Evaluation of NASA POWER Climatic Data against Ground-Based Observations in The Mediterranean and Continental Regions of Turkey

Abstract

The weather reanalysis datasets are very advantageous data types worldwide that fill the gaps of missed measuring data and are alternatives that compensate for the scarcity of observed climate data. The main purpose of this study was to evaluate the effect of horizontal distance, altitude, and climatic regions compared to sea level on NASA POWER reanalysis data for daily temperature variables, relative humidity, and wind speed observed in meteorology stations in the Mediterranean and Continental regions of Turkey. For this purpose, three different meteorology stations (Antalya airport, Elmalı, Teffenni) from the Mediterranean region with different distances and elevations compared to sea level and one station (Ankara) far from the Mediterranean region with continental climate were selected. The statistical approach used to compare observed and estimated values in this study was determination coefficient (R2), Nash-Sutcliffe Efficiency (NSE), Root Mean Square Error (RMSE), Normalized Root Mean Square Error (NRMSE), and Mean Bias Error (MBE). The results showed a high relation between the POWER reanalysis dataset and observed data for all parameters except wind speed. For daily maximum, minimum and mean temperature, the R2 and NSE achieved higher than 0.91 and 0.88 respectively, while the mean bias error MBE ranged between -3 °C up to +2 °C and the RMSE was less than 4 °C in all stations. Additionally, POWER estimated data correlation accuracy for temperature variables increased toward higher altitudes in the study area. Similarly, this performance was followed by relative humidity, increasing relation accuracy toward higher elevated regions. The R2 was higher than 0.69 in higher altitudes and less than 0.4 in lower elevations. The MBE for relative humidity ranges -2% in Antalya to +9% in Ankara, and the RMSE attained less than 13.81% in all regions. The POWER daily wind speed did not show relation with observed data without adjusting for elevation and seasonal bias correction. Overall, it was concluded that the NASA POWER dataset could predict temperature and relative humidity over study area and give a promising result if used in research, water, and agricultural decision-making where observation data are not available.