reactive compositional-flow simulation in porous media is essential to model thermal-oil-recovery processes for extraheavy-hydrocarbon resources, and an example is the in-situ conversion process (ICP) developed by Shell for oil-shale production. Computational costs can be very high for such a complex system, which makes direct fine-scale simulations prohibitively time-consuming for large field-scale applications. This motivates the use of coarse grids for thermal-reactive compositional-flow simulation. However, significant errors are introduced by use of coarse-scale models without carefully computing the appropriate coarse parameters. In this paper, we develop an innovative dual-grid method to effectively capture the fine-scale reaction rates in coarse-scale ICP-simulation models. In our dual-grid method, coupled thermal-reactive compositional-flow equations are solved only on the coarse scale, with the kinetic parameters (frequency factors) calculated on the basis of fine-scale computations, such as temperature downscaling and fine-scale reaction-rate calculation. A dual-grid treatment for the heater-well model is also developed with coarse-scale heater-well indices calculated on the basis of fine-scale well results. The dual-grid heater-well treatment is able to provide accurate heater temperatures. The newly developed dual-grid method is applied to realistic cross-sectional ICP-pattern models with a vertical production well and multiple horizontal heater wells operated subject to fixed and time-varying heater powers. It is shown that the dual-grid model delivers results that are in close agreement with the fine-scale reference results for all quantities of interest. Despite the fact that the dual-grid method is implemented at the simulation-deck level, by use of the flexible scripting and monitor functionalities of our proprietary simulation package, significant computational improvements are achieved for all cases considered.

Quantification of the relative role of land-surface processes and large-scale forcing in dynamic downscaling over the Tibetan Plateau

Dynamical downscaling modeling (DDM) is important to understand regional climate change and develop local mitigation strategies, and the accuracy of DDM depends on the physical processes involved in the regional climate model as well as the forcing datasets derived from global models. This study investigates the relative role of the land surface schemes and forcing datasets in the DDM over the Tibet Plateau (TP), a region complex in topography and vulnerable to climate change. Three Weather Research and Forecasting model dynamical downscaling simulations configured with two land surface schemes [Noah versus Noah with multiparameterization (Noah-MP)] and two forcing datasets are performed over the period of 1980–2005. The downscaled temperature and precipitation are evaluated with observations and inter-compared regarding temporal trends, spatial distributions, and climatology. Results show that the temporal trends of the temperature and precipitation are determined by the forcing datasets, and the forcing dataset with the smallest trend bias performs the best. Relative to the forcing datasets, land surface processes play a more critical role in the DDM over the TP due to the strong heating effects on the atmospheric circulation from a vast area at exceptionally high elevations. By changing the vertical profiles of temperature in the atmosphere and the horizontal patterns of moisture advection during the monsoon seasons, the land surface schemes significantly regulate the downscaled temperature and precipitation in terms of climatology and spatial patterns. This study emphasizes the selection of land surface schemes is of crucial importance in the successful DDM over the TP.

High-Resolution Dynamical Downscaling of ERA-Interim Using the WRF Regional Climate Model for the Area of Poland. Part 1: Model Configuration and Statistical Evaluation for the 1981\u20132010 Period

In this work, we present the results of high-resolution dynamical downscaling of air temperature, relative humidity, wind speed and direction, for the area of Poland, with the Weather Research and Forecasting (WRF) model. The model is configured using three nested domains, with spatial resolution of 45 km × 45 km, 15 km × 15 km and 5 km × 5 km. The ERA-Interim database is used for boundary conditions. The results are evaluated by comparison with station measurements for the period 1981–2010. The model is capable of reproducing the main climatological features of the study area. The results are in very close agreement with the measurements, especially for the air temperature. For all four meteorological variables, the model performance captures seasonal and daily cycles. For the air temperature and winter season, the model underestimates the measurements. For summer, the model shows higher values, compared with the measurements. The opposite is the case for relative humidity. There is a strong diurnal pattern in mean error, which changes seasonally. The agreement with the measurements is worse for the seashore and mountain areas, which suggests that the 5 km × 5 km grid might still have an insufficient spatial resolution. There is no statistically significant temporal trend in the model performance. The larger year-to-year changes in the model performance, e.g. for the years 1982 and 2010 for the air temperature should therefore be linked with the natural variability of meteorological conditions.

