key: cord-0052066-z3hpfzx0
authors: Saxena, Shalini; Rabha, Amit; Tahlani, Preeti; Ray, Shibendu S.
title: Crop Situation in India, Before, During and After COVID-19 Lockdown, as Seen from the Satellite Data of Resourcesat-2 AWiFS
date: 2020-10-27
journal: J Indian Soc Remote Sens
DOI: 10.1007/s12524-020-01213-5
sha: 7c726c1a1b86a3d0509a5293524ff6e746c9f54b
doc_id: 52066
cord_uid: z3hpfzx0

In this study, an attempt has been made to assess the crop situation in India, before the COVID-19 pandemic lockdown, during the lockdown and after the lockdown. In India, the lockdown was imposed from 24 March to 31 May 2020. That was the period of harvesting of Rabi (winter) season crop and sowing of Zaid (summer) season crop. However, the government announced a large number of measures to provide relaxation to agricultural activities during the lockdown. A satellite remote sensing-based assessment was carried out to see the impact of the lockdown and government measures. Normalized Difference Vegetation Index (NDVI) time composite all-India product derived from Resourcesat-2/2A Advanced Wide Field Sensor for February, May, and July 2020 was used to assess the crop situation, representing three crop seasons, i.e. Rabi, Zaid and Kharif (rainy). NDVI images of 2020 were compared with corresponding images of 2019. Change images were generated, and state-level NDVI values were computed. The state-level cropped area proportion was also mapped using the NDVI thresholding approach. The crop sown area statistics, crop cutting experiment (CCE) numbers and rainfall data were also compared for both the years. It was observed that the differences were very low between the NDVI of 2019 and 2020 in February month. However, the differences were high in many states during May and in most of the states during July. A statistical test of significance (paired t test) was carried out for state-level NDVI and crop area values, which validated this result. This NDVI change was mostly due to increased crop area during Zaid and Kharif (rainy) seasons and higher rainfall from May to July. The satellite and other data (crop area and CCE numbers) analysis also showed the sowing and harvesting operations in major parts of the country went on smoothly, during the lockdown period.

The COVID-19 pandemic caused a large-scale global impact this year (Elavarasan and Pugazhendhi 2020) , with more than 34.8 million cases and more than 1.0 million deaths, as on 4 October 2020 (WHO 2020) . This forced various national governments to impose strict lockdown conditions to control the spread of the disease. In India, the lockdown was imposed for a long period of more than 2 months, i.e. from 25 March to 31 May (Patel et al. 2020 ).

All socio-economic activities were suspended during the lockdown period, except for the essential activities. Considering agriculture as an essential and time-bound activity affecting a large fraction of the population, the government made agricultural activities, including farming operations by farmers and farmworkers in the field, fully functional from 20 April. This helped in the smooth functioning of Rabi (winter) crop harvest and Zaid (summer) crop sowing activities. India also declared a host of welfare measures to protect vulnerable people, including smallholder farmers, agricultural labourers, and migrant workers (Padhee and Pingali 2020) .

