By: Zofie C.
Author’s note: As an avid environmentalist and a mathematics devotee, when I first came across genetic algorithms and their applications to environmental science, my interest was immediately piqued. I delved into my research knowing nothing, and tackled several hefty academic papers before reaching a basic level of understanding. This topic is complicated and still developing, and, as such, I am sure that I have made a few errors in my understanding of the topic and in airing what I have learned. Regardless, I hope that this serves as an interesting and comprehensible introduction to the fascinating topic that is genetic algorithms in environmental science.
In order to fully do justice to the subject of genetic algorithms (also referred to as GA) in environmental science, it is first necessary to address the general concept of genetic algorithms and how they work. Genetic algorithms were first invented by John Holland during the 1960s. They were later popularized by his student, David Goldberg, in the late 1980s (Haupt 2003), and are now used within a variety of subjects, including computer science, economics, engineering, bioinformatics, and, of course, environmental science, amongst a host of other fields. To some extent, genetic algorithms attempt to mirror the process of human evolution and optimization, taking a Darwinist stance in their application by acting in a manner that replicates the concept of “survival of the fittest” and mimics natural evolution (Jain 2017). Hence, the name choice of “genetic algorithm” is derived from the way in which the algorithm attempts to replicate the natural processes of evolution that are most commonly associated with genes.
The diagram below illustrates this concept, starting with the initial processes in which the population, which contains a set of individuals, is created. This then leads to the fitness assessment, thereby allowing for the selection of the “fittest” and crossover, or ‘mating’ between a pair of individuals at a crossover to create offspring. This finally results in mutation, in which some of the genes may mutate such that population diversity is preserved and early convergence of the population is prevented (Mallawaarachchi). The process then repeats itself until it meets the criteria in which it is necessary to stop.
Photo from: Neural Designer
Genetic algorithms are particularly useful because they yield better (read: more accurate) results than the results of more traditional techniques used for feature selection, can handle more complicated data sets, don’t require specific knowledge, and are easily applied to various subjects (Gomez et al.).
The application of genetic algorithms to problems of optimization are especially relevant in the field of environmental science. Mulligan and Brown “use a GA to estimate parameters to calibrate a water quality model. … They found that the GA works better than more traditional techniques plus noted the added advantage that the GA can provide information about the search space, enabling them to develop confidence regions and parameter correlations,” (Haupt). The flexibility of genetic algorithms allows them to be applied to a wide variety of circumstances that feature complex data sets, creating a model that is fitted to the data set instead of vice versa.
Genetic algorithms are particularly useful in the sub-areas of environmental science such as climatology, or the study of prevailing weather conditions and the resultant phenomena associated with them (Merriam-Webster), and water resource systems, which is a blanket term for the amalgamation of the study of the circulation, distribution, and management of water (commonly known as hydrology (Merriam Webster)) as well as “infrastructure, ecologic, and human processes involving water” (Brown et al.). One example of the practical application of genetic algorithms is found in their application to climatology, in which they can be used to model global temperature changes, as shown by the research of Karolina Stanislawska, Krzysztof Krawiec, and Zbigniew Kundzewicz. In their study, they concluded that genetic algorithms are “capable of inducing models that mimic the aggregate behavior of a very complex climate system ... In particular, [GA] allows [scientists] to find analytical models that bind the mean global temperature to climate factors, without resorting to historical temperature itself. These outcomes suggest that the approach used in this study allows making interpretations that are potentially useful in climatology,” (Stanislawska et al. 2012). In climatology, genetic algorithms are useful for studying the environment because of their unbiased nature and ability to handle complex data sets. In environmental science, the use (or misuse, depending on how you look at it) of our planet’s finite water resources is a concern of tremendous gravity and impact, especially as Earth’s population continues to increase and global temperatures rise. Genetic algorithms can be used to optimize the efficient use, preservation, and distribution of Earth’s water resources (Tyagi et al. 2019).
So many of the environmental issues we face as a society today are the result of a lack of optimization and planning, such as the pollutive results of urban sprawl and the inadequate use of clean energy. It would be interesting to see genetic algorithms applied to these environmental and social issues and others, including (but most certainly not limited to) waste management, food distribution, energy optimization, urban planning, and pollution. The use of genetic algorithms could help generate solutions to these environmental issues in an unbiased and scientific manner, helping to move the world towards a more sustainable future.
Further Reading:
Discussion Questions:
What are some environmental optimization issues that you can think of that genetic algorithms could be applied to ?
Genetic algorithms are complex and thus not necessarily easy for the average person to create and apply in everyday life. What are some ways that you can optimize your own lifestyle that have a positive environmental impact (i.e. taking shorter showers)?