By Hemendra Mathur
The number of variables impacting the viability of Indian agriculture, as well as the volatility in the magnitude of these variables is increasing.
The key variables impacting Indian agriculture include weather parameters, monsoon, irrigation patterns, soil nutrition, government policy (federal as well as state), commodity prices (global as well as Indian) and consumption patterns.
Changes in any of these variables on either side of a normal curve have a direct bearing on farm economics. The predictability, as well as modeling of risks on above parameters, is a herculean task, which many Indian and global agritech startups are daring to solve for the last few years. There are already some early signs of success, but there is still a long way to go to productise these solutions.
I am confident of the data science capabilities of many of these start-ups (such as Satsure, CropIn, Agnext, Transerve, Agrisk, Harvesting Inc, Oxen, Skymet, Farmguide, Aapah Innovations, VegaMx to name a few) which can lead to higher predictability and better risk management in Indian agricultural supply chain in time to come. The typical customers for such tailor-made data-modelled solutions include banks, insurance companies, agri input companies, food processing companies and the government.
Clearly, there is a huge demand among the agri ecosystem players for “modeled solutions” to solve multiple problems including remote farm monitoring, weather risk management, crop/pest detection, and yield estimation. The demand for these solutions is increasing, and hopefully, modeling will get sharper and accurate with the accumulation and synthesis of multiple “time series” and “spatial” data points enabled by a combination of hardware (such as satellite, drones, sensors, IoTs, cameras, robots) and software (SaaS, AI, ML, deep learning).
Having said that, it is too much of an ask from entrepreneurs to build the solutions at a national or global scale. They have the capability but are constrained by capital and access to data science talent. It is high time that we start thinking about building public digital goods in agriculture which can act as a breeding ground for innovations and multiple APIs to solve problems at scale.
In one of my previous articles, I talked about building an open-source digital platform for real-time access to information on farm, farmers, and crop called Agristack. I already see a few pilots underway to create Agristack, specific to certain crops and regions.
Farmerzone is one of such initiatives to create open-source digital platform for agriculture undertaken by Department of Biotech (DBT). The initial focus is on digitising potato crop in Uttar Pradesh and Punjab. DBT needs to be applauded for conceiving it and developing a working model for implementation involving multiple stakeholders.
A logical progression of Agristack would be the creation of unidimensional and multi-dimensional digital maps specific to agriculture. Agriculture-specific weather maps have already proven their utility. Likewise, it is important to map other parameters which are paramount to farm economics. The creation of these digital tools may sound optional today, but may become a necessity soon. The long-term survival and growth of Indian agriculture will depend on the adoption of such tools.
Though there are multiple such maps needed, but if I have to put my finger on first five to be created, it will be – water stress map, soil nutrition map, crop time series (sowing to harvest) map, mandi map, and consumption map. Let me touch upon on each of these in more detail on why they are needed and possible ways to build these maps.
Water stress map
India is among the most water-stressed countries in the world. As per the study from World Research Institute, 54 percnt of India is water stressed (average per capita availability of water per year in India is about 1540 m3). A recent report by NABARD points to the fact that agriculture productivity in India should be measured in terms of per liter of water rather than per hectare of land as water is scarce and depleting commodity as compared to land.
Over and above, the frequency of droughts is increasing and over 40 percent of cultivable land lies in drought-prone areas. Excessive use of groundwater for irrigation has left many aquifers dry. The spatial distribution of monsoon has also become erratic (for example floods in Kerala and deficit rainfall in northeast region in monsoon months this year).
It is time to start building a repository of groundwater and surface water used specifically for the purpose of agriculture and monitor it on a real-time basis. Satellite imagery can be used to measure and monitor water availability both below as well as above the surface.
These maps can be used as a guide to decide on cropping patterns in the water-stressed and water-rich areas. For example, water-intensive crops such as paddy and sugarcane need to be substituted with water efficient crops such as millets in the high-water stress zones. Given more than 70 percent of available water being used for agriculture, we should be thinking to start tracking “water balance sheet” for the purpose of agriculture so that water usage remains proportionate to available water assets.
Soil nutrition map
The current ratio of key nutrients – nitrogen, phosphorus and potassium (N:P:K) in Indian soil is 6.7:2.4:1 against the recommended ratio of 4:2:1. The ratio is much more distorted in states like Punjab and Haryana where N:P:K use ratios stands at approx. 31:8:1 and 28:6:1, respectively.
The higher skew towards Nitrogen in most states in India can be attributed to higher subsidy on “N” as compared to “P” and “K” (Urea – key source of “N” – is out of nutrient based scheme of fertilizer subsidy, which makes N relatively cheaper to P and K). Such ratios are recipe for disaster which is going to push down soil productivity.
Given significant variation in NPK ratio as well as other critical nutrients (such as Sulphur, Boron, Zinc, Iron, Copper, Manganese), it is important to map their availability in soil. The government’s farmer’s portal is a good attempt to capture the nutritional profile of soil by districts. It also captures pH values to measure acidity and alkalinity in soil; in addition to the nutrients.
The logical progression is to make the soil nutrition data more granular, real-time and actionable. One of the options is to start digitising soil health cards. The soil health card (SHC) scheme launched by the government in 2015, is an excellent government initiative to measure and report soil nutrition.
About 107 million SHC were dispatched in cycle I (2015-17) and about 50 million have been dispatched till date in cycle II (2017-19). The digitisation of SHC and linking it to geo-tagged farms can make data more precise, granular and real-time. The use of technology such as hand-held soil scanners developed by likes of Soilcares can also cut short time and effort in measuring and reporting of soil data.
