Building Digital Maps Of The Farmlands
Sign in

Building Digital Maps Of The Farmlands

 
print Print email Email
IT Consultant in USA

Google has developed digital maps of the cities.  Now it is easy to locate any address and find one’s way in any city.  What if similar digital maps of the farmlands of the world are created and they get updated frequently showing the growth of crops and possibly the health of the crops and that information is sent to the farmers in the form of advice and alerts?  Well, it sure is an ambitious project. That is what Indigo, Boston, USA based company is venturing to do - to create Indigo Atlas – a geospatial map of all the agricultural lands/farms across the globe and the crops.

 Building Digital Maps Of The Farmlands

Indigo collects data from the field, images from satellites and other sources, and uses machine learning techniques to develop the map.  By analysing the maps of the past few years, the model can predict, and has done successfully, the yield for the next season. 

Crop monitoring was first done by the farmers or growers and was later assisted by satellite imagery (telephotogrammetry) which permitted inaccessible and large tracts of land to be monitored efficiently.  Today, that process can almost be automated with “intelligent computers” (artificial intelligence) that monitor, send alerts, advice, and actions to take, to the farmers.

Welcome to the new world - man vs machine in the yesteryears to man and machine today.  Will artificial intelligence assist humans or desist them? 

A crop’s health must be monitored to decide on the amount of water and fertilizer to be added, the herbicide treatments, if needed, to be given.  A number of foliar contact and direct measurements can be done.  But this process becomes cumbersome.  Crop monitoring is physically done by visual inspection.  When a farmer has thousands of acres of crop to take care, physical monitoring is impossible.

However, to automate it, one must identify the appropriate biophysical variables that characterize the health of the crop and devise methods to measure them.  The number of variables measured, the frequency of measurements, the reliability and accuracy of measurements all decide on the outcome of monitoring and how reliable it is to take actions based on. The idea of using artificial intelligence is simple and as follows.

Plants reflect sunlight, near infrared light (NIR), at different frequencies.  Chlorophyll is an indicator of the health of a plant.  It strongly reflects more NIR and less red light.  At the bottom of a leaf is a spongy layer that reflects light.  When a plant is affected, the spongy area is also affected, and its reflectivity varies; the amounts of red light reflected increases and NIR light reflected decreases.  Thus, by measuring this variation in the reflected NIR light compared to red light, the health of the plant can be ascertained.  This is the fundamental principle behind satellite imaging. 

Creating a map of the farmlands is difficult compared to that of cities because the crops change very often.  Indigo combines what the company calls as three truths as described below.

Ground Truth:  These are the important observations made by the farmers, and agriculture experts from Indigo who visit the farms. They include major decisions taken by them.

Field Truth:  These are the data collected from the machines (such as combines, sprayers, planters etc) that are used in the farm.  This involves machine-to-machine communication. 

Space Truth:  These are the satellite images of the crops, the field and adjoining areas. 

Using these three, Indigo calculates a parameter called, ‘Crop Health Index’ (CHI), which is an indicator of the health of the crops.  This can provide information about plant growth, type of soil, topography and water absorption.  An early identification of the health provides earlier intervention to save the plant.  The three - CHI, weather information and US government crop reports, - are fed into a proprietary machine-learning algorithm developed by Indigo.  The details of these calculations are proprietary. 

So far, Indigo has instrumented about 50,000 hectares of land and has collected volumes of data at the rate of 1 billion data points every day.  Indigo has been predicting end-of-the-season yield for many crops.  The company claims that its predictions done four months ahead of the predictions of United States Department of Agriculture’s (USDA) and those of markets are very reliable.  Additionally, the model has also been able to aid the county governments by providing an accurate assessment of damage to grain bins, and farms due to natural calamities like flood.  This was done by combining maps of flooded areas, maps of locations of storage bins and crop growth.

Government agencies can start relief measures and plan for the distribution of monetary relief to the farmers and other affected people.  Indigo feels that this model will create a digital map of world’s farmlands, guide farmers in their farming practices and predict the yield before the season is over.

start_blog_img