Interannual variability of soybean yield in the Argentine Pampas and its relationship with synoptic weather types in Southern South America

The Argentine Pampas Region spreads over 600000 square km and is one of the largest food producers in the world (cereals, crops, and livestock) being the soybean one of the most important crops. The beginning of the soybean crop cultivation in Argentina was in the early 1960s with ten thousand sown hectares. Nowadays, over 12 million hectares are cultivated, contributing in a high percentage to the global trade of soybean and its by-products.

In the region chosen it is often possible to find extreme climate situations, which are persistent in some cases, and which have a significant socio-economic impact. Agriculture is mainly developed without artificial irrigation; therefore, rainfall and its spatial-temporal variability are crucial for successful harvests. Local and daily scale variables depend on the larger scale atmospheric fields, among other conditions. Consequently, an objective classification of daily circulation fields is proposed using a long database in order to relate these weather types to soybean yield variability in the Argentine Humid Pampas.

In order to conduct this study, two datasets were used: the first one it is related to the crop (yield) and the second one to the atmospheric circulation (1000 and 500 hPa geopotential height). Soybean yield series of 58 provincial departments supplied by Argentina's Secretaría de Agricultura, Ganadería, Pesca y Alimentación in the period 1973-1999 were used. These provincial departments are located in the region of interest. This work focuses on yield (estimated as the ratio of total production to area harvested) as an indicator of a crop's vulnerability to climate variability. Previous studies revealed that soybean yield shows spatial coherence at the local scale. For this reason, the index defined as the number of provincial departments with negative soybean yield anomaly (NYI) was considered in order to characterize each soybean campaign in a regional way.

Daily average fields of 1000 and 500 hPa geopotential height in the period 1979-2001 corresponding to the NOAA-CIRES Climate Diagnostics Center NCEP_Reanalysis 2 data provided by the NOAA/OAR/ESRL PSD from their Web site at http://www.cdc.noaa.gov/ were also used. The domain chosen extends from 15° S to 60°S latitude and from 30°W to 90°W longitude and includes 475 grid points (2.5° latitude by 2.5° longitude). This domain spreads over both the Pacific Ocean and the Atlantic Ocean, and the Andes Mountains, all of which have a significant influence on the atmospheric circulation over South America. In order to filter the seasonality of daily fields, the daily average field in the period analyzed was subtracted from each day.

The resulting daily anomaly fields of geopotential heights were used to conduct this study. Principal Component Analysis, in T-mode with the correlation matrix as input, was performed coupled with Cluster Analysis to determine the atmospheric patterns (‘synoptic weather types'- WT). The ‘k-means' partitioning method with Euclidean distance was used. This classification was made during the growing season of the soybean (October-May), which coincides with the time of highest rainfalls in the region. A 7 cluster solution for the 1000 hPa level was found and a total of 5 clusters was considered for the 500 hPa level. The series of seasonal and monthly frequencies were calculated for each WT in order to relate them to the soybean yield index (NYI).

The most outstanding features are that the structures, corresponding to a positive anomaly over the east of the continent and adjacent ocean generating subsidence and warm advection across the layer (facilitating dry conditions over the Humid Pampas) do not benefit soybean yield. This negative effect is manifested when these WT occur mainly in the flowering, pod set and pod filling stages (January and February).

The same negative effect on soybean yield was found for weather types that could be related to intense precipitation events in the Pampas when they have a high frequency of occurrence during the harvest (April and May). These patterns corresponds to: structures of 1000 hPa geopotential height anomalies that present a cyclonic anomaly with axis in the east of the continent over the Atlantic Ocean that affects almost the whole country, probably associated with cold front passages or cyclogenesis; and structures that show negative anomalies at the southwest extreme of the continent intensifying the westerlies south of 40°S. At the upper level, generalized low soybean yield is mainly related to structures that show negative anomalies to the south of the continent favoring the flow from the west over Patagonia. At the same time, a positive anomaly on the Atlantic Ocean induces an anomalous flow from the northeast at lower latitudes.

Furthermore, WT whose anomalies configuration favor cold air incursion, show a negative effect during the sowing stage (November). This negative effect reverts in the rest of the growing season, mainly in January and February. These atmospheric patterns provide properties that reduce thermal stress conditions if they occur during the flowering and pod set stages, while they contribute with low temperatures if they occur during seed germination and early vegetative stages (growth stages sensitive to low temperatures).

In general terms, the soybean yield can be considered as the final result of an addition of different complex processes throughout the growing cycle. In this study, a characterization of the atmospheric circulation in relation to the soybean yield variability was carried out. The high or low occurrence, recurrence and persistence of particular atmospheric structures can condition the success of a campaign. Therefore, the study and monitoring of these weather patterns and their relationship to surface variables constitute a primary element for farming producers and decision makers to plan strategies.