India has three major cropping seasons called Kharif/ Rainy (June to October), Rabi/Winter (November to April), and Zaid/Summer (April to June/July). Thus, before & Shibendu S. Ray shibendu.ray@gmail.com; shibendu.ncfc@nic.in lockdown Rabi season was going on and crop harvesting was scheduled during the lockdown period. The Zaid season sowing was to take place during the lockdown. By the time Kharif sowing started, the lockdown was over. The major crops during the Kharif season are rice, cotton, sugarcane, groundnut, soybean, pigeon pea and coarse cereals. During Rabi season, the important crops are wheat, rapeseed & mustard, gram, sorghum and rice. Crops such as rice, pulses and coarse cereals are mostly grown in the Zaid season. The cropping pattern of a region depends on its soil, water availability, climatic factors (Singh et al. 2011) , socio-economic, political, and historical factors (Wardlow and Egbert 2002) . Though crop production is mostly sensitive to rainfall, the economic variables and the government policies, which can modulate subsidies and support prices, also have control over food production (Ghosh et al. 2019) . Hence, to understand the crop situation, there is a need to analyse the weather data and the government policies. Satellite remote sensing is one of the most effective tools that can provide precise and up-to-date information on the crop situation. The main advantage of remote sensing from satellites is the synoptic and repeated collection of data, which allows retrieval of time-series and reliable information on the temporal and spatial pattern of crop area and crop condition over large areas (Bastiaanssen et al. 2000; Dadhwal et al. 2002; Panigrahy and Ray 2006) . While high-resolution sensors provide detailed information about small areas, the sensors with moderate spatial resolution and high temporal revisits allow monitoring of crop situations in a large area. The Indian satellite Resourcesat-2/2A borne Advanced Wide Field Sensor (AWiFS) data having a spatial resolution of 56 m at nadir and a swath of 740 km with 5-day revisit time have been utilized for various resource monitoring activities (Ray and Neetu 2017) . The all-India Normalized Difference Vegetation Index (NDVI) product derived using the algorithm developed by Misra et al. (2012) is highly suitable for vegetation studies on a national scale. NDVI image is often useful because of its high correlation with crop biomass, vigour and canopy cover (Son et al. 2012; Zhang et al. 2011) . The AWiFS NDVI product, which is available with 10-or 15-day composites, has been operationally used for many agricultural applications, such as crop estimation (Ray and Neetu 2017) , drought assessment (Saxena et al. 2019 ) and extreme weather impact assessment (Ray et al. 2016) .

There have been many satellite-based studies for studying the impact of lockdown in India, especially on environmental implications like air quality (Singh and Chauhan 2020 ) and water quality (Garg et al. 2020) . Additionally, there have been a few publications on possible impacts COVID-19 on Indian agriculture (Dev 2020; Kumar et al. 2020; Padhee and Pingali 2020) . However, there was no study to show the impact of lockdown on the cropping pattern, especially using satellite remote sensing data.

In India and many other countries, crop cutting experiments (CCEs) are carried out to measure crop yields, for production estimation and crop insurance (NSSO 2007; Ray and Dubey 2018) . In a CCE, the crop is cut from a randomly selected plot (generally 5 9 5 m 2 ), and grains are harvested, threshed, dried and weighed. A multiplication factor is used to convert the weight of the grains to per hectare yield. Under the Government of India's flagship crop insurance programme PMFBY, a large number of CCEs are conducted in the country to estimate yield at the Insurance Unit level for insurance claim computation (DAC&FW 2016) . The number of CCEs conducted gives an indication about the harvesting operations.

In this context, this current study was carried out to assess the crop situation before (Rabi season), during (Zaid season) and after (Kharif season) the COVID-19 lockdown using Resourcesat 2 AWiFS all-India NDVI products and crop (crop area statistics and number of CCEs) and weather (rainfall) data. The current year's situation was compared with the previous year (2019).

All-India Resourcesat 2/2A NDVI products for February, May, and July, representing Rabi, Zaid, and Kharif seasons, respectively, were procured from National Remote Sensing Centre (NRSC), ISRO with a temporal frequency of 15 days (2 products in a month) for the years 2019 and 2020. The year 2019 was selected for comparison purposes as it was not only the previous year, it was also agriculturally a normal year. The NDVI products are of 56-m spatial resolution. In total, there were 12 NDVI products (2 years, 3 months in each year, and 2 fortnights in each month). The two fortnightly products of each month were combined to generate monthly products, using the maximum value compositing (MVC) method (Holben 1986 ). The agriculture area mask derived from NRSC's land-use land-cover (LULC) map of a 1:250,000 scale (NRSA 2007) was overlaid on the AWiFS NDVI products to extract NDVI maps for the agricultural area only. NDVI images were used for visualization of green vegetation during February, March, and July 2019 and 2020. To assess the change in NDVI, between respective months of 2020 and 2019, NDVI deviation percentage maps were generated, as given below.

