Tropical weather can be capricious, torrid one minute and drenching the next. Such mercurial meteorology poses challenges to farmers, who need to predict soil moisture to plan irrigation. In a new study, scientists developed a tool for Indian farmers in the region of Nashik to provide weather forecasts and irrigation suggestions at the scale of a single farm. The authors said it could be used across the region’s 6,000 hectares of farmland—and potentially beyond.
“It is an excellent initiative, and the objective to help farmers manage farm-level operations better is no doubt a good one,” said Madhavan Rajeevan, a meteorologist and former secretary of the Ministry of Earth Sciences in India who was not involved in the study.
“Decision Tools in an Uncertain Environment”
The 4-year study encompassed two entire growing seasons, including two monsoon seasons (called kharif) and two winter seasons (called rabi).
The scientists enlisted the help of 10 grape farmers in the district of Nashik who had soil moisture sensors. Using data from just two of the 10 available sensors and estimates from satellites, they quantified current levels of soil moisture. The decision to use a small subset of available data demonstrated that the method could be successful despite limited field observations.
Then, researchers gathered weather-related data such as rainfall, temperature, humidity, and wind from the India Meteorological Department’s hindcasts and forecasts. Weather forecasts typically reach 10-kilometer scales at best, so they integrated these data into a machine learning model that yielded small-scale predictions useful to farmers. The new method forecasted rainfall 1–3 weeks in advance and at the scale of individual farms.
Given current soil moisture and predicted rainfall, the study’s authors then developed a tool that translated forecasts into irrigation decisions. The tool saved water compared with the farmers’ traditional approach, which relies on personal notes of past rainfall, daily weather conditions, and how dry the soil looks. The new forecasting tool reduced water usage by 20%–45% during the monsoon season and by 17%–35% during the winter season.
“The real contribution [of the study] is developing decision tools in an uncertain environment,” said Subimal Ghosh, a coauthor and a hydroclimatologist at the Indian Institute of Technology Bombay. “The idea is to provide a simple tool to help farmers, especially poorer ones who cannot afford [soil moisture] sensors, decide how much water they should use.”
Still, Rajeevan was skeptical. For starters, using large weather circulation patterns to make farm-level predictions introduces inaccuracies. In addition, it’s not clear how easy it will be to scale up this small case study.
Ghosh, though, remained resolute. “No model predictions are perfect,” and finer resolutions always add errors and uncertainties, he said.
Weather Predictions with Climate Change
Early-warning systems like the new forecasting system could gain prominence as the climate changes. For example, the World Meteorological Organization recently announced a global initiative to build and scale early-warning systems, especially in vulnerable regions of Africa, Central and South America, South Asia, and small island states.
The new study is just one step in that direction. The next, the authors said, is to develop an app that broadcasts climate advisories in local languages. “This is very important for Global South countries with resource constraints,” said coauthor Raghu Murtugudde, a climate scientist at the Indian Institute of Technology Bombay. “Adaptation is about learning how to manage the unavoidable.”
—Rishika Pardikar (@rishpardikar), Science Writer