Famed for its tea, Assam’s tea garden records are giving scientists a glimpse into past rainfall changes in northeast India, plagued by scattered and spotty historical rainfall data.

For 12 years, the scientists and students at Cotton University in Guwahati, Assam, combed through handwritten tea garden records spanning 1920 to 2009, to reconstruct a 90-year daily rainfall dataset.

The reconstructed dataset aims to fill in the gaps in historical rainfall observation network by the India Meteorological Department and generate a new baseline for rainfall in the northeast. This will help analyse and address the impacts and risks due to extreme rainfall, in improved ways.

Using the data mined, they were able to identify and highlight the increasing frequency and intensity of extreme rainfall in northeast India through the past century, while the average rainfall has decreased.

“The rain comes in short spells; the spells are no longer well distributed in space and time like they used to be 30-40 years ago,” Rahul Mahanta, who led the reconstruction, told Mongabay-India during a visit to the university just days before the Assam floods disrupted life, in June 2022.

Surrounded by registers from tea gardens with yellowing pages, sprinkled generously with dates and numbers in spindly handwriting, Mahanta narrates the story behind the reconstruction.

Historical rainfall data

“While investigating extremes of rainfall [flood and drought] in northeast India, we observed that continuous rainfall data by the India Meteorological Department was only available for 15 stations for 32 years, starting from 1975. But before 1975, the number of stations with rainfall measurement data in a given year decreased, and stations with continuous measurements are even fewer,” Mahanta said.

“To get a true picture of extremes, we need long-term datasets. So, we set about to reconstruct the historical rainfall database using tea garden records. We collected records from private tea gardens and sourced British-administered IMD data collected from British-owned tea gardens in Assam. By the time we collected the data and digitised them, it was 12 years. We had accessed about 547 stations across tea gardens and other documentary sources such as records kept by Jesuit missionaries, by the time we got done,” added Mahanta, who worked with his students to keep the data mining effort going without any funding.

The reconstruction builds on continuous rainfall data obtained from 24 India Meteorological Department stations from 1920 to 2010 including stations in Assam, Manipur, Tripura, Meghalaya, Nagaland, and Arunachal Pradesh at elevations ranging from 16 metres to more than 2,000 metres above sea level.

The group had to put up with yellowing, disintegrating pages and deciphered bad handwriting; but that was the least of their worries. “We had to also standardise the data and correct errors which was a time-consuming exercise,” added Mahanta.

Handwritten tea garden records. Credit: Sahana Ghosh/Mongabay.

“Organised meteorological records were facilitated by the British with the establishment of the IMD in 1875. Institutions such as tea gardens retain archives with handwritten records going back to the late 1870s. Because tea and timber production were the main drivers of the colonial economy, and because most of the cultivable land was allocated for them, tea plantation records make up a significant part of northeast India’s documented heritage. Of almost 750 tea estates in Assam, more than 100 have been in existence for over a century and have daily temperature and rainfall records,” according to a paper by Rahul Mahanta.

Private diaries, periodicals, and journals, and logbooks of medical and scientific research found in missionary hospitals in the region, Jesuit libraries, newspapers, and personal diaries are also rich repositories of temperature and rainfall data.

Analysing the 90-year-long data also reveals that the variability in rainfall observed is part of some global-scale natural forcings, not anthropogenic. “If there are some [anthropogenic forcings] we need more sophisticated methods to decipher them,” he added.

Currently, Mahanta and his team are looking at the wet and dry spells using the reconstructed dataset (paper under review) and have found changing patterns in the monsoon season since 1970: the intensity of extremely wet spells and the number of extremely dry spells during the monsoon season have both been increasing in recent decades, which in turn increase the risk of both drought and flood in the region.

“Rainfall extremes during the months of the monsoon season can be as important as how much total water is received,” says Mahanta. For example, this year during critical crop growth stages, too many days without rain reduced yields in some districts and led to crop failure in some others, which impacts the region’s agriculture-dependent economy. “At the same time, short periods of very heavy rainfall like the ones we witnessed in April, May, and June this year, created humanitarian disasters, when massive flooding killed hundreds of people in the state,” he said.