Journal of the Indian Society of Remote Sensing

The deviation per cent was divided into three classes (increase, no change, and decrease), as given in Table 1 .

The average NDVI value for each major agricultural state was computed for different months of 2019 and 2020 and compared graphically.

To find the cropped area in different NDVI images, a thresholding approach was used. Various authors have used the NDVI thresholding approach for crop area assessments (Basnyat et al. 2004; Xiao et al. 2006) . Mondal et al. (2014) , using Resourcesat 2 LISS III data, found that the NDVI threshold value for vegetative pixels was [ 0.39 in Kharif, [ 0.23 in Rabi and [ 0.25 in Zaid seasons. Patel and Oza (2014) analysed multi-temporal MODIS data and found that if it was a cropped area, then NDVI should be greater than 0.3. Accordingly, the NDVI threshold value of 0.3 was used to extract the crop area in each season for 2019 and 2020. However, as this value was derived from the literature survey and not based on ground truth, instead of crop area, crop area proportions, i.e. crop area divided by the total agricultural area (taken from NRSC LULC map), were estimated and compared. Figure 1 provides the flow diagram of the steps followed in the satellite data analysis. Satellite image analysis was carried out using ERDAS Imagine image processing software.

A paired t test was carried out to assess the statistical significance of the differences in state-level NDVI values and crop area proportions between 2019 and 2020.

In addition to satellite data, crop-wise state-level area statistics were also collected from the Crops Division of (Saxena et al. 2019) were also used for a year-to-year comparison. District-level rainfall deviation data were collected from IMD, and maps were generated in GIS.

The cumulative rainfall remained above normal (long-period average) throughout the period from 1 March to 30 May 2020. Central India received 104% higher rainfall than the average, the north-west region received 31% higher, eastern and north-eastern India got 6% higher, while the Southern peninsula received 4% lower rainfall. The district-level rainfall deviation map showed that for most of the districts of the country, except parts of Andhra Pradesh, Telangana, Gujarat and the north-east, the rainfall was normal (-19 to ? 19% deviation) or excess (C ? 20% deviation) (Fig. 2) . Similarly, from 1 June to 29 July, the cumulative rainfall was 1% higher than the long-period average. However, from 1 June to 1 July, when the sowing for Kharif season starts, the rainfall at all-India level was 15% higher than the average. The district-level deviation maps show that, except for some parts of Uttar Pradesh, Tamil Nadu, Rajasthan and north-eastern states, most of the districts had normal or excess rainfall.

A comparison of 2 years' rainfall situation showed that there were a larger number of meteorological subdivisions with excess/large excess or normal rainfall in 2020 compared to 2019. Out of 36 meteorological subdivisions in the country, 29 subdivisions had normal or excess/large excess rainfall from 1 March to 31 May and from 1 June to 31 July, while in 2019 the numbers were 13 and 22, respectively (Table 2 ). This showed a much better rainfall situation in 2020 during the whole Zaid season and the sowing period of Kharif season, compared to 2019.

Since only 47.2% of the net sown area of the country is irrigated, the crops grown in Zaid and Kharif season are highly dependent upon rainfall. Hence, this high rainfall during March to May and June-July has resulted not only in an increase in crop area but may also result in an increase in crop vigour and production in both Zaid and Kharif seasons.

