Predicting average regional yield and production of wheat in the Argentine Pampas by an artificial neural network approach
A regional analysis of the effects of soil and climate factors on wheat yield was performed in the Argentine Pampas in order to obtain models suitable for yield estimation and regional grain production prediction. Soil data from soil surveys and climate data from meteorological records were employed...
| Autor: | |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2009 |
| País: | Argentina |
| Institución: | Consejo Nacional de Investigaciones Científicas y Técnicas |
| Repositorio: | CONICET Digital (CONICET) |
| Idioma: | inglés |
| OAI Identifier: | oai:ri.conicet.gov.ar:11336/112424 |
| Acceso en línea: | http://hdl.handle.net/11336/112424 |
| Access Level: | acceso abierto |
| Palabra clave: | ARGENTINE PAMPAS WHEAT YIELD ESTIMATION https://purl.org/becyt/ford/4.1 https://purl.org/becyt/ford/4 |
| Sumario: | A regional analysis of the effects of soil and climate factors on wheat yield was performed in the Argentine Pampas in order to obtain models suitable for yield estimation and regional grain production prediction. Soil data from soil surveys and climate data from meteorological records were employed. Grain production information from statistics at county level was integrated at a geomorphological level. The Pampas was divided into 10 geographical units and data from 10 growing season were used (1995-2004). Surface regression and artificial neural networks (ANN) methodologies were tested for analyzing the data. Wheat yield was correlated to soil available water holding capacity (SAWHC) in the upper 100 cm of the profiles (r2 = 0.39) and soil organic carbon (SOC) content (r2 = 0.26). The climate factor with stronger effect on yield was the rainfall/crop potential evapotranspiration ratio (R/CPET) during the fallow and vegetative crop growing cycle periods summed (r2 = 0.31). The phototermal quotient (PQ) during the pre-anthesis period had also a significant effect on yield (r2 = 0.05). A surface regression response model was developed that account for 64% of spatial and interannual yield variance, but this model could not perform a better yield prediction than the blind guess technique. An ANN was fitted to the data that accounted for 76% of yield variability. Comparing predicted versus observed yield a lower RMSE (P = 0.05) was obtained using the ANN than using the regression or the blind guess methods. Regional production estimations performed by the ANN showed a good agreement with observed data with a RMSE equivalent to 7% of the whole surveyed area production. As variables used for the ANN development may be available around 40-60 days before wheat harvest, the methodology may be used for wheat production forecasting in the Pampas. |
|---|