2022 Volume 10 Pages 1-13
Sustainable use and management of fertilizers depend on the plant nutrition and fertilizer recommendation approaches. The Build-Up and Maintenance approaches are based on the concept of “Feed the soil and Feed the crop” where fertilizers are applied over the selected time period until nutrient levels are raised to the critical soil test levels, then applications are continued at a rate to maintain the nutrient levels to sustain that soil test. The common approach of fertilizer adjustments based on Soil test rating equates general or blanket recommendation to medium fertility status of soil available NPK. For soils testing low or high category, the fertilizer recommendations are increased or decreased by 30 per cent of the general recommendations. Nutrient recommendations through post-harvest soil test values prediction equations have much practical significance. The relevance and value of soil testing increases through yield targeting based fertilizer recommendations. Futuristic approaches like precision agriculture, use of GIS, GPS and RS, Cloud computing, and Big data are versatile components of plant nutrients / fertilizer recommendation. In these approaches, variability of conditions in each field is accounted for and fertilizer recommendation is made precisely so as to optimize the use of soil resources, increase yield, reduce production costs, minimize negative impacts to the environment and ensure optimum returns from its investment. Digital Ag and Big Data analytics bring new opportunities to yield optimization by precision fertilizer management.
The future after the pandemic COVID-19 remains a nightmare. In the midst of this global crisis, we are in a necessity to understand the short-term and long-term impacts on global food security. This pandemic lines up with the already devastating conflicts, economic shocks, climate change and natural disasters which continue to put too many lives and livelihoods at risk of hunger around the world. With only a slight fall in the number of people facing acute levels of hunger globally – from 124 million in 2017 to 113 million in 2018, the number technically in crisis has surpassed 100 million for the past three years [1].
To evade malnutrition, Food and Agricultural Organization of the United Nations (FAO), recommends medium- and long-term policies to boost domestic production of food, such as free or subsidized input distribution, import-tariff or value-added tax cuts on fertilizers and technology for agricultural production, government-funded agricultural research and extension activities, and subsidies for the adoption of new technologies and irrigation [2]. The United Nations (UN) Agencies and their member countries are working towards achieving the vision of the 2030 Agenda for Sustainable Development and its 17 SDGs and Land Degradation Neutrality by responding with various actions and recommendations in relation to sustainable soil and nutrient management. The Committee of Agriculture (COAG) of FAO, during its 25th session held on 26–30 September 2016, requested the FAO to “intensify its food safety work and technical support to smallholders at the local level concerning the safe use of fertilizers and pesticides” [3].
By dramatically increasing the availability of crop nutrition, fertilizers improve ecosystem services of the soils which contribute to 95% of global food production either directly and indirectly. Proper use of nutrients may also promote biomass production and contribute to increased soil organic matter and soil health. However, impacts of fertilizers, if not properly used, include contribution to global climate change, degradation of soil and water resources and air quality, soil-nutrient depletion and potential harm to human, animal and soil health. The perturbations to the biogeochemical flows of nitrogen and phosphorus due to their production for agricultural use have exceeded safe margins for human activities.
Thus, to have the best of fertilizer use, knowledge on fertilizer recommendation approaches is essential. This article provides a comprehensive view on the various approaches of fertilizer recommendation. Efforts to achieve freedom from hunger became successful only after the discovery of the nutritional needs of crops in the mid-nineteenth century. In order to supplement plant nutrients of low fertility soils or poor soils, the value of manures was stressed and mineral fertilizers were developed. Furthermore, the farmer’s economic returns have increased substantially due to fertilizer use in crop production [6].
The Global Soil Partnership (GSP) and FAO subsequently produced the Voluntary Guidelines on Sustainable Soil Management (VGSSM) as a first step to addressing these threats, two of which are ‘nutrient imbalances’ and ‘soil pollution’ and involve fertilizer applications that can be excessive, insufficient or polluting (none of which are sustainable) [7]. The International Code of Conduct for the Sustainable Use and Management of Fertilizers provides a locally adaptable framework and voluntary set of practices with which governments, the fertilizer industry, Agricultural Extension and Advisory Services (AEAS), supporting academic and research institutions, actors in the nutrient recycling industry, civil society and end-users can contribute to sustainable agriculture and food security from a nutrient management perspective by following or adhering to the guidelines and recommendations provided [8].
