Task 140

Earth Model-Human Model

Principal Investigator(s):

E. Kalnay


S. Motesharrei


R. Cahalan

Last Updated:

October 26, 2012 15:25:38

Description of Problem

Human population and consumption has grown significantly over the past few decades. The Earth’s natural resources were assumed to be practically infinite for the whole length of the history, but we are now realizing that they may be scarce. This has rung a bell for the policy makers, scientists, economists, and all other conscious individuals. Economic growth has reached an “uneconomic growth” phase. To cope with such issues, new fields of study like “ecological economics” are born. Several research groups around the globe have developed (mathematical) models to predict the future of human population and nature. Such models have helped scholars to understand and investigate possible scenarios for the future of life on our planet more thoroughly.

The most complete versions of such models incorporate population, climate, energy, and agriculture as main variables. However, some of these variables, like population, are taken as exogenous variables and therefore, the coupling between the variables is uni-directional. This means that, for example, increased population can affect climate by creating more pollution, but the climate change does not feed back on the population.

Scientific Objectives and Approach

We are developing a prototype Earth System Model (ESM) coupled with a Human model where all of the above-mentioned variables are fully coupled to each other. As a result, we do not need to import the values for any of the variables from external sources, e.g. United Nations population projections for the next 50 years. Instead, the model internally determines any of the variables, say population, in the coming decades based on the dynamical relations with the rest of the variables. When the prototype model is ready we will work with the Climate@Home team to post the model and allow many members of the public to test the impact of the model assumptions and parameters on the model behavior. Calibration of models is a fundamental issue, and we plan to apply the powerful tool of Local Ensemble Transform Kalman Filter (LETKF) to calibrate the parameters of the model to the past 40 years or so of observed data.


The project has just started. At this time, we have succeeded in developing a basic toy version of our model, with two variables, population and nature. Nature represents an aggregate variable which implicitly includes climate, agriculture, and energy, i.e., resources for human beings. Through this basic model, we have successfully observed several types of scenarios of periods of growth and decay, overshoot and collapse, and convergence to a steady-state, with a much richer behavior than the popular predator-prey model. We have coupled the Lorenz 1963 chaotic model (as well as the predator-prey model) to the Local Ensemble Transform Kalman Filter (LETKF) and assimilated noisy observations. These simulations show that the model parameters are much better estimated by the LETKF than by the standard observations/parameters regressions.