Name: Rahul Raj
Program: Ph. D.
Name of supervisor: Prof. J. Adinarayana (IITB) and Prof. Jeffrey Walker (Monash University)
Agriculture has been an essential factor of the Indian GDP as well as employment. While India marks the pinnacle globally in the production of pulses, rice, wheat, spices and spice products, ironically Indian agriculture’s operational efficiency has not been able to mark an international stand. Friends, nearly half of India’s population directly or indirectly works in the agriculture sector, but still, the agriculture sector contributes around only 18% of India’s GDP. Earlier, i.e. before 1970, Indian agriculture was greatly influenced by the monsoon conditions which has led to low productivity of the farms. Thanks to the first and second green revolution that the growth rate of agricultural productivity rose quickly with increased storage and transportation facilities. However, excess use of fertilisers, pesticides, and heavy machinery has made the existing practices unsustainable. Now its time to optimise the farm management practices to maintain the harmful pesticide residues in the crop below an acceptable limit, without decreasing the yield of the crop.
In this research, a model has been developed for identification of crop water and nitrogen stress, using UAV based high-resolution RGB, and hyperspectral reflectance data in the VIS and NIR regions. An image of the research farm is shown in figure 1. The study was focussed on the analysis of growth-stage based behaviour of crop biophysical and biochemical properties in conjunction with the input resources supplied in the farm. The biophysical properties like canopy height, leaf area index (LAI), and tassel counts were estimated from drone-based RGB images and validated using ground-based measurements. Tassel counting was also done through the one-band hyperspectral image. Biochemical properties like leaf water content and leaf nitrogen content were calculated using drone-based canopy hyperspectral reflectance data and validated using CHNS-based lab chemical analysis results. The analysis of field-based and drone-based leaf hyperspectral reflectance spectra and CHNS analysis of the leaves make this research data novel. New indices were identified and tested for estimating leaf water and nitrogen content. Since the data is high dimensional, advanced visualisation techniques were used to determine the most useful set of indices to be fed to the machine learning algorithm to estimate water and nitrogen content in the leaves.
The biophysical and biochemical properties extracted through drone-based data were used along with APSIM simulated crop and soil properties. The introduction of APSIM crop simulation model enables the architecture to incorporate weather data for making crop properties/management-related decisions. Based on estimated and simulated results, a Bayesian model was used to create a probabilistic decision on the stress-status of the crop for 1 m × 1 m plot area, as shown in figure 2.
This research will enable a farmer to know when, where and how much fertiliser/pesticides/water need to be supplied on the farm. With such critical information, farmers will not only save money but also optimal use of input resources will improve the crop yield, stop the wastage of groundwater, and reduce the agriculture-related soil and water pollution. The three other sectors which would be induced positively by this research are:
Commercial benefit: Development of such technologies may lead to the building of startups in the field of precision agriculture. Fertiliser companies can use this to help farmers to suggest the amount, location and time of fertiliser application.
Scientific benefit: The data visualisation of complex datasets and extracting information from drone-based hyperspectral data by comparing it to the lab-based chemical analysis of the same crop is new to agricultural research. This research will set a landmark on which many other scientific objectives can be fulfilled.
Community benefit: The optimal use of pesticides and fertilisers will limit the harmful residue to be absorbed by the grains/fruits, which further improve the quality of food and contribute to our healthy lifestyle.
Green revolution one and two have done their job by improving the productivity of agriculture, but to make it sustainable and to optimise the operational efficiency of Indian agriculture, we need DIGITAL REVOLUTION in agriculture, only then, we will be able to achieve - “more crop per drop”!