Implementation of Smart Sampling

The vagaries of weather, especially in the background of climate change, has forced Indian Government to take firm steps to relieve farmers from the related stress. The situation gets complicated when farmers are forced to commit suicide in the want of money due to failed crops. By keeping this in the view, the Indian Government brought a very ambitious insurance scheme namely: Pradhan Mantri Fasal Bima Yojana (PMFBY). This is not only an insurance scheme but it does push for adoption of new and cutting edge technologies such as Geographic Information System (GIS), Remote Sensing, Drones, Artificial Intelligence (AI) etc. for the better assessment of nation-wide agriculture conditions and productivity.

    Furthermore, PMFBY is a yield-based crop insurance scheme that uses Crop Cutting Experiments (CCEs) to determine the yield loss suffered by the farmers due to natural catastrophes and adverse weather conditions. As an improvement on the CCEs, various agencies are trying to perfect the method of yield calculation and reporting. One such method is to amalgamate CCE with technology to select the sample sites for CCEs. These samples must be representative of the crop conditions in a given area, normally a Gram Panchayat(GP). A method in this regarding was proposed by Murthy et al. (1997) whereby Normalized Differential Vegetation Index (NDVI) from satellite images is to be used for assessing the conditions (thus creating strata) of crop and then taking a random sample from each such strata. In a sense, smart sampling is nothing but stratified sampling method. Now, the question comes how to implement this. Since there are no off-the-shelf tools available for smart sampling, a tool need to be produced by combining other smaller tools. This can be done in ArcGIS, R, Python etc. platforms.

Here I am showing how smart sampling can be done in freely available statistical programming language R:

R code for Smart Sampling using NDVI

#First Libraries
library(raster)

## Loading required package: sp

library(rgdal)

## rgdal: version: 1.5-17, (SVN revision 1070)
## Geospatial Data Abstraction Library extensions to R successfully loaded
## Loaded GDAL runtime: GDAL 3.0.4, released 2020/01/28
## Path to GDAL shared files: C:/Users/Gulshan/Documents/R/win-library/4.0/rgdal/gdal
## GDAL binary built with GEOS: TRUE 
## Loaded PROJ runtime: Rel. 6.3.1, February 10th, 2020, [PJ_VERSION: 631]
## Path to PROJ shared files: C:/Users/Gulshan/Documents/R/win-library/4.0/rgdal/proj
## Linking to sp version:1.4-4
## To mute warnings of possible GDAL/OSR exportToProj4() degradation,
## use options("rgdal_show_exportToProj4_warnings"="none") before loading rgdal.

# read the NDVI data
ndvi <- raster("path_for_NDVI_image.tif")
gp <- readOGR("path_of_directory_whereyour_gp_locationis_stored", "shapefilename")

## OGR data source with driver: ESRI Shapefile 
## Source: "path shapefile will appear here", layer: "name of shapefile will appear here"
## with ## features
## It has ## fields
## Integer64 fields read as strings:  Tabulat_sh Tabulat__1

#If you NDVI image doesnot have the same projection as your shapefile, make them into one projection system.
# Reproject the NDVI image - one time process.

#newproj <- "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0"
#ndvi_repro <- projectRaster(ndvi, crs = newproj)

