STATISTICAL ANALYSIS OF ET EXPLANATORY VARIABLES IN CONFLUENCE STATE, LOKOJA, SAVANNAH REGION OF NIGERIA

Abstract

This observational study investigated trends of Lokoja climate variables relating to evapotranspiration from 1989 to 2019. Studying evapotranspiration in the savannah region of Nigeria is critical for understanding the local climate, water availability, and ecosystem dynamics, and can inform sustainable development and adaptation strategies in the face of climate change. The FAO-56 Penman-Monteith model (P-M) was used to estimate the ET from the ERA5 reanalysis monthly mean temperature, precipitation, wind speed and atmospheric pressure data of Lokoja. Statistically significant explanatory variables were determined using multiple regression analysis, and multicollinearity and heteroscedasticity tests were conducted on the results. Clustered column charts used to visualize the performance of the model revealed that increase in precipitation did not automatically translate to increase in ET. Linear regression of ET against temperature revealed that temperature explains approximately 29% of the variability in ET.  At 95% confidence level and 251 degree of freedom, the R² (0.98) with standard error 0.11 indicate that the statistical analysis of the ET explanatory variables is robust and reliable, and the model is able to accurately predict the values of the ET. The multiple regression analysis result revealed that a mean daily increase of ET, starting from 0.54mm per day, is affected positively by average wind speed (u) of about 0.31mm/day for every unit increase in wind speed (u) at two meters height, 0.27mm/day for a unit increase in net radiation (Rn), 0.98mm/day for a unit increase in vapour pressure deficit (VPD), but a decrease of 0.01mm/day for every unit increase...