Evaluation of Climate Change Impact on Runoff: A Case Study

Background/Objectives: In this study we have investigating Evaluation of Climate Change Impact on runoff. Methods/Statistical Analysis: In this study, outputs of HADCM3 (Hadley Center Coupled Model, version 3) GCM (General Circulation Model) under A2 scenario is used to investigate the climate change impact on runoff of Gaveh Rood catchment in western Iran. The SDSM (Statistical Down Scaling Model) is employed to statistically downscale climate variables of temperature and precipitation. A runoff-rainfall Hydrological Model, named HYMOD is used for developing runoff time series under climate change impacts regarding the downscaled climate variables. Findings: The performance of calibrated SDSM model shows its well performance in rainfall and temperature downscaling in the study area. The downscaling results show considerable changes in temperature and precipitation in future under climate change impacts. In the period of 2020-2099, the mean temperature and precipitation is increased by 5-8.5% and 3-5%, respectively, relative to the observed period of 1989-2000. The calibration and validation results of the developed HYMOD show that it the mean runoff decreased by 7-16% (11.5%, on average) in the predicted period relative to the observed period. Applications/Improvement: These changes highly affect the water resources systems and special attention should be given to them in future planning of water resources.

Dynamical Downscaling of ERA-Interim Temperature and Precipitation for Alaska

AbstractThe European Centre for Medium-Range Weather Forecasts interim reanalysis (ERA-Interim) has been downscaled using a regional model covering Alaska at 20-km spatial and hourly temporal resolution for 1979–2013. Stakeholders can utilize these enhanced-resolution data to investigate climate- and weather-related phenomena in Alaska. Temperature and precipitation are analyzed and compared among ERA-Interim, WRF Model downscaling, and in situ observations. Relative to ERA-Interim, the downscaling is shown to improve the spatial representation of temperature and precipitation around Alaska’s complex terrain. Improvements include increased winter and decreased summer higher-elevation downscaled seasonal average temperatures. Precipitation is also enhanced over higher elevations in all seasons relative to the reanalysis. These spatial distributions of temperature and precipitation are consistent with the few available gridded observational datasets that account for topography. The downscaled precipitation generally exceeds observationally derived estimates in all seasons over mainland Alaska, and it is less than observations in the southeast. Temperature biases tended to be more mixed, and the downscaling reduces absolute bias at higher elevations, especially in winter. Careful selection of data for local site analysis from the downscaling can help to reduce these biases, especially those due to inconsistencies in elevation. Improved meteorological station coverage at higher elevations will be necessary to better evaluate gridded downscaled products in Alaska because biases vary and may even change sign with elevation.

Statistical Downscaling of Wintertime Temperatures over South Korea

AbstractReanalysis data have global coverage and faithfully render large-scale phenomena. On the other hand, regional and small-scale characteristics of atmospheric variability are poorly resolved. In an attempt to improve reanalysis data for regional use, a statistical downscaling strategy is developed based on cyclostationary empirical orthogonal function (CSEOF) analysis. The developed algorithm is applied to the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data and to the European Centre for Medium-Range Weather Forecast (ECMWF) Interim Re-Analysis (ERA-Interim) data in order to produce winter temperatures at 60 Korea Meteorological Administration (KMA) stations over the Korean Peninsula. The developed downscaling algorithm is evaluated by predicting winter daily temperatures from 17 November to 16 March for 35 years (1979–2014). For validating the downscaling algorithm the jackknife method is used, in which winter daily temperature is predicted over a 1-yr period not used for training. This procedure is repeated for the entire data period. The mean and variance of the resulting downscaled temperatures match reasonably well with those of the KMA measurements. Validation based on correlation and error variance shows that the temperatures at 60 KMA stations are faithfully reproduced based on coarse reanalysis data. The utility of this technique for downscaling model predictions based on future scenarios is also addressed.

Statistical downscaling of precipitation and temperature using sparse Bayesian learning, multiple linear regression and genetic programming frameworks

This study attempted to investigate two approaches to downscale temperature and four approaches to downscale precipitation. The first approach was an implementation of multiple linear regression (MLR) in the form of backward stepwise regression. The second approach applied canonical correlation analysis (CCA) with a sparse Bayesian learning (SBL) approach called relevance vector machine (RVM). For precipitation downscaling, two additional approaches which combined genetic programming (GP) as the predictor processing method with sparse Bayesian learning (SBLGP) and multiple linear regression (MLRGP) were also presented. The results showed that SBL outperformed MLR in downscaling temperature. For all stations, temperature was better downscaled than precipitation. For precipitation downscaling, the SBLGP approach outperformed all other approaches. MLRGP, on the other hand, did not bring about much improvement in the results and was in many cases outperformed by MLR.