The season-wise comparison of the area of crops/crop groups between 2020 and 2019 is presented in Fig. 3 . Though during the Rabi season, i.e. before lockdown, there has been an increase in area, the increase was between 0 and 17% for different crops with an overall increase of 9.5%. In fact, for oilseeds, the area showed a decrease though it was very low (0.09%). However, during the Zaid season, i.e. during the lockdown period, most of the crop/ crop groups showed an increase in the area of more than 25% compared to 2019. The overall increase of area for all the crops during the Zaid season was 35%, compared to 2019. The major increase in the Rice area was found in Similarly, during the Kharif season (i.e. post-lockdown period), as per the sowing progression data of 17 July 2020, there was an increase in overall crop sowing of 21.2%, compared to the same period in 2019. All crops/crop groups together showed an increase of more than 12%, while Kharif pulses and oilseeds showed an increase of more than 30%. As per government statistics, 4.8 million ha higher Kharif area coverage was reported in 2020 (as on 1 October), i.e. 112.2 million ha as compared to 107.3 million ha of the previous year. More area coverage was reported for rice, pigeon pea, mung bean, groundnut, soybean and cotton.

Under PMFBY, the CCE data are collected using an Android App (CCE-Agri) and are uploaded to a central server in real time. These PMFBY CCE data are used for crop insurance purpose. The number of CCEs conducted also indicates the success of the harvesting operations. A comparison of the number of CCEs conducted, under PMFBY, showed that during the Rabi season of 2019-2020 the number of CCEs was 274,769, while in 2018-2019 it was 276,261. The Rabi season CCEs are generally conducted from March to May. Thus, from March to May of 2020, most of which was a lockdown period, there was no significant reduction (only 0.5%) in the number CCEs conducted compared to the similar period in 2019. It indirectly showed that the government policy of relaxation for agricultural activities has helped to smoothen the harvesting operations, and therefore, there was no reduction in CCE numbers.

The Normalized Deviation Vegetation Index (NDVI) images derived from Resourcesat 2 AWiFS data for February, May and July representing pre-lockdown (Rabi season), lockdown (Zaid season) and post-lockdown (Kharif season) periods for the year 2020 are presented in Fig. 4 , along with the images of the similar period of 2019. These NDVI values are only for the agricultural area, as agriculture mask has been used.

NDVI is a proven indicator of vegetation condition or crop health (Rouse et al. 1974 ). The higher green and blue regions in May and July indicate higher crop areas and better crop conditions during this period. Even in February, i.e. before lockdown, the crop condition was better in the central and southern states because of good post-monsoon rainfall in 2019, which resulted in good soil moisture, helping the growth of rainfed Rabi season crops.

The NDVI difference image of 2020 vis-à-vis 2019 shows that most of the area during May and July have more than 5% positive NDVI deviation compared to the previous year. During May, most of the eastern and southern states had higher NDVI in 2020 as compared to 2019. In July, in the NDVI change image, most of the states have a bluish tinge, showing an increase in NDVI compared to 2019.

The state-wise NDVI comparisons, between 2019 and 2020, for February, May and July, are presented in Fig. 5 . The comparison of NDVI values of 17 States shows that (Table 3) . The t test analysis showed that the difference between state-level NDVI values of 2019 and 2020 was highly significant for May and July but not significant for February. During the post-lockdown period, i.e. in Kharif season, because of the good rainfall in June and July, most of the States had increased NDVI compared to that of 2019, except for the states of Bihar, Assam and West Bengal. Lower NDVI in Bihar and Assam might have been because of the impact of floods in these two states (Sphere India 2020) or due to the reduction in the jute and mesta area, which are mostly grown in Bihar, West Bengal and Assam. As per the estimates under the FASAL project (Ray and Neetu 2017) , jute area in these states was 720.6 thousand hectare in 2019, while it was 676.6 thousand hectare in 2020.

To assess the cropped area, NDVI thresholding (with NDVI value C 0.3) was carried out on the images of 2019 and 2020 for February, May and July. The images are presented in Fig. 6 , and the state-wise statistics of cropped area % (cropped area from image divided by total agricultural area from NRSC LULC map) are shown in Fig. 7 . Though this may not exactly compute the crop area, it indicates the cropped area.

