S. S. Ray, S. Panigrahy and J. S. Parihar
Space Applications Centre, ISRO, Ahmedabad-380015 [email protected]
Precision farming is one of the most scientific and modern approaches to sustainable agriculture that has gained momentum in 21st century. Consider this situation: ‘A farmer goes to his field with his GPS (Global Positioning System) guided tractor. The GPS senses the exact location of the tractor within the field. It sends the signal to a computer on the tractor, which has a GIS, storing the soil nutrient requirement map in it. The GIS, in consultation with a Decision Support System, decides what is the exact requirement of the fertilizer for that location and then commands a variable rate fertilizer applicator, attached with the tractor, to apply the exact dose at that precise location. And all this is done within a second, before the tractor moves further.’ Sounds like an excerpt from a scientific fiction. But this is what precision farming (PF) means to the large growers of the US or European countries. Hence, the first thing that comes to mind is that, this system is not for developing countries, especially India, where the farmers are poor, farming is mostly subsistent and the land holding size is small. But, this is far from the truth as this approach has a large potential for improving the agricultural production in developing world. This article is an attempt to explain the possibility of adopting PF in India and the role of satellite based remote sensing in facilitating that.
Precision Farming and its Objective
Precision farming aims to improve crop performance and environmental quality. It is defined as the application of technologies and principles to manage spatial and temporal variability associated with all aspects of agricultural production (Pierce and Nowak, 1999). In other words, precision farming is the matching of resource application and agronomic practices with soil attributes and crop requirements as they vary across a field.
Thus, the concepts of precision farming include:
- Variations occur in crop or soil properties within a field.
- These variations are noted, and often mapped.
- Management actions are taken as a consequence of the spatial variability within the field.
A host of terms have been used to describe the concept of precision farming. Generally all these terms are combinations of two phrases. The first phrase is ‘Spatially variable’, ‘GPS based’, ‘Prescription’, ‘Site-specific’ or ‘Precision’, whereas the second phrase can be ‘Farming’, ‘Agriculture’ or ‘Crop production’.
Though, the 20th century agriculture had been characterized by the increase in land and labour productivity, the use of external inputs, an increase in efficiency and efficacy of external inputs, it has also been associated with the stimulation of uniformity in agricultural production areas and the negative side-effects of agriculture. The PF techniques, by appreciating the variability within the field and adopting management practices to cater the variability, are serving the dual purpose of enhancing productivity and reducing ecological degradation. The real value from precision farming is that the farmer can perform more timely tillage, adjust seeding rates, fertilizer application according to soil conditions, plan more crop protection programs with more precision, and know the yield variation within a field. These benefits can enhance the overall cost effectiveness of crop production, however the grower must be willing to make adjustments in his management styles to make it work.
Fig. 1: Structure of LORIS – Local resource Information System (Source: Schroder, 1997)
Developments which prompted PF
Many technological developments, which occurred in 20th century contributed to the development of the concept of precision farming. These technological developments are as follows.
Global Navigation Satellite System
The Global Navigation Satellite Systems (GNSS), such as the NAVSTAR GPS of the USA and the GLONASS of the former Soviet Union have been helpful in pinpointing the precise location. In the post- S/A the positional (horizontal) accuracy of the GPS can be of the order of 20 m, where as that for GPS operating in Differential mode is around 1 m.
GPS-Guided agricultural machinery
This is the major development, which brought the concept of precision farming. There are basically two categories of agricultural positioning system The first category is for monitoring or sensing pertinent soil and crop parameters, such as soil moisture content, nutrient availability weed location or yield mapping. The second category, where in-field position is required, is for control of precision application machinery (variable rate applicators) where actual position of the field vehicle must be related to a digital field map of the relevant parameter. For monitoring/mapping operations only the position in real time is required, where as for precision application, position in real time along with forward prediction of positioning is required.
Fig: 2. Soil spectral variability map of Srirampuram village, Dindigul district, Tamil Nadu, generated using merged data of IRS LISS III and Pan.
Geographical Information Systems (GIS)
GIS has two different roles in precision farming vehicles (Bregt, 1997). First, a combination of GIS and simulation models is highly relevant for precision farming. There are many simulation models for different purpose like the flow of water, crop growth, soil erosion, nutrient and pesticide leaching. GIS helps in integrating geographical data on various aspects such as soil, crop, weather and field history along with simulation models. Another aspect of GIS support to precision agriculture is the engineering component, in which the research findings are translated into operational systems for use at farm level. GIS can support this engineering activity by providing a good platform for storage of base data, simple modelling, presentation of results, development of a user interface, and, in combination with a GPS, controlling the navigation of farm. On the basis of GIS, a decision support system can be developed for operationalisation of precision farming at farm level.
