Using Earth Science Tools to Improve Seasonal Climate Prediction for Agriculture

Cynthia Rosenzweig, Walter Baethgen, Antonio Busalacchi, Mark A. Cane, David Rind, and Compton J. Tucker

El Niño events are responsible for dramatic changes in rainfall in many regions and consequently for large effects on agricultural productivity. Droughts and floods are often products of these “teleconnections” between El Niño events and regional climate. The economic and social consequences of these events can be devastating. In 1988, a La Niña year, drought caused agricultural losses totaling $39 million in the Midwestern United States. In 1991 and 1992, an El Niño event caused another severe drought, this time in southern Africa. NASA’s Earth Science Enterprise is developing techniques for using satellite observations to improve climate forecasts in the NASA Seasonal-to-Interannual Prediction Project (NSIPP) and investigating ways to utilize these forecasts to enable governments and agencies to plan for their consequences on agricultural production and on societies.
Climate fluctuations, such as droughts and floods, affect agricultural regions around the world. These fluctuations cause yields, production, and farm incomes to vary. In many regions, a major component of climate fluctuation is the El Niño Southern Oscillation (ENSO), a large-scale interaction between the Pacific Ocean and atmosphere occurring every 2 to 9 years. Recent advances in satellite remote sensing, ocean observations, and Earth science modeling offer the possibility of forecasting these ENSO-related regional seasonal events. Such predictive models have the potential to improve agricultural planning. The ability to warn farmers, food security specialists, and regional planners in advance of droughts and floods could allow timely action to mitigate negative ENSO impacts and to obtain benefits from positive ones.
To help enhance these opportunities, the NASA Earth Observing System (EOS) Interdisciplinary Science CAFÉ Project (Climate Variability, Anthropogenic Forcings, and Agricultural/Marine Ecosystem Interactions) is developing a coordinated framework for linking observational and modeling tools used to predict ENSO events, their climatic effects, and their impacts on people and agriculture around the world. The interdisciplinary team consists of oceanographers, climate modelers, marine biologists, and agro-ecologists. This project is analyzing regional case studies to serve as prototypes for the characterization of ENSO effects on agriculture. This article concentrates on the use of these integrated observations and models in Latin America, a region where agriculture is strongly affected by ENSO events (Figure 1).

ENSO Prediction and Monitoring
The first model used to predict ENSO events is the Cane-Zebiak model. This model depicts the fluctuations of temperature and sea level in the tropical Pacific Ocean and the overlying atmosphere. The Cane-Zebiak model is a dynamic simulation tool that relies on governing physical equations rather than on statistical relationships, thus providing the means to interpret and understand the ocean-atmosphere processes it simulates. For use in the predictive mode, the Cane-Zebiak model initially relied solely on observations of surface winds over the ocean made from merchant ships. These winds are coupled to an ocean component that generates currents, depths of thermocline (the dividing layer within the ocean between the upper layer of warm water and deeper cold water), and temperatures.
Satellites now routinely monitor sea-surface temperatures (SSTs), sea levels, and winds. These data are used to improve the performance of ENSO prediction models. For example, the TOPEX/Poseidon joint mission to measure sea level height and the QuikSCAT mission to measure sea winds help the Cane-Zebiak model simulate sea level heights and wind velocities. The Cane-Zebiak model generates ENSO predictions up to 1 year before an event.
NSIPP attempts to improve upon the Cane intermediate model forecasts by using comprehensive coupled models of the ocean/atmosphere/land surface for 1-year predictions. Coupled models such as this are necessary to predict the impacts of ENSO events over continental regions. The Goddard Institute for Space Studies (GISS) also uses coupled models for their agricultural forecasts (Figure 2).

Climate Modeling of ENSO Teleconnections
Climate modeling of ENSO teleconnections for agriculture is carried out with global and regional models. Global climate models (GCMs) calculate how El Niño and La Niña events change large-scale atmospheric dynamics, while finer-resolution regional models provide more detailed localized projections.