# Once the data is prepared loop the data through a for loop as below and it should work.



for(i in 1:dim(gp)[1]){
  a <- crop(ndvi_repro, extent(gp[i,]))
  b <- rasterize(gp[i,], a)
  a <- mask(x = a, mask = b)
  #here I am reclassifying the NDVI.
  class_diff <- (max(getValues(a), na.rm = TRUE) - min(getValues(a), na.rm = TRUE))/5
  z <- c(min(getValues(a), na.rm = TRUE), (min(getValues(a), na.rm = TRUE) + class_diff), 1,
      (min(getValues(a), na.rm = TRUE) + class_diff), (min(getValues(a), na.rm = TRUE) + 2 * class_diff), 2,
      (min(getValues(a), na.rm = TRUE) + 2 * class_diff), (min(getValues(a), na.rm = TRUE) + 3 * class_diff), 3,
      (min(getValues(a), na.rm = TRUE) + 3 * class_diff), (min(getValues(a), na.rm = TRUE) + 4 * class_diff), 4,
      (min(getValues(a), na.rm = TRUE) + 4 * class_diff), (max(getValues(a), na.rm = TRUE)), 5)
  reclass_z <- matrix(z, ncol = 3, byrow = TRUE)
  a <- reclassify(a, reclass_z, right = NA)
  x11() # this will pop up a new plotting window.
  plot(a)
  # if you wish to save the classified NDVI image; just remove # and give appropriate path.
  #writeRaster(a, filename = "givepathnameofclassifiedndvi", overwrite = TRUE)
  sampleSp <- sampleStratified(x = a, size = 1, xy = TRUE, sp = TRUE, na.rm=TRUE)
  sampleSp@data$area_per <- round((sampleSp@data$cell/sum(sampleSp@data$cell)) * 100, 1)
  plot(sampleSp, add = TRUE, pch = 18, col = 'red')
  title(i)
  # if you wish to save the ss points; just remove # and give appropriate path.
  #writeOGR(obj = sampleSp, dsn = "givedirectorypath_for_points_to_be_stored", layer = nameofthesamplepoints[i], driver="ESRI Shapefile", overwrite_layer = TRUE)
}

The points (selected smartly ;-)) will appear in red colour. Your output should look like the view in the video.

References:

Murthy, C. S.; Thiruvengadachari, S.; Jonna, S.; Raju, P. V. Design of Crop Cutting Experiments with Satellite Data for Crop Yield Estimation in Irrigated Command Areas. Geocarto International, 1997, 12 (2), 5–11.

Super Cyclone: A powerful destructive tropical Cyclone "Amphan"

There are two types of cyclones i.e. 1) Tropical Cyclones and 2) Extra-tropical cyclones. Tropical cyclones, as name suggests, forms in tropical region while the extra-tropical form in mid-latitudes. Tropical cyclones are comparatively smaller, longer-lived and are more intense. On 20 May 2020, a tropical cyclone named Amphan made landfall on eastern part of Indian Sub-continent. 

Amphan Cyclone
Image Courtesy: NASA, satellite image from MODIS onboard Aqua satellite

Development:

For the development of a tropical cyclone, following conditions are to be met:

1. This type of storm typically develops over oceans between 10° and 20° N and S latitudes and no closer than 5° from the Equator.
2. Tropical cyclones form over the sea where a threshold sea surface temperature of ~26.5° Celsius is reached. 
3. An unstable environmental lapse rate is also required
4. Weak winds with little change in speed or direction with height are also needed 
5. The atmosphere must have a high-water content through the bottom 5 km of the atmosphere.

Category:

 To assess the intensity of tropical cyclones ranked on Saffir–Simpson scale which rank storms from category 1 (weak) to category 5 (devastating). Amphan has been ranked category 5. The intensity of a tropical cyclone is based on the central pressure of the storm, mean wind speed, and height of the accompanying sea-level rise. Further, Amphan has been classified as "Super Cyclone"; cyclones in the North Indian Ocean with winds surpassing 220 kilometers per hour are considered “super cyclonic”—the highest classification in the north Indian basin. On May 18, Amphan sustained winds of 265 km per hour. Amphan and, the last super cyclonic storm recorded was Odisha in 1999.

Track of Amphan:

Tropical cyclones move from west to east. Amphan made landfall between Digha in West Bengal and Hatiya Island in Bangladesh. After landfall it traversed as shown below. On the approach of cyclone, there is sudden rise of coastal water which is known as storm surge. Storm surge take sea water far inland and cause damage. 

Track of Amphan
Image Courtesy: The Hindu Newspaper, dated 21-05-2020

Destruction:

Since the category of Amphan is 5, devastation is very high wherever Amphan visited, particularly south West Bengal and 24 Pargana district.