Many farm information systems (FIS) are available, which use simple programmes to create a farm level database. One example of such FIS is LORIS (Figure 1). LORIS (Local Resources Information System) consists of several modules, which enable the data import; generation of raster files by different gridding methods; the storage of raster information in a database; the generation of digital agro-resource maps; the creation of operational maps etc. (Schroder et al., 1997)
Precision farming needs information about mean characteristics of small, relatively homogeneous management zones. These mean characteristics may be obtained from soil tests for nutrient availability, yield monitors for crop yield, soil samples for organic matter content, information in soil maps, or ground conductivity meters for soil moisture. Generally, the fields are manually sampled along a regular grid and the analysed results of the samples are interpolated using geostatistical techniques. Geostatistical modelling of soil, water and crop variability requires that large number of samples at close intervals are collected throughout the agricultural landscape. Such samplings are costly and time consuming. Various workers have shown the advantages of using remote sensing technology to obtain spatially and temporally variable information for precision farming. Remote sensing imagery for PF can be obtained either through satellite-based sensors or CIR video digital cameras on board small aircraft. Moran et al. (1997) in their review paper summarized the applications of remote sensing for precision farming. There are, basically, three approaches for use of remote sensing for precision farming (Barnes et al., 1996).
In the first approach, the multi-spectral images can be used for anomaly detection. These anomalies can be in the forms of disease/pest, weed growth, water stress, etc. Using the reflectance measurements in the visible part of the spectrum, it has been possible to detect diseases and identify weeds from crops. The difference between remotely sensed surface temperature and ground-based measurements of air temperature has been established as a method to detect water stress in plants. However, such type of anomaly detection needs regular observation of the crop through remote sensing sensor. This calls for use of a remote sensing system with high temporal resolution, which can provide at-least 5-6 observation per season. Hence the temporal resolution needed is of the order of a fortnight.
The next approach is based on correlating variationas in spectral response to specific variables such as soil properties or crop yield. Soil physical properties such as soil water, organic matter, soil texture can be correlated to spectral reflectance. Vegetation spectral response has also been used to infer other soil conditions. Crop yields for many crops like, rice, wheat etc. have been found to be highly correlated with spectral vegetation index during maximum vegetative cover. Thus, the yield map generated from spectral images can be used to form management units. To find out within field variability, the remote sensing data should have high spatial resolution. Typically to analyse the variability one is looking for about 750 to 1,500 data points per hectare. With current satellites, one can see areas that are 30 meters x 30 meters (11.1 measurements/ha), 23 x 23 meters (18 measurements/ha), 10 x 10 meters (100 measurements/ha) and 5 x 5 meters (400 measurements/ha). With future satellites, we will be receiving data that have a variety of spatial resolutions (Table 1) that in some cases will be as detailed as 1 x 1 meter or over 10000 data points per hectare.
|Mission/ Agency||Major Specifications|
|PAN (Resolution: 3 m, 5m, Swath: 120 km),
MSS (Resolution: 10, 20 m, Swath: 120 km)
VEGETATION payload (Resolution: 1 km, Swath: 2200 km)
|ORBVIEW-3, Orbital Science Inc., US.A.||PAN (Resolution: 1m, 2 m, Swath: 8 km)MSS (Resolution: 8 m, Swath: 8 km)|
|QUICK BIRD, Earthwatch Inc., U.S.A.||PAN (Resolution: 1m, 2 m, Swath: 36 km)MSS (Resolution: 4 m, Swath: 36 km)|
|RESOURCESAT-1ISRO, India||LISS-IV (Resolution: 6m, Swath: 25 km)LISS-III (Resolution: 23m, Swath: 140 km)AWiFS (Resolution: 60m, Swath: 740 km)|
|CARTOSAT-1ISRO, India||PAN Stereo (Resolution: 2.5 m, Swath: 30 km)|
|CARTOSAT-2ISRO, India||Panchromatic (Resolution: 1m, Swath: 12 km)|
The third approach is to integrate biophysical parameters (such as Leaf Area Index or temperature) derived from high-resolution satellite based remote sensing data, with physical crop growth modeling towards an operational decision support system for precision farming. For example, Moran et al. (1995) utilized remotely sensed estimates of LAI and evapotranspiration as inputs to a simple alfalfa growth model. To derive biophysical parameter, the remote sensing system need to have high spectral resolution, covering the whole range of optical and thermal region.
However, use of RS data for mapping has many inherent limitations, which includes, requirements for instrument calibration, atmospheric correction, normalization of off-nadir effects on optical data, cloud screening for data especially during monsoon period, processing images from airborne video and digital cameras (Moran et al, 1997). Keeping in view the agricultural scenario in developing countries, the requirement for a marketable RS technology for precision agriculture is the delivery of information with the following characteristics:
- Low turn around time(acquired, corrected and processed) ~ 24-48 hrs
- Low data cost ( ~ 100 Rs./acre/season )
- High spatial resolution (at least 2m multi-spectral for 1 ha field size)
- High spectral resolution (10-20 nm for retrieving biophysical parameters)
- High temporal resolution (at least 5-6 dates per season)
- Delivery of analytical products in simpler format
- Prospects of Precision Farming in Indian Agricultural Situation
Precision farming, though in many cases a proven technology, is still mostly restricted to developed (American and European) countries. The reasons for limited implementation of PF in Asian countries are following:
- Small land holdings
- Cost/benefit aspect of PF system
- Heterogeneity of cropping systems
- Lack of local technical expertise
- Knowledge and technological gaps
Out of these, the two major problems for implementing PF in Indian agriculture are small land holdings and cost of PF system. We shall discuss these two and see how remote sensing can help.