Fertilizer recommendations have evolved out as a combined research efforts of soil chemistry and soil fertility. Soil testing forms the basis for fertilizer recommendation. Many approaches based on soil nutrient status have been worked out to achieve precise and workable fertilizer recommendations based on the requirement of the crops. In this article, the various fertilizer recommendation approaches which were/ are practiced in India are discussed along with the foreseen futuristic approaches.
The basic philosophies based on which the fertilizer prescriptions are made by the policy makers, Research Institutions and the State Department of Agriculture are as mentioned below.
The Sufficiency Level of Available Nutrient (SLAN) is also called the Percent Sufficiency concept or the Crop Nutrient Requirement (CNR) concept. The goal of the sufficiency approach is to apply enough fertilizer to maximize profitability in the given year of application, while minimizing nutrient applications and fertilizer costs at the same time. The sufficiency approach is commonly referred to as the “feed-the-crop” approach for efficient management of fertilizers [9].
In the SLAN approach, the soil test indices are interpreted as very low, low, medium, high, or very high, and an associated nutrient recommendation (amount) is made. The soil test levels must be calibrated by using yield response trials to determine sufficiency levels. In these trials, the point at which there is no increase in yield is identified as the critical level. A positive aspect of the sufficiency concept is that yields are maximized while annual inputs are minimized. However, applications will need to be made every year to maintain those yield levels. Fertilization based on sufficiency levels is well-suited for short term leases [6].
2.2 Maintenance approachIn maintenance approach of nutrient management, fertilizer is applied, based on the amount of nutrients removed from the field to maintain the soil nutrient level. This approach does not recommend fertilizer application when soil nutrient levels are above the critical level as the soil itself will be able to supply the nutrients required by the crop and no fertilizer response would be expected. It is assumed that crop nutrient removal rate is accurate and allows maintaining soil tests at the critical level. To account for some of the applied nutrients being tied up in chemical interactions in the soil, the actual amount needed for maintenance may be slightly higher than the amount removed by the harvested crop [6].
2.3 The build-up and maintenance approachThe build‐up approach to nutrient management is based upon “feed-the-soil”, in contrast to the “feed-the-crop” concept of sufficiency approach. In this approach, fertilizers are applied over the selected time period until nutrient levels are raised to the critical soil test levels, then applications are continued at a rate to maintain the nutrient levels to sustain that soil test. The process to build-up soil test values is usually spread over four to eight years, depending upon the farmer’s economic situation. Longer term build-up programs help farmers manage their finances by spreading build-up fertilizer costs over several years. But shorter build-up programs can provide earlier benefits from higher soil tests. The build-up and maintenance philosophy has been applied since the 1940’s [9] and is most suitable for immobile nutrients like P and K [10]. However, there are limitations for this approach among the soils with excessive leaching of the nutrients or having the nutrients ‘tied-up’ in unavailable forms. This demands fertilizer application at the time of cultivation.
The build-up and maintenance approach results in establishing soil test levels in a range where a yield response to applied fertilizer is not expected. Soil test levels are maintained to support optimum yields and ensure that nutrients are not limiting. This approach usually works well for phosphorus and potassium, but is not appropriate for N, since N soil tests cannot be built up or maintained. This means P and K would be raised to the critical soil test levels, by applying fertilizer during a long period of time, avoiding a one-time high application rate [10].
2.4 The Basic Cation Saturation Ratio (BCSR) conceptAlso called the Cation Ratio Concept, this focuses on the cations, potassium (K), magnesium (Mg), and calcium (Ca), and, in principle, attempts to maintain desired ratios of these cations on the soil cation-exchange complex. With this philosophy, the desirable distribution of exchangeable nutrients is 65% Ca, 10% Mg, 5% K and 20% H. The resulting desired ratios are 6.5Ca : 1Mg, 13Ca : 1K, and 2Mg : 1K [9]. The Base Cation Saturation Ratio (BCSR) approach (particularly a ‘soil audit’) is a more holistic and comprehensive soil assessment than standard routine soil analysis; it aims to take the ecology of the soil much more into account, and is therefore more suitable to inform fertilizer strategies and sustainable soil management that aims to build long-term soil health, rather than just feeding agricultural crops [11].