In February, i.e. before lockdown, the cropped area in northern states was similar for both years. However, area sown was higher in western states like Gujarat, Maharashtra, Telangana and Tamil Nadu. As mentioned earlier, this may be because of good post-monsoon rainfall in 2019, b Fig. 4 Resourcesat-2 AWiFS NDVI images during February, May and July of 2019 and 2020 which helped in the increase in Rabi sown area of rainfed crops because of high residual moisture (Figs. 6, 7) . In May, during the lockdown, most of the eastern states, i.e. Assam, Bihar, Chhattisgarh, Jharkhand and Odisha, had a significantly higher crop area in 2020 than in 2019 (Figs. 6, 7) . Figure 2 shows that all these states had very good rainfall in the March-to-May period. The southern states like Andhra Pradesh, Telangana and Tamil Nadu also had a higher area in 2020. This has also been found from the government statistics (Sect. 3.2). State-level (for 17 states) average cropped area proportion (%), derived from satellite data, for 2020 was 5.7% higher than 2019 in February, while it was 6.8% and 14.7% higher in May and July, respectively (Table 3 ). Though this shows a similar trend as the government statistics, discussed in Sect. 3.2, the values cannot be compared, as the government statistics is for total all-India cropped area, while the remote sensing-derived estimates are the average state-level area per cent of 17 major agricultural states. The t test analysis showed that the differences between the state-level cropped area (%) values of 2019 and 2020 were highly significant for February and very highly significant for May and July.

In July 2020 image, most of the states had higher areas compared to the corresponding period of 2019, except for Assam, Bihar, West Bengal and Rajasthan (Figs. 6, 7) . As mentioned earlier, the higher area is due to very good rainfall in June and July (the sowing period), while the low area in three eastern states is due to the impacts of flood and cyclone.

The lockdown to control the spread of COVID-19 pandemic was enforced in India on March 24th and continued up to May 31st. This lockdown time coincided with harvesting operations of Rabi/winter season crop and sowing time of Zaid/summer season crops. Globally, there are rising concerns over the impact of COVID-19 on agricultural production (Pu and Zhong 2020) . However, in India, a large number of initiatives were taken to ease the situation (Nayyar 2020; Padhee and Pingali 2020). These included:

(1) exemption of farm operations and supply chains from lockdown restrictions, (2) Indian Council of Agricultural Research (ICAR) issuing guidelines about harvesting and threshing of Rabi crops along with the advisories on handling post-harvest operations like storage and marketing of the farm produce more efficiently (ICAR 2020), (3) the government releasing the first instalment of the income support scheme (PM-Kisan Yojana) payment to farmers, (4) the central bank announcing a moratorium on agricultural loans, and many other initiatives. In addition to these government initiatives, there was a good amount of rainfall and its proper distribution during March-to-May period for the Zaid crop and June and July for the Kharif crop. Overall, the pre-monsoon period (March-May) had 20 per cent excess rainfall, and the monsoon period (June-September) had 9 per cent excess. Thus, the sowing of Zaid and harvesting of Rabi crops progressed well after the initial hiccup during lockdown (Mohan 2020) . Due to intermittent rains during this period, the moisture content in the soil became suitable for the growth of crops (ET Bureau 2020).

All these factors caused a significant increase in crop area, as seen from the Agriculture Department's statistics and also satellite remote sensing data. The good rainfall not only increased the crop area but also improved the crop condition, as seen from the NDVI values and sown area maps and estimates. Also the CCE data (number of CCEs) derived from PMFBY portal showed that there was no impact on harvesting operations. Thus, the year 2020, even with COVID-19 lockdown, not only maintained the agricultural sowing and harvesting operations but also improved upon the previous year, which was also an agriculturally good year.

