Can paddy yield be predicted two months before harvest?

A new research finds this possible.

A groundbreaking study combining advanced remote sensing technology with machine learning shows promising results for predicting paddy yield two months before harvest.

The study uses solar-induced chlorophyll fluorescence (SIF), which directly measures plants' photosynthetic activity, offering real-time insights into their health and stress levels.

This differs from traditional methods, which rely on canopy characteristics like greenness and often miss important physiological changes in plants.

“SIF gives us a clearer picture of crop potential, especially under stressful environmental conditions,” said Manoj Lamichhane, a PhD student at South Dakota State University in the US, one of the study's authors.

The study presents a robust data-driven approach to predicting rice yield in Nepal by integrating SIF with traditional remote sensing and environmental data. A key strength of the research is utilising a multi-source dataset, which includes SIF, and biophysical, climatic and soil variables.

The study provides valuable insights into paddy yield prediction by exploring various time windows in machine learning models for growing periods of these variables.

Analysis revealed that soil variables, mainly sand and P2O5, are the most critical factors in predicting paddy yield. Climatic properties, vegetation indices, and SIF variables such as precipitation and water deficit follow them.

According to the research, integrating SIF with traditional remote sensing and environmental variables reduces the prediction error by 21 percent.

“When tested on unseen data, our model accurately predicted rice yield across all districts in Nepal, aligning closely with observed data. This fitted model can predict paddy yield using features from July to September,” according to the study.

“Given that the rice harvest in Nepal occurs in November, yield predictions can be made as early as two months before harvest, providing valuable lead time for decision-making.”

The study said that future research could explore applying this approach in other major paddy-producing countries across Southeast Asia to evaluate its effectiveness and adaptability to different regional conditions.

Experts say new research conducted in the Tarai offers hope for improving agricultural productivity and food security in Nepal, where rice plays a vital role in the diet, rural income, and economic growth.

The plains account for 71 percent of Nepal’s total paddy acreage, the hills for 25 percent, and the mountain region for 4 percent.

Paddy is Nepal’s biggest earning farm commodity, with tens of thousands of farmers relying on its income. According to economists, the high output may cool down inflation and pump up the economy.

Paddy production in Nepal, particularly in the Tarai belt, faces increasing challenges due to climate change and unpredictable weather patterns, which can severely impact yields.

Using machine learning models, researchers say they can predict rice yields based on various environmental factors, such as rainfall, temperature, and soil conditions.

The study's findings underscore the importance of adequate rainfall during August and September.

“Our research shows that rainfall during these critical months directly influences yields. This means that efficient water management during this period is crucial,” said Dr Sushant Mehan, assistant professor at South Dakota State University.

Soil health also emerged as a key factor in the study. The research highlights the importance of maintaining phosphorus levels, a nutrient essential for rice growth.

“Farmers should regularly test their soil and apply the necessary fertilisers, particularly phosphorus-based ones,” added Mehan, emphasising the need for sustainable soil management practices.

This research has significant implications not only for farmers but also for policymakers.

By providing early yield predictions, the study enables better planning for food security and supply chains, helping to avert potential shortages.

Rajendra Uprety, a rice expert, told the Post that early yield predictions could primarily benefit a country like Nepal.

For instance, according to Uprety, farmers will be encouraged to insure their paddy based on early yield predictions. “Besides, if farmers know that they are getting a good yield two months before the harvest, they can increase investment in paddy. The investment will enable farmers to adopt commercial farming from the subsistence farming method.”

Since paddy output is directly linked to the country’s economic growth, the government and policymakers can accurately forecast the country's economic growth, including inflation. “Low output means it will put pressure on the price, and accordingly, the government can plan for imports,” said Uprety. “Thus, prices will remain stable.”

He said that neighbouring India has recently lifted the ban on the export of rice following a projection that it will have a bumper rice harvest this year.

On September 28, the Indian government lifted the non-basmati white rice export ban. This change follows a prohibition to enhance domestic supply since July 20, 2023.

“Early prediction will enable Nepal, just like India, to know what Nepal’s requirement is in case of low production,” said Uprety.

“These early predictions allow farmers and the government to prepare in advance, ensuring that enough rice is available even during years of low production,” said Sagar Kafle, assistant professor at Tribhuvan University, one of the study's authors.

“As climate change continues to impact agriculture in Nepal, this study offers a roadmap for increasing resilience in rice production. The findings could play a crucial role in helping Nepal maintain stable paddy yields, securing both livelihoods and food supplies for the country’s population.”

Despite the El Niño phenomenon, Nepali farmers harvested a record 5.72 million tonnes of paddy in the last fiscal year, ended mid-July.

Paddy production grew 4.33 percent year-on-year. The country has 1.43 million hectares of land suitable for growing paddy. The grain is generally planted twice a year—in June and March.

However, spring paddy is planted only on 104,712 hectares. According to the agriculture ministry, paddy's contribution to the country’s gross domestic product amounts to 4 percent.