The approach is based on the work of William A. Albrecht, a doctor of medicine whose work also aimed to link soil health to human health. The BCSR concept suggests that soils have a balanced or an optimal ratio (or range of ratios) of exchangeable cations in solution, and that there is an ideal ratio between the total of the four base cations Calcium, Magnesium, Potassium and Sodium (Ca, Mg, K and Na) and the total cation exchange capacity of the soil.
Research in the field of fertilizer recommendation is a continuous process and various new approaches, and models are being employed from time to time in order to make the right fertilizer prescription. The practice of providing fertilizer recommendations following scientific principles was developed in 1953 in India. Since then, the serious research and developmental activities performed gave rise to various approaches and models for fertilizer recommendation, which is discussed as follows.
3.1 General / Blanket recommendationThis practice relates the knowledge of soils to the judicious use of fertilizers. This has been followed in India since 1953. Recommendations are arrived at based on the results of model agronomic experiments on government farms and simple fertilizer trials on cultivator's field. Since the fertility variations were not accounted for, uniform adoption of this kind of recommendations did not ensure economy and efficiency of fertilizer use.
Although this is a sound recommendation for majority of the situations, it leads to either over or under usage of fertilizer nutrients [12].
3.2 The probability of response approachThe approach of Fitts [13] relates the probability of obtaining increase in yield from a given fertilizer input, which is plotted against soil test values. The probability of getting a profitable response to fertilizer is high in fields whose soil test is low, and vice versa.
3.3 Soil test rating and fertilizer adjustmentsThe soil testing programme was started in India during the year 1955–56 with the setting up of 16 soil testing laboratories under the Indo-US Operational Agreement for “Determination of Soil Fertility and Fertilizer Use”. In 1965, five of the existing laboratories were strengthened and nine new laboratories were established to serve the Intensive Agricultural District Programme (IADP) in selected districts. Making use of the services of soil testing labs at Indian Agricultural Research Institute, New Delhi and the results of Adhoc Research Projects, tentative soil testing procedures were identified and soil test values were empirically grouped into categories like low, medium and high. The general or blanket recommendation is equated to medium fertility status of soil available NPK. For soils testing low or high category, the fertilizer recommendations are increased or decreased by 25% of the general recommendations [14].
3.4 Nutrient indexing systemParker et al. [15] developed the nutrient indexing system which is very useful in formulating area wise fertilizer recommendations and in comparing fertility status of different areas. Nutrient index is calculated by giving weightage to number of samples falling in low, medium and high soil fertility classes. Indices are calculated separately for different nutrients. The index value of less than 1.67 and more than 2.33 indicates low and high fertility class respectively. Nutrient index value ranging from 1.67 to 2.33 indicates medium fertility class.
3.5 Fertilizer recommendations for a cropping sequence based on initial soil test values through Post-Harvest Soil Test Values Prediction Equations (PHSTVs)Sustainable agriculture requires soil test based nutrient management practices to be adopted for better productivity, profitability, sustainability and environmental safety [16]. Soil test-based approach for nutrient management requires indices related to crop yield, effective nutrient supply during crop growth period, proper monitoring of soil nutrient status and well-developed service infrastructure with excellent quality control which is not feasible in farmers point of view. Hence, it is necessary to predict the soil nutrient status after the harvest of a crop. Also, though there are numerous soil testing laboratories in operation, in a vast country like India with millions of hectares of cultivated land, soil testing for each field season after season and prior to the cultivation of each crop seems to be practically impossible for the want of time, money and labour.
Under these circumstances, the prediction of post-harvest soil fertility status using the pre-sowing soil test values, fertilizer doses and yield or uptake by the crop has much of practical significance. Studies on this aspect were carried out by many workers for various cropping sequences and soil types which has been documented by Subba Rao and Rathore [17] and Muralidharudu [18]. In Tamil Nadu, fertilizer recommendations for cropping sequences based on initial soil test values has been developed by Santhi for onion [19].