The soil nutrition map can help farmers in deciding fertiliser application rate customised to the nutrition deficiency in their respective farms. The fertilisers companies will also be big beneficiaries as this data will help in developing custom-made products and identifying the right target geographies. The availability of such data can also play a big role in designing the fertiliser-specific policies of the country.
Crop time series map
Farmer selection of the crop is usually driven by last year’s prices of commodities, which usually lead to glut/deficit situations and non-remunerative prices. The “crop dashboards” depicting area under production at any given point of time can make crop selection more scientific.
The crop dashboards can be displayed on state highways, village panchayat offices as well as sent to farmer mobile phones. Real-time access to “sowing to harvest data” for different crops in Rabi, Kharif and Zaid can address demand-supply imbalance. The wide variations in prices in TOP crops (Tomato, Onion, Potato), seen almost every year, can be taken care of with farmer’s access to this information.
The data of area under production under different crops is currently reported by the Ministry of Agriculture. The data reported on the website is very rich, but the UI and UX of the data can be improved to make it farmer-friendly and farmer-interpretable.
The accuracy of the time series data can be further improved by using satellite imagery and drone monitoring in addition to ground truthing. The crop detection algorithms developed on the basis of imagery by many start-ups are improving by the day, which can over a period of time reduce the need for ground truthing efforts.
The other aggregation / validation points for developing time series data on area and production under a given crop could be the agri-input dealers who on the basis of advance booking and sales of seeds can predict sowing intention of farmers.
This is probably easiest of the five maps listed. The agricultural marketing in India is regulated by Agricultural Produce Market Committee (APMC) Act enacted by the state governments. There are approximately 2,500 principal regulated markets and around 4,800 sub-market yards regulated by the respective APMCs in India. It is possible that some of them are not operational. So essentially, there are approximately 7,000 agricultural mandis under APMC Act, which needs to be geo-tagged along with arrival volumes and pricing details.
Though infrastructure at many of these mandis has not kept pace with the increasing arrivals, village mandis still continue to be key aggregation points for farm produce and account for the bulk of purchase from farmers. The arrival and pricing information across mandis on a real time is of significant value to farmers in deciding on “where”, “when” and “what” price to sell.
A farmer-friendly mandi digital map can enable this decision making. Though, Agmarknet provides some of these data points, it is important to make it easily accessible and interpretable by farmers.
Increasing farmer avenues to sell and ultimately her / his income is possible if she / he has the option to sell at multiple mandis, which may not necessarily be the nearest one. With improving road infrastructure and access to third party logistics company, a farmer has much better access to mandis which may be located in other districts / states. In addition to the mandis, the location of warehouses and cold storages can also be mapped; which can give farmer option to store and transport farm produce in the most optimal way.
The mandi density analysis from such maps in the context of arrival volumes will also help policy makers in re-designing infrastructure available in existing mandis as well as developing a roadmap for the number and locations of new mandis required.
Consumption maps are becoming important because of a significant shift in consumer diet over the last couple of decades. The increasing share of proteins (plant, milk and animal) and fibre at the expense of carbohydrates is an indicator of evolving consumer demand and hence calls for commensurate supply chain alignment.
Also, regional variations and trends in consumption such as increasing consumption of wheat flour in southern India (traditionally a rice eating belt) and growing demand for fish and seafood in non-coastal belts are leading to some sort of homogenisation of food consumption behavior across the country driven by higher awareness and better accessibility. These trends call for building supply chains (including warehouses, processing facility, silos) to cater to evolving demand which is likely to be more dispersed whereas production regions will remain geographically concentrated.
However, multiple layers of intermediaries between consumer and farmer continue to cause distortions in the flow of information and product. There is no way a farmer has the sound basis to plan his production and crop selection on the basis of knowledge of latent and evolving consumer demand trends.
Consumption maps are expensive to build through consumer research route. The National Sample Survey Office (NSSO) collects consumption data from both urban and rural consumer at a frequency of five years. NSSO gives a estimates of household consumption expenditure for both food and non-food categories. Though this is very useful data, but much more is needed to get deeper into categories and to increase the frequency of data availability.
One of the cost-efficient ways to capture consumption data is through aggregation points (including mandis in urban areas) such as Azadpur (Delhi) or Vashi (Mumbai). Kirana stores (accounting for bulk of purchase of grocery items for Indian consumers) could be another source point for data collection to estimate consumption at a more granular level.
The kirana sales of daily grocery is a good reflection of consumption of staples by people living in the catchment area. A sampling of a few kirana stores from each pincode (there are 19,100 pincodes in India) can lead to much more manageable sample size than comprehensive consumer surveys.
Also, there are many POS (Point of Sale) and Kirana-centric start-ups (likes of Jumbotail, Superzop, Shopkirana, Peelworks) who can be partnered with in the future for such data collection. Modern trade and online grocery channels can also be tapped into augment the data collection for the purpose of building consumption maps.
The development of these five digital maps as public goods can be a game changer for agriculture supply chain. The challenge is to make these maps granular, accurate, real-time and user-friendly with least possible cost through optimal utilisation of technology and resources. Given the importance of the sector, this is a good investment into the future for making the supply chain more efficient and transparent.
In addition to stand-alone maps, multi-dimensional analysis of above digital maps (such as mandi density x production, soil nutrition x crop type) can be very insightful to all the supply chain members and policymakers in particular. There is a huge possibility of designing a host of innovations and applications on top of these digital maps by start-ups ecosystem in agritech as well as consumer-tech.
These maps can catapult the much-needed digitization of Indian agriculture to pave way for data-enabled supply chain – for the benefit of farmers, consumers and policy makers.