A play of thoughts and an escape route

Since this blog has two words namely Spatial and Ideas, I find it possible and interesting to delve into these separately and at the same time together. Consider this, in 1947, a geographer named John Kirkland Wright coined a term "geosophy" to describe our geographical imaginations. He suggested that geographers should use the geographical knowledge of non-geographers such a fisherman, farmers, drivers etc. to get their insight on geographical landscape. His concept, however, did not find immediate taker but later on a whole new school of thought i.e. Humanistic geography based on such a concept, apart from philosophical foundations of Phenomenology, wherein concepts like mental maps became very influential.

The idea is that we constantly gather information about our surroundings and tend to connect these to form our own concepts about any said thing or phenomenon. As a fact, same information produces different concepts for different people, this not only enriches our understanding of the world we live in but enriches a discipline the ideas belong to.

Having said this, in this post I want to explore the nature of our ideas and our thoughts leaving spatial part for an another post. Our thoughts are like unbroken string whereby one thought is connected to another to produce an understanding of a phenomenon. Thoughts, in this sense are similar to our breath which is also a chain of inhalation and exhalation. We cannot stop our breath, or may be for some time, similarly we can not stop our thoughts but may be partially during our sleep. This is how we have evolved, our mental energy consumption is highest among animals. A human brain can consume as much as 25% of energy while idling as compared to 5% in case of an elephant. Now, interestingly, our thoughts tend to produce emotions leading to production of moods. Many of us find ourselves invested in thinking about past or future which is like day dreaming as per Sam Harris, a leading philosopher who also runs a meditation App called Waking Up. This day dreaming sails us away from our present and produces negative or positive emotions as per our negative or positive thoughts. Once an emotion is formed, we tend to sustain it by producing similar kind of thoughts thus creating a gloomy or happy mood. The problem arises when our decisions are ridden on these emotions, our decisions are clouded by emotions producing bad/wrong decisions. If a person can take decision setting aside one's emotions, she is called an emotional intelligent and said to have high Emotional Quotient (EQ). A common thing noticed in leaders is that they have high EQ and their decisions are not overpowered by emotions.

Some people are more gloomy while others seems to be happy, this is because of the interplay of such thoughts. Since thoughts are a continuous chain, it seems to be difficult to get out of a mood. In a similar argument, it has been contested that once a behavior of a person is set e.g. gloomy or happy-going, it cannot be changed. Such people find themselves trapped in thoughts, extreme cases of them turn suicidal. However, as per Matthew Ricard, a known western face of Buddhist monks, meditation provide a ray of light in this dark tunnel. Even though, Yuval Noah Harari, in his famous book Sapiens a brief history of Humankind, contested that 99% of buddhist monks do not reach nirvana. Sam Harris on his App is positive that on continuous practice of meditation, it is possible to make changes in our brain's functioning where old neurons may disappear while new one will be created. To the depressed and their extreme cases i.e. suicidal people, this brings a ray of hope. The key point here is people do not need to attain nirvana but to keep on living their life as happy as possible by identifying their thoughts and escaping from potential dangerous ones. This possibility comes through a regular practice of meditation. As a disclaimer I declare that I too find myself engaged into day dreaming of past or future thoughts and try to cut them as short as possible by practicing meditation.  

Ironically, Meditation which is an inseparable tradition of Buddhist culture, seems to have been forgotten in its own motherland i.e. India. When EQ turn people into better decision makers, why meditation is not included in schools? The reason for including meditation courses in schools is that by the time children grow up, they will have clear understanding of their emotional flush. This will create a possibility of germinating leadership qualities in our future generations. However, there is no right time to start, people of all ages can start to map their emotions through meditation. Since I am writing this blog during the peak of COVID-19 pandemic, meditation can be helpful to deal with this stressful and uncertain situation.

On how to start meditation there are number of good sources such as Matthew Ricard's book: The art of meditation or Swami Vivekananda's book: Meditation and its methods. It is a bit easier to start with guided meditation, for that, one can look up Sam Harris's Waking Up App.