3.6 Fertilizer recommendation for maintenance of soil fertility through yield targetingIt is prudent to adjust the fertilizer practices over seasons in such a way so as to strike a balance between high yields and maintenance of soil fertility. The generation of basic data for targeted yield of crops in a crop rotation would hence enable application of fertilizer for appropriate yield targets in multiple cropping for maintenance of soil fertility. Ramamoorthy and Mahajan [20] showed that the yield target and the required fertilizer dose for maintenance of soil fertility can be calculated from the equations:
where, n is ratio between the percent contribution from soil and fertilizer nutrient; r is nutrient requirement in kg q-1; m is ratio between nutrient requirement and contribution from fertilizer nutrient; s is soil test value in kg ha-1
Praveena [21] reported that the targeted yield approach was more balanced, profitable and helpful in checking soil nutrient mining as compared to fertilizer recommendations for economic yield based on regression approach.
3.7 Yield targeting for a fixed cost of fertilizer investmentThe relevance and value of soil testing increases by choosing the yield target at such a level so that the cost of fertilizer requirement becomes more or less same as what was being practiced by the farmer already. The results of such demonstration trials conducted in Delhi villages reveal that the response per unit fertilizer is higher than that from other practices, when balanced fertilization is adopted for targeted yield [22].
3.8 Yield targeting under resource (fertilizer/credit) constraintsWhen fertilizer availability is limited or the resources of the farmers are also limited, planning for moderate yield targets, which are at the same time higher than the yield levels normally obtained by the farmers or the average yield of the district, provides means for saturating more areas with the available fertilizers and ensuring increased total production also [22]. This approach allows balanced fertilizer use under resource constraints and maintenance of soil fertility ensuring efficient and economic use of available fertilizers [23].
3.9 Fertility Capability ClassificationThe Fertility Capability Soil Classification System (FCC) was developed as an attempt to bridge the gap between the sub-disciplines of soil classification and soil fertility. Later different category and descriptions of FCC system were formulated by Buol et al. [24] and modified by Sanchez et al. [25].
FCC is a technical system for grouping soils according to the kinds of problems they present for agronomic management of their chemical and physical properties. It emphasizes quantifiable top soil parameters as well as subsoil parameters directly relevant to plant growth. This system consists of three categories viz., type, substrata type and modifiers. Type and substrata type consider the soil texture in the upper 20 cm and 20–50 cm depths, respectively. Modifiers account for the physical and chemical properties of the plough layer and decide the specific fertility limitations.
FCC system helps the fertility scientists to group experimental sites that are expected to respond similarly to soil fertility management practices and extrapolate their findings to soils that can be expected to behave in a similar manner. For studying the response to fertilizer application for different crops, Lin [26] and Denton [27] used the FCC system.
Precision agriculture is gaining a momentum in the country as it provides efficient management of natural resources leading to increasing yields and economic returns in agricultural production. Variability of conditions in each field is accounted for and fertilizer recommendation is made precisely so as to optimize the use of soil resources, increase yield, reduce production costs, minimize negative impacts to the environment and ensure optimum returns from its investment. The parameters include yield variability, physical parameters of the field, soil chemical and physical properties, crop variability such as plant height, planting density, nutrient stress, water stress and chlorophyll content, anomalous factors such as weed infestation, pest and disease infection and wind damage and variations in management practices including fertilizer and pesticide application, tillage operations, crop seeding rate and irrigation practices [28]. The components of precision farming and site-specific crop management include Geographic Information System, Global Positioning System and Remote Sensing technologies.
The applications of precision farming are not only meant for large areas but fits equally well for small farmers also. In Asia, the International Rice Research Institute (IRRI) has promoted site-specific nutrient management (SSNM) with a view to integrate research and education/outreach, using simple technologies such as a leaf color chart, to be sure the practices and information get to the farmer level. A software called Nutrient Expert (NE) has been developed and introduced in several Asian countries to help crop advisors with a simpler and faster way to use SSNM [29].