Evaluation of ensemble NWP models for dynamical downscaling of air temperature over complex topography in a hot climate: A case study from the Sultanate of Oman

This paper evaluates the use of ensemble numerical weather prediction (NWP) models for dynamical downscaling of temperature over a complex, hot region. This approach delivers information about the uncertainty of the NWP models and provides probabilistic information for comparison with the currently used single NWP model. An ensemble system was constructed using four members with a 7km resolution over Oman. Two limited-area models (LAMs), the high-resolution model (HRM) and the model from the Consortium for Small-Scale Modeling (COSMO) formed the ensemble members. The two LAMs were derived and initialized using the general circulation model (GCM) data from the German Global Model (GME), which runs at 40km resolution, using two different initial atmospheric states. The first initial state was provided by the 3Dvar data assimilation system at the German Weather Service (Deutscher Wetterdienst, DWD), and the second initial state was provided from the reanalysis data (ERA-Interim) from the European Centre for Medium-Range Weather Forecasts (ECMWF). The results reveal the uncertainty in temperature prediction related to the uncertainty of the NWP models that were used and indicate that there is no best model for the entire domain. On average, the ensemble mean performed better than individual members.

Assessment of climate change downscaling and non-stationarity on the spatial pattern of a mangrove ecosystem in an arid coastal region of southern Iran

Mangrove wetlands exist in the transition zone between terrestrial and marine environments and have remarkable ecological and socio-economic value. This study uses climate change downscaling to address the question of non-stationarity influences on mangrove variations (expansion and contraction) within an arid coastal region. Our two-step approach includes downscaling models and uncertainty assessment, followed by a non-stationary and trend procedure using the Extreme Value Analysis (extRemes code). The Long Ashton Research Station Weather Generator (LARS-WG) model along with two different general circulation model (GCMs) (MIRH and HadCM3) were used to downscale climatic variables during current (1968–2011) and future (2011–2030, 2045–2065, and 2080–2099) periods. Parametric and non-parametric bootstrapping uncertainty tests demonstrated that the LARS-WGS model skillfully downscaled climatic variables at the 95 % significance level. Downscaling results using MIHR model show that minimum and maximum temperatures will increase in the future (2011–2030, 2045–2065, and 2080–2099) during winter and summer in a range of +4.21 and +4.7 °C, and +3.62 and +3.55 °C, respectively. HadCM3 analysis also revealed an increase in minimum (∼+3.03 °C) and maximum (∼+3.3 °C) temperatures during wet and dry seasons. In addition, we examined how much mangrove area has changed during the past decades and, thus, if climate change non-stationarity impacts mangrove ecosystems. Our results using remote sensing techniques and the non-parametric Mann–Whitney two-sample test indicated a sharp decline in mangrove area during 1972,1987, and 1997 periods (p value = 0.002). Non-stationary assessment using the generalized extreme value (GEV) distributions by including mangrove area as a covariate further indicated that the null hypothesis of the stationary climate (no trend) should be rejected due to the very low p values for precipitation (p value = 0.0027), minimum (p value = 0.000000029) and maximum (p value = 0.00016) temperatures. Based on non-stationary analysis and an upward trend in downscaled temperature extremes, climate change may control mangrove development in the future.

Statistical downscaling and projection of future temperature and precipitation change in middle catchment of Sutlej River Basin, India

Ensembles of two Global Climate Models (GCMs), CGCM3 and HadCM3, are used to project future maximum temperature (T Max), minimum temperature (T Min) and precipitation in a part of Sutlej River Basin, northwestern Himalayan region, India. Large scale atmospheric variables of CGCM3 and HadCM3 under different emission scenarios and the National Centre for Environmental Prediction/National Centre for Atmospheric Research reanalysis datasets are downscaled using Statistical Downscaling Model (SDSM). Variability and changes in T Max, T Min and precipitation under scenarios A1B and A2 of CGCM3 model and A2 and B2 of HadCM3 model are presented for future periods: 2020s, 2050s and 2080s. The study reveals rise in annual average T Max, T Min and precipitation under scenarios A1B and A2 for CGCM3 model as well as under A2 and B2 scenarios for HadCM3 model in 2020s, 2050s and 2080s. Increase in mean monthly T Min is also observed for all months of the year under all scenarios of both the models. This is followed by decrease in T Max during June, July August and September. However, the model projects rise in precipitation in months of July, August and September under A1B and A2 scenarios of CGCM3 model and A2 and B2 of HadCM3 model for future periods.

Statistical downscaling and attribution of air temperature change patterns in the Valencia region (1948–2011)

This study is based on the statistical downscaling and spatial interpolation of high-resolution temperatures (90 m) over the 1948–2011 period performed for the Valencia Region (east Iberian Peninsula) after considering local topographical factors in the fine-scale distribution of temperatures. The objective was to detect the areas that were potentially more vulnerable to air temperature change. This allowed the detection of local climate change patterns, which were analyzed and found to be consistent in spatial and temporal terms. These patterns indicate a more marked warming tendency in higher parts of reliefs and their slopes. However, this tendency is less pronounced in bottoms of valleys and on coastal plains, particularly for minimum temperatures, while the tendency for increasing maximum temperatures becomes more generalised. These patterns seem to connect well with regional changes in pressure fields, wind frequency, precipitation patterns and sea surface temperature.

