Estimating daily solar radiation in the Argentine Pampas

Solar radiation is an important input to crop growth models used for risk management and assessment purposes. Methods are explored to estimate daily solar radiation in the Argentine Pampas, one of the most important agricultural areas in the world. Two scenarios are considered: (i) sunshine duration data are available for a given location, or (ii) only daily temperature (minimum and maximum) and precipitation records exist. If sunshine duration data are available, an association between this quantity and atmospheric transmissivity yields daily radiation estimates with a root mean square error (RMSE) of 1.5 MJ m−2 per day. Without sunshine duration records, daily temperature and precipitation can be used to estimate atmospheric transmittance and then compute daily radiation values. A model linking predictors that are proxies of cloudiness and atmospheric humidity to atmospheric transmittance was fitted using Generalized Additive Models (GAMs), a modern statistical technique that does not assume any a priori functional forms for the association between predictors and predictand. The errors in radiation estimates using temperature and precipitation are larger (RMSE of 3.2 MJ m−2 per day) than those derived from sunshine duration, but they are comparable to results for other locations and methods. Most importantly, daily radiation estimates have small bias and the errors show no systematic patterns with season or other variables.