Year: 2007
Type: article
Abstract: The amount of solar radiation received at the earth's surface (R s , MJ m −2 day −1 ) sets the potential for crop growth, is a key driver of evapotranspiration and, consequently, is an important input variable for crop simulation models. However, values for R s are seldom included in historical weather records. This paper assesses two methods proposed to estimate R s as the product of extraterrestrial radiation (R.J and a transmissivity coefficient (T t ), using data from 10 New Zealand sites (37.5°S‐6.1°S). Both the Hargreaves‐Samani (HS) and the Weiss et al. (WS) models use the difference between daily maximum and minimum temperature (AT) to estimate T t . The HS model uses the square root of AT and a single empirical coefficient, k. Reported values of k are 0.16 for continental and 0.19 for coastal locations. The WS model uses the AT squared and a single empirical coefficient, b , to estimate R s . Reported values of b range between 0.20 and 0.75. The current research used a sensitivity analysis of k and b to determine the accuracy of the two models at the 10 sites and to determine the optimum values for the coefficients k and b. Sensitivity analyses showed the HS model gave more accurate estimates of R s than the WS model. Root mean squared deviations (RMSD) were 2.43 and 3.50 (MJ m −2 ) across all 10 sites for the HS and WS models, respectively. The range of values of k in the HS model was 0.14–0.17, which is typical of continental sites. The mean value (k =0.15) is suggested as appropriate for other New Zealand sites. The range for b in the WS model was 0.35–0.65. The principal use for these estimated R s values would be in crop simulation models. Therefore, a simulation analysis was undertaken with these data. This compared simulated wheat (Triticum aestivum) biomass and grain yields with measured yields from an experiment at Lincoln, New Zealand. Simulations used observed R s values and those estimated with the HS and WS models. Simulated biomass and grain yields from the HS model were virtually identical to those using the observed R s data, whereas those from the WS model were 5–6% less accurate.
Authors: A. L. Fletcher, D. J. Moot
Institutions: Lincoln University, University of Florida
Language: English
Field: Energy
Domain: Physical Sciences
SDGs: Climate action, Zero hunger
Open Access status: bronze
OpenAlex ID: W2019815142
