We are being bombarded with massive amounts of information. This is no different for businesses or organizations. The collection and analysis of this data is called Big Data. It has already been applied in finance, government, and healthcare. How can it change research and development within Environment and Water Management?
Big Data? Deep Learning?
The internet and faster technology has opened up a large stream of data from remote sensing, cameras, wireless sensor networks and more. The resultant large data sets were too difficult to deal with using traditional data processing software. Big Data is a method of extracting desired information from these large data sets. It uses predictive analytics or other advanced data analytics methods. The characterization of big data includes high volume, velocity, variety, and veracity. In other words, Big Data is quickly analyzing lots of different types of highly accurate data. Big Data is only possible now thanks to our current fast data processing capabilities.
Within Big Data, we have the concept of deep learning. Deep learning is essentially computers learning by example. It is most widely recognized as the software leading the development of self-driving cars. A computer is set to look at a series of pictures, words, or data sets. From there it can develop and perform tasks based on the data being presented to it. These tasks could be developed and performed with zero intervention by programmers.
Should it be used in Environmental and Water Management?
A review conducted in 2019 of over 1000 studies sought to “(1) examine the potential and benefits of data-driven research in EWM, (2) give a synopsis of key concepts and approaches in Big Data and ML, (3) provide a systematic review of current applications, and finally (4) discuss major issues and challenges, and recommend future research directions.”
The results of this review show that application of big data can and likely will change the way EWM experiments are run. Deep learning has already been successfully used for applications in remote sensing image classification, high-dimensional spatial and temporal data fusion and multisource data predictive analytics. The studies find that a digital revolution within EWM is necessary.
In the past, empirical or hypothesis-driven studies were done at the speed of a human brain. Now, with how fast these data can be processed and with the potential automation of data discovery through deep learning, computers can filter out, organize, and send only the applicable data to researchers. These techniques could be used for monitoring and understanding environmental changes, and to help improve predictions of short-term changes, like floods.
Actually applying Big Data and deep learning techniques to EWM still has a ways to go. The lack of quality within the existing data sets is a major issue. There is high cost to the existing Big Data software and this would make it unusable for most organizations. Once the cost can be significantly reduced this would remove a huge roadblock. The study also recognized the lack of a governing body to regulate the data and the lack of engagement from a wide variety of stakeholders. These data solutions must be designed by people who understand the algorithms and people who understand the problems and context these programs are attempting to solve. A collaboration between data scientist, domain experts, governments, the public, and private sectors is necessary to make these programs work well.