Global Scale
GCMs are comprehensive mathematical formulations that calculate the movement over time of heat, moisture, and momentum throughout the Earth’s atmosphere and its surface, including the continents and oceans. GCMs make these calculations through solving, sequentially or simultaneously, equations that represent radiation, turbulent transfers at the ground/atmosphere boundary, cloud formations, condensation and precipitation of moisture, and transport of heat by ocean currents and by winds in the atmosphere. The equations are solved for a number of vertical layers in the atmosphere and for grid points in finite difference models at the surface of the earth. The spacing between grid points ranges from less than 100 km to 500 km.
To use the GISS GCM for studying the climate teleconnections associated with ENSO, climate modelers start with observed monthly SSTs from El Niño and La Niña years as the driving force for atmospheric changes. Climate results from GCM simulations with the observed SSTs are compared with the observed climate to test how well the model is performing. These global models generally do a good job of simulating large-scale climate dynamics that influence regional climate patterns (Figure 3).

Regional Scale
Results from the GISS GCM are used to drive a regional-scale model (RCM) to provide more detailed characterizations of ENSO events for a given agricultural region. RCMs are finer resolution climate models able to calculate processes such as convection at scales closer to reality. The RCMs also solve the fundamental equations more realistically; for example, storms move faster through the RCM. The GISS/Center for Climate Systems Research (CCSR) RCM runs at a resolution approximately 50 x 50 km (Figure 4).

Agricultural Users
Seasonal climate predictions related to ENSO can contribute to planning by a variety of agricultural decision makers, including individual farms, agricultural industries, and regional and national agencies. While ENSO events occur cyclically but irregularly, agriculture follows a more predictable annual cycle of crop management decisions (Figure 5 and Table 1).
Opposite parts of the ENSO cycle can cause important constraints for farmers. For instance, El Niño events tend to bring droughts to northeast Brazil but tend to bring rains to Uruguay. La Niña events bring drought to Uruguay and more rain to northeast Brazil.
Dynamic crop models are used with ENSO and climate model simulations to help regional agricultural planners prepare for upcoming growing seasons. The Crop Environment Resource Synthesis-Maize (CERES-Maize) predictive model for corn was used for Nordeste, Brazil. The simulated corn yields using the observed weather dataset from the region compare well to observed yields for that region. The model correctly simulated low yields in 1993 due to low rainfall associated with El Niño conditions. In Uruguay, using CERES-Maize with RCM results captures La Niña drought effects, allowing such adaptation strategies as short-season crops to be tested (Figure 6).

MODIS
The project team is working closely with researchers from the Instituto National de Investigacion Agropecuaria of the Ministry of Agriculture in Uruguay to develop new technologies and knowledge based on EOS products for decision-making in agricultural production. With the MODIS remote sensor, the combination of high spatial and spectral resolution and a high signal-to-noise ratio offers the potential for improved understanding of climate/crop relationships. Use of the MODIS data in conjunction with agricultural land-use maps and soil types differentiates agricultural land cover from the generalized vegetation signal. These methods are helping to make accurate within-season estimates of agricultural plant conditions and their relationship to climate variability, especially as manifested by the ENSO, at national and regional scales (Figure 7).

Summary
Earth science observational and modeling tools are improving seasonal climate predictions for agriculture. However, acknowledging the limitations of forecasts is essential, especially when human livelihoods are at stake. By working directly with researchers in regions affected by ENSO, we hope to develop appropriate utilization modes of these advanced tools.

Acknowledgments
NASA EOS IDS Project No. 291-01-08-00 provides funding for this research. Special thanks to Leonard Druyan, Radley Horton, Spiro Papanikolau, Richard Goldberg, and Daniel Hillel for their contributions.

About the Authors
Cynthia Rosenzweig is a Research Scientist at NASA/Goddard Institute for Space Studies and a Senior Research Scientist at Columbia University.
Walter Baethgen is a Senior Scientist in the Research and Development Division of the International Soil Fertility and Agricultural Development Center.
Antonio Busalacchi is the Director of the Earth System Science Interdisciplinary Center (ESSIC), and Professor in the Department of Meteorology, at the University of Maryland. ESSIC is a joint center among the Departments of Meteorology, Geology, and Geography at the University of Maryland in collaboration with the Earth Sciences Directorate at NASA’s Goddard Space Flight Center.
Mark A. Cane, G. Unger Vetlesen Professor of Earth and Climate Sciences in the Department of Earth and Environmental Sciences and the Department of Applied Physics and Applied Mathematics at Columbia University, works at the Lamont-Doherty Earth Observatory of Columbia University.
David Rind is a Senior Research Scientist at the NASA/Goddard Institute for Space Studies and a Professor in the Department of Earth and Environmental Sciences, Columbia University.
Compton J. Tucker is a Senior Earth Scientist at the Laboratory for Terrestrial Physics, NASA/Goddard Space Flight Center.

Back