Spotty records

Water resources expert Manabendra Saharia, who was not associated with the data mining effort says that an immediate application that could work is preparing new precipitation products that will leverage this hitherto unused source of historical precipitation data.

“Rainfall varies from one place to another substantially. So, the denser the network of stations, the better your understanding of rainfall patterns over an area. A precipitation product essentially is a collection of data from various gauge stations in a single interoperable format. Since IMD has had limited locations for collecting data in northeast India historically, these tea garden stations act as valuable resources for reconstructing climate history,” Saharia at IIT-Delhi added.

Indian meteorologist BN Goswami, who was also leading the reconstruction research, suggests that the India Meteorological Department should add the 90-year-data daily data to their repository.

“We need long-term data to understand the rainfall variability. In recent years, the data gathering has improved but long-term data for 100 years in northeast is only available for 12 stations in the region. This is not sufficient because the region has a lot of variabilities. IMD needs to try and improve this availability of long-term data for northeast India across all its observatories and stations,” said Goswami, a former director of the Indian Institute of Tropical Meteorology.

Cherapunji in Meghalaya. Credit: Nazmul Ahmed, CC BY-SA 4.0, via Wikimedia Commons.

“Before the 1950s, IMD had more than 100 rain-gauge sites across northeast India. They were very efficient. Even in Cherrapunji, Meghalaya, they had five rain gauges before 1940. But after 1950 there is a drop in the records,” says Mahanta.

The drop in rainfall records corresponds to war, social movements, and disasters such as the Great Assam Earthquake in 1950, the 1962 India-China War, the language movement and Bangladesh Liberation War in the 1970s, the 1980s Assam foreigner’s movement followed by other ethnic tensions.

“Cherrapunji has a lot of data gaps which is why it is an outlier. If you miss one year that will completely mess up your understanding of the inter-annual variability of rainfall and extreme events. We should not use Cherrapunji in long-term variability studies because there are large data gaps. Then only we can understand the climatic drivers,” said Goswami in a webinar organised last year by the India Meteorological Department. He also emphasised redefining the rainy season in northeast India.

Augmenting weather services

“Climatological rain is more in April in the northeast than June in central India. So, the rainy season starts in April. Even if we leave out April, we get substantial rain in May compared to June in central India. We have to include May in the rainy season for northeast India. We have to define the rainy season from May to September,” he reiterated to Mongabay-India in an interview. Goswami says improving modeling skills is also crucial to augmenting weather forecast services.

Arup Kumar Sarma who works on water resources in the Brahmaputra river basin applied a rainwater harvesting system (Sustainable Approach of Rainwater Management and Application) to conserve water in tea gardens based on information collected from tea garden owners.

“Waterlogging due to short-duration, high-intensity rainfall is affecting tea gardens in the monsoon. We found that there was no way to let the water pass because there were other cultivation areas or roads. So, we designed a water harvesting system in a way to let the water go into the ground so it addresses the drainage issue and also the drought during the winter season,” Sarma at the Indian Institute of Technology, Guwahati, said.

A tea garden worker at the Aideobarie Tea Estate in Jorhat in Assam. Credit: Reuters.

“We have a two-chamber system where a part of the water goes underground to recharge the aquifer. The water that remains in the upper chamber [the pond portion] can be supplied for the lean season. We used historical rainfall data from the tea garden to get an idea of the maximum precipitation. We used this data to feed into the hydrological model to understand how much water may accumulate and the rate of flow. This helped us design the chambers of the water harvesting system,” he added.

The India Meteorological Department has a network of 18 climate reference stations or manual observatories, 156 daily manual rainfall stations, 82 automatic weather stations, 146 automatic rain stations, and five automatic agro stations (soil parameters, rain and atmosphere) in northeast India. It also has upper air observatories with sophisticated instrumentation.

The India Meteorological Department is extending the automatic agro stations by another 18 to get a complete picture of the soil moisture at various depths across the region; it is also installing 40 automatic weather stations along the India border in Arunachal Pradesh, adding more doppler weather radars and getting a data processing server for the reception of meteorological data from IMD’s network and other data from the region.

This article was first published on Mongabay.