Weather-driven simulation models usually require daily solar radiation (Q, MJ m-2 day-1) as a driving variable, but this parameter is recorded on a relatively sparse network throughout the world. To develop a means of estimating Q for locations in northern Australia where measurements have not been made, regression analyses were conducted on data from 15 agrometeorological stations, spanning a range from tropical to temperate latitudes. Relationships were sought between Q and extraterrestrial radiation (Qext), calculated from latitude and time of year, including coefficients to account for reduced radiation when rainfall was recorded. The final form adopted was: Qt=a Qext + b R′t-1 + c R′t+1 where Qt (MJ m-2 day-1) is the radiation recorded on day t, R′ is the transformed rainfall data and its subscripts t - 1, t ̀ and t + 1 refer to the previous, current and next days, and a, b, c and d are coefficients determined by the regression. The transformation used to calculate R′ from rainfall data was to encode raindays (≥ 0.1 mm day-1) as '1.0', and rainless days as '0.0'. Each rainday reduced Q by an average of 8.6 MJ m-2 day-1 over a 3-day period. This comprised a reduction of 1.8 MJ m-2 day-1 (d coefficient) on the day before rainfall was recorded, 4.7 MJ m-2 day-1 on the day rain fell (c), and 2.1 MJ m-2 day-1 on the day after (b). On the remaining days, Q averaged 66% of Qext (a). These coefficients were extremely stable across most of the 15 stations examined, giving confidence in their application to other stations within the region. No attempt was made to reintroduce the daily variation in Q. The technique should find application for models which are sensitive to the monthly but not the daily variation in Q, by supplying estimates of Q where the parameter has either not been recorded or is missing through instrument failure. The methodology of finding regionally stable coefficients for meteorological variables could have application elsewhere in the world where the network of recording stations is also sparse. © 1990.