In this study, an attempt was made to use satellite remote sensing data from the Indian Resourcesat 2 AWiFS sensor to assess the impact of COVID-19 lockdown, from 24 March to 31 May 2020 on agricultural sowing and harvesting operations. Satellite data of before (February), during (May), and after (July) the lockdown period was used. Data were compared with the previous year, i.e. 2019, which was also an agriculturally good year. Crop sown area and crop condition (from NDVI data) were compared. Along with the satellite data, rainfall and crop statistics, including the number of CCEs, were compared. The results showed that the crop area and crop condition were much better during and post-lockdown period compared to 2019. This increase in crop area and enhancement in crop conditions were mostly due to better rainfall situations and the policy of the government providing relaxation for agricultural activities during the lockdown period. The results also showed that the crop harvesting operations must have been conducted smoothly, as there was no significant change in the number of CCEs conducted under PMFBY (using Smartphone).

The lockdown might have got long-term socio-economic impact on the livelihood of the farmers and agricultural workers, as discussed by Kumar et al. (2020) and FAO (2020) . However, as of now, the analysis shows that India had good cropping season both in Zaid (lockdown period) and in Kharif (post-lockdown period) seasons, after an agriculturally very good Rabi season (pre-lockdown period).

Agriculture field characterization using aerial photograph and satellite imagery

Remote sensing for irrigated agriculture: Examples from research and possible applications

Weekly reports of crop weather watch group. Department of Agriculture, Cooperation and Farmers' Welfare

New Delhi: Department of Agriculture, Cooperation and Farmers Welfare

Remote sensing-based crop inventory: A review of Indian experience

Addressing COVID-19 impacts on agriculture, food security, and livelihoods in India, IFPRI book chapters, In: COVID-19 and global food security

Restructured society and environment: A review on potential technological strategies to control the COVID-19 pandemic

Summer crop sowing in full swing, procurement picks up

Impact of the Covid-19 pandemic on food security and food systems

Changes in turbidity along Ganga River using Sentinel-2 satellite data during lockdown associated with COVID-19

Rainfall implications for food production in India: When and where. International Archives of the Photogrammetry, Remote Sensing & Spatial Information Sciences

Characterization of maximum value composites from temporal AVHRR data

State-wise guidelines and advisory for farmers and farming sector during lockdown period due to COVID-19. Indian Council of Agricultural Research

How Indian agriculture should change after COVID-19

Full India NDVI map generation using Resourcesat-2 AWiFS data

Covid-19: Harvesting of wheat on track amid lockdown

Extracting seasonal cropping patterns using multi-temporal vegetation indices from IRS LISS-III data in Muzaffarpur District of Bihar

COVID-19 and agriculture: Strategies to mitigate farmers' distress

Natural resources census: National land use and land cover mapping using multi-temporal AWiFS data (LULC-AWiFS)

Manual on area and crop production statistics

Lessons from a pandemic to repurpose India's agricultural policy

Remote sensing

Some respite for India's dirtiest river? Examining the Yamuna's water quality at Delhi during the COVID-19 lockdown period

Deriving crop calendar using NDVI time-series. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

Rising concerns over agricultural production as COVID-19 spreads: Lessons from China. Global Food Security, 26, 100409

Space technology use in crop insurance: Experiences from India

Handbook on remote sensing for agricultural statistics

Establishing an operational system for assessment and forecasting the impact of extreme weather events on crop production

Monitoring vegetation systems in the Great Plains with ERTS

Comparison of agricultural situation of india for two years (2017 and 2018) using various drought assessment indicators during south west monsoon season in India. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

Cropping pattern of Uttar Pradesh using IRS-P6 (AWiFS) data. International Journal of Remote Sensing

Impact of lockdown on air quality in India during COVID-19 pandemic. Air Quality, Atmosphere and Health

Mapping major cropping patterns in Southeast Asia from Modis data using wavelet transform and artificial neural networks. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

Situation report 5: Flood situation in India

Discriminating cropping patterns in the US Central Great Plains region using time-series MODIS 250-meter NDVI data-Preliminary results

Coronavirus disease (COVID-19) weekly epidemiological update. Geneva: World Health Organization

Mapping paddy rice agriculture in South and Southeast Asia using multi-temporal MODIS images. Remote Sensing of Environment

Spatial analysis of NDVI readings with different sampling densities

The study was carried out under the FASAL