In India more than 57.8 per cent of operational holdings has size less than 1 ha. With this field size, and the farming being mostly subsistent farming, it is difficult task to adopt the techniques PF at individual field level. However, for adoption of PF, one can consider, instead of individual fields, contiguous fields, with same crop, under similar management practices. Since, management practices, like seed rate, fertilizer rate etc. are mostly based upon the agro-ecological units, they remain similar for a large area. In these cases the PF can be adopted as Information based agricultural system, i.e. at least the farmer has the information about the soil type of his field before adopting the fertilizer practices. Currently, testing of a large number of soil samples may be time-consuming and costly and it may not catch the variability if sampling is not proper. A remote sensing based soil classification will be able to target the samplings towards the variability pattern and thus overcome the above problems. The figure 2 shows the soil spectral variability map of Srirampuram village of Dindigul district, Tamil Nadu. Even at the first level of classification of merged data IRS LISS III (23 m resolution) and Pan (5.8 m resolution) shows that there are at least four types of soil in this village. However the whole village used to apply a similar fertilizer dose for the only crop gro wn in the Rabi season, i.e. Bengal gram. Presently, this village has been adopted by MS Swaminathan Research Foundation for implementing variable rate application technology.
Tamil Nadu, generated using merged data of IRS LISS III and Pan
Cost is the other major hurdle in implementing PF techniques. The cost of the full-fledged technology was as high as 21050 UK Pounds as on 1997, out of which 13000 pound is for mapping devices alone. This cost is too high considering the economic status of Indian farmers. In this context, remote sensing data provides a cheaper mapping alternative.
IKONOS data (1 m resolution) cost is Rs. 1600 per sq. km. IRS Pan (5.6 m resolution) data is available at a cost of Rs. 15 per sq. km, which is much cheaper than IKONOS. With the launching of Resourcesat -1, one can expect to get 6 m multi-spectral data at less than Rs. 100/sq km. Even after adding up the analysis and the ground truth data cost, the remote sensing based mapping will be far less than the on-field mapping devices and thus can be affordable under Indian condition. The implementation of precision farming in India should have two different strategies – one for the low input subsistent agriculture and the other for input intensive profit making agriculture. In case of the former the increase in productivity is the prime concern. Here, the system has to be converted to information based agriculture, where farmer has spatial information about the soil and crop. This information can be used for efficient input application. Since the field sizes are small in this situation, individually bunded field or a group of fields can be considered as a unit for variable rate application. However, for the later case, such as rice and wheat of Indo-Gangetic belt and the horticultural crops like grape (Mahrashtra), potato (Punjab), tea (Assam), the field sizes are large and the farmers are rich. Already input for farming is high and thereby causing ecological imbalances in many places. Thus the input use efficiency is the prime concern here, apart from enhancing the productivity. Here, remote-sensing data can be used to identify the spatial and temporal variability and necessary actions can be adopted using sophisticated instruments like variable rate applicators.
Precision farming is essential for serving dual purpose of enhancing productivity and reducing ecological degradation. Though it is widely practiced for commercial crops in developed countries, it is still at a nascent stage in most of the developing countries. Remote sensing can provide a key input (variability map) for the implementation of precision farming at a lower cost. The study on precision agriculture has already been initiated in India, in many research institutes, such as Space Applications Centre (ISRO), MS Swmainathan Research Foundation, Chennai, Indian Agricultural Research Institute, New Delhi, Project Directorate of Cropping Systems Research, Modipuram. In coming few years PF may help the Indian farmers to harvest the fruits of frontier technologies without compromising the quality of land and thereby turning the green revolution into an evergreen revolution.
- Barnes, E.M., Moran, M.S., Pinter, P.J. Jr and Clark, T.R. 1996. Multispectral remote sensing and site-specific agriculture: examples of current technology and future possibilities. Published in Proc. of 3rd Int. Conf. on Precision Agriculture, June 23-26, 1996, Minneapolis, Minnesota, ASA. pp.843-854.
- Bregt, A.K. 1997. GIS support for precision agriculture: problems and possibilities. In Precision Agriculture: Spatial and Temporal Variability of Environmental Quality (Eds. J.V. Lake, G.R. Bock and J.A. Goode) John Wiley & Sons, NewYork, pp. 173-181.
- Moran, M.S. , Inoue, Y. and Barnes, E.M. 1997. Opportunities and limitations for image –based remote sensing in precision crop management. Remote Sensing of Environment. 61: 319-346.
- Pierce, F. J. and Nowak, P. 1999. Aspects of precision agriculture. Advances in Agronomy. V. 67. pp. 1-85.
- Schroder, D., Haneklaus, S. and Schung, E. (1997) Information management in precision agriculture with LORIS. In Precision Agriculture’97, Vol.II: Technology, IT and Management (Ed. J.V. Stafford). BIOS Scientific Publishers Ltd., Oxford, UK. pp.821-826.