Climate Change Impacts on Streamflow Drought: A Case Study in Tseng-Wen Reservoir Catchment in Southern Taiwan

The study aims to assess climate change impacts on streamflow drought in a catchment upstream of Tseng-Wen Reservoir which is the main water supplier in southern Taiwan. A singular-value-decomposition statistical downscaling method and a stochastic weather generator were applied to six different general circulation models for rainfall and temperature downscaling. Two emission climate change scenarios (A1B and B1) were used for rainfall and temperature projections during the two periods, 2010–2045 and 2081–2100. After rainfall and temperature projections, the HBV-based hydrological model was used to simulate future streamflows. By using the threshold level method, the characteristics of streamflow drought during the baseline period (1975–2000) and the future periods were calculated and compared. Important findings include: (1) the start and end times of scenario droughts occur earlier than those of baseline droughts; (2) streamflow under future scenarios during the dry period tends to decrease in January and February, but to increase in March and April; and (3) most of general circulation models’ results support that the drought duration and magnitude tend to decrease for a given return period in the future, and the occurrence probability of severer drought tends to decrease.

Using fuzzified regression trees for statistical downscaling and regionalization of near surface temperatures in complex terrain

High mountain regions are characterized by a large climatic heterogeneity which is not sufficiently represented by state-of-the-art climate models or reanalysis products. With regard to the increasing demand for high-resolution temperature data for climate impact studies, a statistical approach is presented, which allows estimating high-resolution near-surface temperature fields in complex terrain. High-resolution free air temperatures are derived from climate model data by considering the current stratification of the atmosphere. The residuals compared with in situ observation of near-surface temperatures are subsequently analyzed using a regression tree approach with suitable large-scale atmospheric and local-scale terrain parameters as predictors. The model identifies the predominant synoptic and topographic controls for the local-scale distribution of residuals and can be used to regionalize residual fields with high spatial resolution. The disadvantage that a tree-structured model generates stepwise constant predictant values can be overcome by integrating a fuzzifying routine. A fuzzified regression tree model was applied to analyze and predict the spatial and temporal variability of topographically induced temperatures for a target area in the Central Himalayas. Large-scale atmospheric variables, derived from the ERA-Interim reanalysis, and local terrain parameters were used as potential predictors. The model sufficiently identified the main influencing factors for the temperature heterogeneity. The potential solar insolation was found to be the predominant predictor, but also, hydroclimatic large-scale variables were found to be crucial. During clear nights, the model showed a distinct elevation dependency of residuals which indicates the importance of nocturnal cold air drainage and accumulation for the local-scale temperature distribution in the highly structured target area.

Statistical downscaling of temperature using three techniques in the Tons River basin in Central India

In this study, downscaling models were developed for the projections of monthly maximum and minimum air temperature for three stations, namely, Allahabad, Satna, and Rewa in Tons River basin, which is a sub-basin of the Ganges River in Central India. The three downscaling techniques, namely, multiple linear regression (MLR), artificial neural network (ANN), and least square support vector machine (LS-SVM), were used for the development of models, and best identified model was used for simulations of future predictand (temperature) using third-generation Canadian Coupled Global Climate Model (CGCM3) simulation of A2 emission scenario for the period 2001–2100. The performance of the models was evaluated based on four statistical performance indicators. To reduce the bias in monthly projected temperature series, bias correction technique was employed. The results show that all the models are able to simulate temperature; however, LS-SVM models perform slightly better than ANN and MLR. The best identified LS-SVM models are then employed to project future temperature. The results of future projections show the increasing trends in maximum and minimum temperature for A2 scenario. Further, it is observed that minimum temperature will increase at greater rate than maximum temperature.

Evaluation of climate change impacts on streamflow to a multiple reservoir system using a data-based mechanistic model

Recent investigations have demonstrated scientists' consensus on the increase in global mean temperature and climate variability. These changes alter the hydro-climatic condition of regions. Investigation of surface water changes is an important issue in water resources planning as well as for the operation of reservoirs. In this study a data-based mechanistic (DBM) model has been used for daily streamflow simulation. This model is a data-driven statistical base simulation model that can take advantage of additional climate variables with time variable configurations. The model has been developed for simulation of streamflow to three reservoirs, located in central Iran, using the daily rainfall, temperature and streamflow data. Comparison of the DBM results with the autoregressive integrated moving average model, as an alternative model, shows its higher performance. To include climate change impacts in study, an artificial neural network-based statistical downscaling model is developed for rainfall and temperature downscaling. The downscaled temperature and rainfall data under climate change scenarios based on HadCM3 general circulation model outputs are used to evaluate the climate change impacts on streamflow for the 2000–2050 time horizon. The results demonstrate the considerable impact of climate change on streamflow variability with significantly different behaviour in the three adjacent basins.