top of page
  • Zakariyya Tewari and Tehniyat Ikram

Tackling climate change with big data

Updated: Jan 25


With rising geopolitical tensions during COP27 negotiations, an unstable energy landscape and waves of never-ending climate extremes, 2022 has been a defining year for attempting to save a planet on the brink of collapse. As a result, many organisations have risen to the challenge in an attempt to apply their technological expertise. We take a deep dive into the SERVIR Global initiative, discussing the use of Big Data and Artificial Intelligence to mitigate the effects of climate change.

It is no secret that tackling climate change has become one of the toughest global problems of our time, posing significant threats and a forced reapproach towards the way we utilise fuels and energy. The effects, whilst differing in levels of severity, are apparent everywhere. In the UK this year alone, we have experienced intense wind (e.g. Storm Eunice), long spells of hot weather and the more recent torrential rains and unusually high temperatures during the month of October. However, this pales in comparison to countries such as Pakistan and Bangladesh, where humid climates and mass flooding have become increasingly frequent and alarmingly destructive.

Fortunately, the rapid-advancement of technology, particularly in regards to Big Data and Artificial Intelligence (AI), has innovated an array of unique tools and techniques that present a promising approach to combat these climate challenges.

Originally founded in 2005, SERVIR GLOBAL is a joint initiative of NASA, USAID (United States Agency for International Development) and other leading geospatial institutions in Asia, Africa and Latin America. Through leveraging massive satellite and geospatial datasets, SERVIR focuses on providing developing regions the key insights and forecasts to address the severe climate crises they are faced with. Today, SERVIR has formed a global network of regional hubs and has developed over 30 services, partnering with over 500 institutions. We take a deeper look into two of their widely-used solutions: ClimateSERV and HIWAT.

ClimateSERV is a web application that allows for the visualisation and analysis of satellite data to derive agricultural metrics and time series forecasting. The major datasets used include Climate Hazards group InfraRed Precipitation with Stations (CHIRPS), which consists of over 30 years of satellite-derived rainfall data, North American Multi-Model Ensemble (NMME), a collection of 180-day seasonal forecasts, and MODIS-derived Normalised Difference Vegetation Index (eMODIS NDVI) and a compilation of vegetation growth data from the last 15 years. With these data-rich sources, researchers and scientists are able to conduct time series analyses and forecasting to guide agricultural growth and predict the likelihood of drought and severe weather conditions. ClimateSERV has played a key role in several SERVIR projects, including flood forecasting in Cambodia and groundwater analysis in Niger for agricultural production.

Observation of average precipitation from the CHIRPS data using ClimateSERV.

One of the most dangerous weather events affecting Bangladesh are known as Kalbaishakhi or Nor’westers - small storms bringing gale winds and torrential rain. Without the necessary weather forecasting technology, it can be close to impossible to predict the occurrence of these storms, leading to catastrophic consequences. SERVIR Global recently launched the High Impact Weather Assessment Tool (HIWAT) which utilises NASA satellite data with observations from the Bangladesh Meteorological Department (BMD). Until recently, Bangladesh lacked both the data required to provide accurate weather forecasts and the funding that would bring observation satellites and powerful computers to analyse this data. However, with the HIWAT technology provided by SERVIR Global, the BMD is able to produce weather predictions and, as a result, provide the government with a strategic edge when responding to climate emergencies. With a large proportion of Bangladesh’s population being dependent on agriculture, HIWAT ensures the planning of appropriate protocols in response to flooding. From 13:00 (GMT) everyday, HIWAT runs a 54-hour probabilistic forecast for improved decision making, with toggles for lightning, hail threat and supercell threat organised by district. The government is then able to plan short-term flood responses such as food aid, water purification tablets and free seeds given to farmers.

With rapid advancements in the practical applications of Machine Learning (ML) and AI, the future only becomes more exhilarating. The potential of the CHIRPS data has been illustrated through an experiment for short-term streamflow forecasting using both linear (ARIMA, SARIMA) and nonlinear (Random Forest, CHAID) ML models, resulting in a 25% increase in prediction accuracy as compared to previous methods. Furthermore, while ML has shown to be a powerful solution for predictions and mitigating the impacts of climate change, there is also potential for its use to tackle the cause. The emergence of autonomous vehicles and AI-enabled transport can play a key role in reducing greenhouse emissions, and recent investigations by DeepMind have highlighted the benefit of incorporating ML programs into renewable energy sources. However, with these exciting new innovations and technologies comes a major price, which is the fact that these practical ML models may involve the processing of millions of rows of data, and as such can generate dangerously high carbon emissions if done without regulation. Therefore it is imperative to remember the full picture, and assess all the impacts AI may bring for a given situation before any implementation.

As the intensity of the effects of climate change increase, breakthroughs in AI only become more significant. The utilisation of Big Data and ML techniques have the potential to tackle both the dangerous causes and destructive consequences that climate change brings, but we require the expertise and co-operation of organisations such as SERVIR to leverage these tools in the essential areas.

bottom of page