Schepers demonstrated that fertilization rates of maize could be adjusted by tracking crop needs with a chlorophyll meter to schedule nitrogen fertigation through center-pivot systems [30]. They reported that fertilizing according to chlorophyll meter readings allowed a savings of 168 kg N ha-1 in the first year and 105 kg N ha-1 in the second year without reducing yields (compared with standard practices). The adoption of these sensor-based technologies can be profitable compared with non-precise fertilizer application, depending on crop and fertilizer prices [31].
Thus, precision farming takes nutrient stewardship to another level, adding the provisions of using the right tools with the right information in the hands of the right people to fine-tune the nutrient management plan for a given field.
4.2 Use of GIS, GPS and RS as tools for fertilizer recommendationGeographic Information System (GIS), Global Positioning System (GPS) and remote sensing technologies serve as versatile tools for the creation of fertilizer variable rates map, foliar nutrient maps, decision support system for fertilizer recommendation practices. GIS applications enable the storage, management, and analysis of large quantities of spatially distributed data in association with their respective geographical features. Variable parameters that can affect agricultural production can be evaluated using GIS analytical capabilities. The ability to depict different, spatially coincident features. The ability to depict different, spatially coincident features is not unique to a GIS, as various computer aided drafting (CAD) applications can achieve the same result.
GPS technology is a versatile tool for management of agricultural and natural resources. GPS is a satellite-and ground-based radio navigation and locational system that enable the user to determine very accurate locations on the surface of the earth. Although GPS is a complex and sophisticated technology, user interfaces have evolved to become very accessible to the non-technical user. Simple and inexpensive GPS units are available with accuracies of 10 to 20 meter, and more sophisticated precision agriculture systems can obtain centimetre level accuracies.
Remote sensing technologies are used to gather information about the surface of the earth from a distant platform, usually a satellite or airborne sensor. Most remotely sensed data used for mapping and spatial analysis is collected as reflected electromagnetic radiation, which is processed into a digital image that can be overlaid with other spatial data [32]. Reflected radiation in the infrared part of the electro-magnetic spectrum, which is invisible to the human eye, is of particular importance for vegetation studies. For example, chlorophyll strongly absorbs blue (0.48 mm) and red (0.68 mm) wavelength radiation and reflects near-infrared radiation (0.75–1.35 mm). Leaf vacuole water absorbs radiation in the infrared region from 1.35–2.5 mm [33]. The health of the crops, vegetation and forests can be obtained by the interpreting the spectral properties of vegetation in different parts of the spectrum.
In India, crop production forecast by remotely sensed satellite data was initiated by the Indian Space Research Organization (ISRO) in the early 1980s leading to the development of operational satellite-based systems for monitoring crop production, horticulture, and crop insurance, assisted by several government organizations and national institutes. Ministry of Agriculture and Farmers’ Welfare (MoA & FW) efficiently employs satellite remote sensing for procuring information on crop statistics needed for agricultural input planning and decision-making.
Pre-harvest crop production forecasts are done for wheat, rice, jute, mustard, cotton, sugarcane, and sorghum, based on spectral indices and weather parameters. This forecast project called “FASAL” (Forecasting Agricultural Output using Space, Agro-meteorology, and Land-based Observations) is undertaken by Mahalanobis National Crop Forecast Centre (MNCFC) established in 2012. The project aims to gather monsoon data and monitor crop growth and production [34].
The project “KISAN” (Farmer) began in 2015 by the MNCFC for optimum crop cutting experiment (CCE) plan and enhanced yield predictions by using high-resolution remote sensing images from satellites and Unmanned Aerial Vehicles (UAVs) [35]. CCE locations were established using several factors, such as the sowing date, Normalized Difference Vegetation Index (NDVI), Biomass, and Leaf Area Index (LAI) obtained by remote sensing. Around 250 CCEs were done in the selected districts of four different states: Haryana, Karnataka, Maharashtra, and Madhya Pradesh. Such projects enable the farmers to make crucial decisions within the season, essentially before crop harvesting [36].
