Behind differing farm fire counts: Satellite resolution, methodology

The National Aeronautics and Space Administration (Nasa)’s VIIRS-SNPP satellite showed 74,015 farm fires in Punjab between October 1 and November 16 .

But data collated by the Indian Agricultural Research Institute, using a combination of both VIIRS-SNPP and the MODIS satellite and sensor systems, which is used by the Commission for Air Quality Management (CAQM) in the National Capital Region, and the Central Pollution Control Board, showed cumulative fire counts in Punjab only touched 68,922 till November 16.

Then there is data from System of Air Quality and Weather Forecasting and Research (Safar), part of the Union ministry of earth sciences, whose data doesn’t match that of Nasa or IARI. For instance, the number of fires, according to Safar, peaked on November 7 (4,189), but this doesn’t tally with the fire counts being released by IARI or Nasa.

CAQM has now asked states to only rely on data being collated by IARI.

Experts say a difference in methodology, satellite resolution, and the area being scanned can explain the differing farm fire figures, with Safar’s daily count not only comprising the fires recorded in Punjab and Haryana, but also Uttar Pradesh and Jharkhand. It also excludes some fires.

“Ours is a difference in methodology, as we combine data from two Nasa satellites and two Isro satellites, using a combination of the VIIRS, MODIS and INSAT-3D/3DR (Indian National Satellite System) to generate a fire count. This count only includes those that can influence Delhi’s air, so some fires are omitted,” explains Gufran Beig, founder and project director at Safar.

Beig says while the resolution of NASA’s SNPP satellite can go down to 300 x 300 metres, thereby being able to capture even smaller fire counts close to the surface, Indian satellites have a resolution of 3x3 kilometres, 10 times lower, which can lead to a lower fire count. It’s also important to see how often the images are captured, he adds.

CAQM, when contacted, clarified that all states had been asked to adhere to a framework laid down by Isro before August 30 which will ensure a uniform process is followed in detecting farm fires. “A combination of SNPP-VIIRS and MODIS (Terra and aqua) satellites and sensors is in place, which is being shared by IARI. Overall numbers so far show a reduced count this year as compared to last year.”

According to a research paper, “Objective evaluation of stubble emission of North India and quantifying its impact on air quality of Delhi” which was published in the Science of the Total Environment journal last year, Indian satellites have an advantage in terms of a geostationary orbit and capturing data every 15 to 30 minutes, but have a disadvantage because of coarser resolution.

“MODIS and VIIRS can detect a larger number of fires, and INSAT-3D due to its better temporal resolution could detect even those fire events which MODIS missed. There are number of instances where anomalies were found on day-to-day basis and rectified carefully. VIIRS was often found to detect additional counts which were never confirmed by ground surveys that might have occurred due to festival related firecrackers or sources other than stubble burning,” the paper states.

Dr Mahesh Narang, principal extension scientist at the Punjab Agricultural University (PAU) in Ludhiana says IARI uses satellite data that captures fire counts multiple times a day, but adds that there is need for a common methodology across all states to ensure numbers can be correctly calculated.

“There is no right or wrong system, but if we are comparing satellite data or remote sensing data, we need to analyse it with the same satellite data from last year, to get accurate numbers. Data for Punjab this year shows a 13.2% reduction and if the same satellite is analysed next year, you can get accurate figures to compare,” Narang says.

The Council on Energy, Environment and Water (CEEW), which analysed Nasa’s VIIRS satellite data between September 15 and November 21 showed Punjab recording 76,320 fires with Haryana recording 9,409 fires during the same period. The figures for the corresponding period under the Isro protocol shows 71,024 fires in Punjab and 6,642 fires for Haryana.

If November is analysed, Safar’s monthly fire count recorded 64,051 fires, however, this figure includes fires from UP and Jharkhand as well, while omitting certain fires with ‘low’ intensity. The Isro fire count showed 61,272 fires in Punjab and Haryana, a difference of around 6,000 fires from Nasa’s VIIRS, which recorded 67,754 fires.

S Narayanan, Member Secretary, Haryana State Pollution Control Board (HSPCB), said the state was acting as per the CAQM’s guidelines and using the ISRO satellite numbers, being collated by IARI. “Last year, we were using data from Haryana Space Applications Centre (HARSAC) in Hisar and the numbers were not as accurate. This year, the Commission has taken this decision to have a uniform system across all states. The figures being compared for this year are being done with the same satellite data for last year now,” says Narayanan.

According to LS Kurinji, programme associate at CEEW, Sangrur in Punjab is leading the fire count so far this year with 8,872 logged between September 1 and November 22, followed by Bathinda, Moga and Ferozepur districts.

“Ludhiana also reported a marked rise in farm fires this year, reaching 5,395 compared to the last year’s figure of 3,885. In Haryana, Fatehabad reported a maximum of 1,968 fires, followed by Kaithal, Jind and Karnal districts,” she says.