Estimation of diurnal patterns of global solar radiation, temperature, relative humidity and wind speed from daily datasets at a humid tropical location

Abstract

The primary focus of this study is to simulate, characterize and validate diurnal patterns of global solar radiation (GSR), temperature, relative humidity and wind speed from daily ground observations from 2017 to 2020 at Obafemi Awolowo University, Ile-Ife, Nigeria. The study also estimates diurnal variations of the meteorological parameters from three daily reanalyses (NASA, NCEP/NCAR and ERA5) and compares the results with hourly station observations (using performance evaluation indices such as mean bias, BIAS; percentage mean bias, PBIAS; root-mean-square-error, RMSE; Nash-Sutcliffe coefficient, NSE; normalised standard deviation, NSD and correlation coefficient, r). Results showed that the empirical models adequately captured the observed diurnal characteristics of the meteorological variables. Good estimates of diurnal patterns of GSR (BIAS = 6.77 Wm−2; RMSE =9.20 Wm−2; PBIAS = 5.68%; NSE =0.52; NSD = 0.62; r = 0.86), temperature (BIAS = -0.79°C; RMSE = 1.90°C; PBIAS = -2.43%; NSE =0.60; NSD =1.09; r = 0.86), and humidity (BIAS = 2.81%; RMSE =6.10%; PBIAS = 2.51%; NSE =0.43; NSD =0.83; r = 0.73) were obtained. Statistics suggested very strong model fits and close agreements with observations. Large and significant discrepancies (at p ≤ 0.05), however, were obtained for diurnal simulations of wind speed (BIAS = 0.28 ms−1; RMSE =0.40 ms−1; PBIAS = 9.79%; NSE = -0.10; NSD =1.36; r = 0.65). Furthermore, the model performances for hourly disaggregation of the parameters varied amongst the reanalyses. The findings provide good basis for generating sub-daily meteorological data resources with wide range of applications in hydrology, climate modelling, and plant growth simulation.