Assessment of three models for estimating daily net radiation in southern Africa

Abstract Accurate quantification of net radiation flux ( R n ) is of paramount importance for the estimation of reference evapotranspiration ( E T 0 ) rate, which is used to estimate crop water use. A widely recommended Penman-Monteith procedure outlined in the Food and Agriculture Organization (FAO) Irrigation and Drainage Paper No. 56 for estimation of R n (FAO56- R n ) is often used to estimate R n . However, the FAO56- R n model is data-intensive and also requires site-specific calibrations of coefficients to attain a high level of accuracy of R n estimates. These coefficients are often used without site-specific calibrations as a result of' the lack of large and representative radiative flux measurements in data-scarce regions such as southern Africa and R n estimates are therefore considered questionable. Assessment of different models in-situ measurements of R n is critical to identify an alternative approach that could be used for accurate estimation of R n with minimal data input and without any site-specific calibrations in this region. In this study, two new R n models, which differ only in the procedures used to compute atmospheric emissivity were proposed. The first model requires measurements of solar irradiance ( R s ) maximum and minimum air temperatures ( T a i r m i n and T a i r m a x ) while the second model requires additional measurements of relative humidity ( R H m i n and R H m a x ) for estimation of the actual vapour pressure ( e a i r ). Two new R n models and a widely recommended FAO56- R n model were evaluated using daily R n measurements acquired from five sites which represent different climatic and land cover conditions of southern Africa. The results showed that the first model performed better than all the evaluated models at four sites, with regression coefficient ( r 2 ) values greater than 0.90 and index of agreement ( d ) values greater than 0.97. These findings suggest that the first model presented here is the most promising and suitable to estimate R n with minimum input data in southern Africa without any site-specific calibrations. The findings of this study can be used to inform the decision on selecting a model to be used for reliable estimates of R n for improved estimation of crop water requirement in climatic conditions similar to this region.