4.3 Cloud computingCloud computing creates a whole ecosystem, from sensors and monitoring tools that collect soil data to agricultural field images and observations from human actors on the ground accurately feeding data repositories along with their GPS coordinates. As an example, sensors are now able to detect the location of a bale of hay in a field, as well as the amount of moisture it contains [37]. The cloud can be used in a versatile manner by the farmer. Information can be accessed from predictive analysis institutes, whereby they can have an accurate prediction on products that are in demand by different markets and adjust production accordingly. Cloud computing has perspective insight on climatic conditions and other parameters which affects crop production. Clouds are knowledge-based repositories containing a wealth of information related to farming practices, crops input, agricultural innovations, pesticides, seeds, fertilizers, nutrients and weed resistance, as well as on equipment. Along with this, expert advice can be sought from a wide range of sources, on farming and processing of agricultural products. This scenario can also be extended to include access to consumer databases, supply chains and billing systems. Information Communication and Technology (ICT) provides greater role in offering greater expertise to producers regarding pricing, good quality seed information, fertilizers, disease detail, sharing new discoveries of scientists working at various Agricultural Institutes. An effective implementation of cloud computing in agricultural sector is encouraging and required for overall development of agricultural sector of India. There are potential risks in cloud computing which if properly addressed can be a potent ICT tool in agricultural sector in India [38].
4.4 Big data, data revolution and analyticsBig data provides for precision data storage, processing and analyzing the data. It also enables searching, aggregating, relating different agricultural datasets to get optimum conclusions in farming. Decisions on crop recommendations, yield prediction, fertilizer recommendation, pest management, forecasting prices, and policy recommendation are achieved through relating factors such as remotely sensed data (crop health, Leaf Area Index, soil mapping, etc.) with the statistical data (rainfall, temperature, and previous yields) [39]. In contrast to most other tools in agriculture. software algorithms and data analytics do not require any hardware installation.
Ideally, software solution should be easily integrated with any external source of data, such as satellites, sensors, drones, machinery and robots. There are many ways farms use big data in making decisions to improve production and profits. For instance, farmers can use the data from satellite imagery to monitor surface temperatures in fields and decide the necessary management practices. They can use sensors in fields or on crops to gather information related to water availability, pest incidence or nutrient requirement [40]. Big data is even used after crops are harvested. Scales that weigh trucks loaded with crop produce can also analyse factors such as moisture content of the products and the production rates. Data can also be used to monitor the crop standards to meet the preference of the consumers which could be included on food packaging enabling the consumers to know their foods. The digital revolution in agriculture is focused on low-cost data collection of soil conditions, weather station and data collected by the satellites [41].
Digital Ag and Big Data analytics bring new opportunities to yield optimization by precision fertilizer management. Fertilizer optimization requires a continuous and complex analysis of large, dynamic data sets. Digital-Ag software are designed to help farmers tackle complex problems in crop production, utilizing the accumulating big data and knowledge. With little effort, farmers can benefit from such analysis and substantially improve their efficiency and decision making. Such software can incorporate inputs on climate, soil, water, genetics, energy, economic resources, field history, yields and more. For example, nitrogen use can become more efficient by analysing soil properties and the rate in which water infiltrates through it, and linking it with intra-field soil testing, mapping of soil and plant nitrogen content, data on nitrogen uptake rates by the crop, yield maps, temperatures, precipitation and more data.
The rapid analysis of the large sets of data also enables farmers to react to real-time events. For instance, the efficiency of nitrogen fertilizers is affected by temperature; a rain event might change the decision regarding the timing of fertilizer application and which type of fertilizer to apply; analysis of satellite imagery can detect problems in specific areas in the field, helping the farmer to locate the problem more easily than ever before etc. Another example for using Big Data analysis is the ability to estimate the yield potential for each area of the field. This greatly affects the optimal fertilizer rates needed.
4.5 Predictive analyticsPredictive analytics uses statistical models and algorithms to predict future events and behaviours. This capability was made possible due to Big data collection. Analysing historical data, such as yields, weather, trends in soil, fertilizer inputs and more, together with real-time data, gives the farmer powerful tools to make informed decisions and manage risks. For instance, minimizing nitrogen leaching can be achieved by using predictive analytics models.
