ASSESSING THE IMPACT OF CLIMATIC VARIABILITY ON WHEAT YIELDS BY USING CROP SIMULATION MODELS AND WEATHER GENERATORS CONDITIONED ON EL NINO PHASES. A CASE STUDY FOR CHILE'S AGRICULTURAL CENTRAL VALLEY

摘要

Agriculture is one of the human activities more susceptible to climatic variability. Even with advances in the fields of plant breeding, soil fertility, weed science, among others, uncertainty in the final outcome remains present. Daily meteorological variables are the most important source of temporal variability for crop growth and development. Among them temperature and precipitation have been identified as the major driven forces, being included in many studies to investigate the main effects of climate variability on agriculture by using crop simulation models. (Riha et al, 1996; Wheeler et al, 2000). With the current level of understanding of climatic variability, especially in those cases where a subset of events within the regional climate can be identified, a good opportunity to explore its impacts over agricultural systems has appeared. This is the case of the El Nino-Southern Oscillation phenomenon where a set of mutually exclusive and collectively exhaustive events has been defined. If a particular region shows an ENSO climatic footprint meteorological variables classified into its different phases (i.e. El Nino, La Nina or Normal) can be used to forecast crop yields and to study the main components of their temporal variability. Studies like those mentioned above require the inclusion of daily weather information as the main input data. Such records are not always easily available and/or their lengths are insufficient to allow the study of climate variability effects on agricultural crops. To solve this problem stochastic synthetic series can be generated by using random number generators whose outputs have the property of reproducing the main statistical characteristics of the series from which their parameters were fit (Wilks, 1999). Those algorithms are commonly known as Weather Generators when used for Monte-Carlo simulation and introduced into the crop-modeling scheme by Richardson (1981). In Chile, precipitation is identified as one of the most important meteorological variables for agricultural productivity not only because it represents the direct supply of water for rain fed agriculture but also because its amount and distribution on winter season determines the amount of water available for irrigation when snow melts during spring and summer. Although simple empirical relationships can be built using monthly or seasonal total precipitation as a predictor variable, a more comprehensive study on the effects of climate variability on agricultural crops must include all the meteorological variables that determine biomass accumulation. Rainfall anomalies in central Chile have been investigated and associated to ocean and atmospheric phenomena like Southern Oscillation Index (SOI) and El Nifio events, concluding that anomalously dry conditions are found during positive SO phase (Rubin, 1955; Pittock, 1980a) and exceptionally abundant during El Nino years (Quinn and Neal, 1982; Kane, 1999). Chile's agricultural central valley is mainly located between latitudes 29 and 40 south. It is a complex agricultural system where irrigated and rain fed land are used to satisfy the internal demand for agricultural products as well as to produce highly valuable crops that are exported, representing one of the most important sources of income for the country. The purpose of this study is to characterize the variability of wheat yields as well as the main hydrological features of Chile's central Valley, by using daily weather generators conditioned on El Nino phases and crop simulation models. This study is a first step towards exploration of the value of climate forecasts and the development of management strategies to take advantage of them.