4.6 Decision Support Systems in agricultureAgricultural Decision Support Systems (Agri DSS) are Information Technology (IT) resources that are designed to help farmers tackle complex problems in crop production, utilizing the best available data and knowledge about scientifically-sound best practices. These technological systems, support precision agriculture or smart farming approach, which can reduce labour and fertilizer inputs, minimize negative environmental impacts, and also increase yields.
Agricultural decision support systems can incorporate inputs on climate, water, genetic, energy, landscape, human, and economic resources, and ideally provide an analysis of how these factors work together in influencing productivity. There are a number of systems on the market today. Although the current acceptance of such products among farmers is low, it is expected to change in the future. Smart use of each piece of land is essential to feed the growing world population amidst diminishing availability of arable land.
The Decision Support System for Agro Technology Transfer (DSSAT) of International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT), was designed for users to easily create ‘experiments’ to simulate on computers, outcome of complex interactions between various agricultural practices, soil and weather conditions and suggest appropriate solutions to site specific problems [42]. Selvamani calibrated the DSSAT-CROPGRO, a peanut growth model using the field observations in Inceptisols of Tamil Nadu [43]
The model QUEFTS (Quantitative Evaluation of the Fertility of Tropical Soils) developed by Janssen [44] calculates the yields of crops as a function of availability of soil and fertilizer N, P and K. Smaling calibrated the model based on fertilizer trials conducted in Kenya and made some modifications and validated the model. This model could be one of the tools in IPNS and is being used by IRRI scientists [45].
Computer software DSSIFER (Decision Support System for Integrated Fertilizer Recommendation) was developed to generate crop, site and situation specific balanced fertilizer prescriptions in Tamil Nadu [46] and it has been revised and improved as DSSIFER 2010 [47]. This software utilizes the crop and location specific fertilizer prescription equations based on targeted yield model developed by the AICRP-STCR scheme, Department of Soil Science and Agricultural Chemistry, TNAU and Mitscherlich and Bray percentage sufficiency recommendation equations developed by soil testing wing of the Department of Agriculture, Government of Tamil Nadu and also the blanket fertilizer recommendation adjusted to soil test values.
4.7 Data mining in agricultureMining the large amount of existing crop, soil and climatic data, and analyzing new, non-experimental data optimizes the production and makes agriculture more resilient to climatic change [48]. Data mining is useful in decision making for complex agricultural problems. Some research challenges of agriculture can be solved by data mining algorithms. The k-means algorithm is distance-based clustering techniques. By applying this algorithm, k clusters are formed. Based on Euclidean distance, object is placed into the respective cluster [49]. The k-means algorithm is used to classify soil in combination with GPS. Classification of plant and soil, grading apples before marketing, monitoring water quality change, detecting weeds in precision agriculture, and the prediction of wine fermentation problems can be performed by using a k-means approach [50].
In data mining a statistical model known as Artificial Neural Network is a non-linear predictive model that learns through training and resembles biological neural networks in structure [51]. The neural network is used in prediction of flowering and maturity dates of soybean and in forecasting of water resources variables in agriculture.
Support Vector Machine (SVM) which was originally developed by Vapnik [52] has been widely applied to many different fields, such as signal process and time series analysis. SVMs are one of the newest supervised machine learning techniques. Decision Tree is tree-shaped structures that represent sets of decisions and generate rules for the classification of a dataset. Specific decision tree methods include Classification and Regression Trees (CART) and Chi Square Automatic Interaction Detection (CHAID). In agriculture, Decision tree algorithm is used for predicting soil fertility [38].
Agriculture is a realm where farmers and agribusinesses have to make innumerable decisions regarding fertilizer recommendation and application every day and face intricate complexities as these involve various factors influencing them. The right fertilizer recommendation approach is crucial in sustaining the population of the world by supporting food security, enhancing the livelihood of farmers, providing necessary human nutrition, providing nutrients for the production of renewable materials such as timber, fibre and biofuels and more importantly sustain the health of the soil. A leap towards more futuristic approaches like precision farming, Data mining, use of GIS, GPS and remote sensing, decision support systems and Big data –analytics incorporating the basics of established approaches such as Soil Test Crop Response, Targeted Yield Approach and so on would pave way for the success in the complex event of fertilizer recommendation.
