eChapters

The calculation of raster data statistics is very essential before performing some geoprocessing operations. Raster data statistics involve knowing the distribution of pixel values in a band, comparing the pixel values of two bands, extracting values of all bands for a particular geographical location or point, and computing statistics like mean, median, standard deviation, etc. at pixel level across all bands of raster dataset.

In R, codes to perform raster data statistics are present in raster package. The visualisation of raster data statistics can be done by rasterVis package. So, we need to import these packages in R in order to perform this exercise.

Many times, users have to work with raster datasets of different spatial and radiometric resolutions. So, there is requirement of bringing their resolutions to the same resolution in order to facilitate their processing. Resampling is the process of deriving pixel values from the existing image to the new raster image of different pixel size. Resampling is most common pre-processing step in remote sensing. It is done both at spatial and radiometric scale. At spatial scale, resampling is done either to increase the spatial resolution i.e. decrease pixel size known as Upsampling or to decrease the spatial resolution i.e. increase pixel size known as Downsampling. Radiometric resampling is performed to increase or decrease the radiometric resolution of the image. In R, resampling of raster image is performed by raster package. The details of the resampling methods and their respective R – code with examples are discussed in this chapter.

First of all, we need to import the raster image on which we wish to perform resampling operations by using following R – code. In R, resampling is feasible on one band only. So, we will import single band raster image.

Mosaic is the combination of two or more raster images into single image. Sometimes, single image captured by remote sensing satellite doesn’t completely covers the desired study area mostly. Therefore, users resort for mosaicking two or more images captured by satellite sensors from different paths so that the whole study area is merged into a single raster image to facilitate the image processing and analysis. The pixel size of the raster images should be same for mosaicking.

In most of the cases, there is spatial overlapping among the raster images from which we want to create mosaic. These overlapping areas are handled in multiple ways which is being discussed further using R – code example. For mosaicking raster images in R, mosaic function of the raster package is used.

Vector data analysis is performed to convert raw input vector data to output that provides new and meaningful information which ultimately supports decision making and disclose hidden patterns. There are various techniques to analyse vector data like buffering, intersect, union, and so on. Some of the most commonly used techniques with their R-code examples are explained in this chapter.

First of all, you need to import all the packages which are required for performing this exercise. These packages include rgdal and sp for importing and re-projecting vector data as well as raster and rgeos for performing vector analysis operations.

Raster sub-setting is done to select the desired area or bands of the raster image on which the user wants to perform analysis. Raster sub-setting is done in two ways:

Spectral subset: Spectral sub-setting is done to extract few desired bands from the multiband raster image which are required for analysis.

Spatial subset: Spatial sub-setting is done to minimize the area or extent of raster image to the desired area or extent for further analysis.

Raster sub-setting in R is done by functions available in the ‘raster’ package. The following packages are required in this chapter:

Coordinate Reference System (CRS) is a framework of coordinate based local, regional or global reference system used to represent the locations of geographic features. Coordinate Reference Systems (CRS) provide a standardized way to define the real-world geographic locations which further enable us to use common locations of geographic datasets for integration. The datasets from different sources and CRS are transformed to a common CRS in order to perform various integrated analytical operations like overlaying of data layers. Coordinate Reference Systems are mainly of two types:

1. Geographic Coordinate System: This is global or spherical coordinate system in which locations of features on earth are defined in terms of latitude and longitude values. Latitude and longitude are angles based on a point at the centre of the earth. In this system, locations are defined by degrees, therefore, distances cannot be measured accurately. The most common geographic coordinate system is the World Geodetic System 84 (WGS84).
2. Projected Coordinate System: In this system, locations of features are defined using cartesian x, y coordinates on a flat, two – dimensional surface. This system is based on sphere or ellipsoid projected on flat surface. Projected Coordinate System are generally referred as projec- tion. It is required for preparing maps on paper or computer screen. Un- like Geographic Coordinate System, projected coordinate system has ac- curate measurements of distance, angles and areas across the two dimen- sions. However, there is alwaystrade-offs between area, direction, shape, and distance in this system due to distortions created while representing ellipsoid or spheroid on a flat map. So, the projection type is selected based on its application. Most common projections are Universal Trans- verse Mercator (UTM), Lambert Conformal Conic and Albers Equal Area.

Vector data consists of points, lines and polygons with x, y coordinates known as vertices representing the geographic location i.e. latitude and longitude. Vector data depicts the shape of the spatial feature. Points consist of single x, y locations. Examples of points are tower location, sampling location, etc. Lines represent 1 – dimensional features like roads, streams, etc., and are composed of many or at least two connected points or vertices. Polygons denote 2 – dimensional features like boundaries with three or more vertices that are connected and closed. The vector data are represented in the following figure (fig. 5.1).

Sensors onboard the satellite records the intensity of emitted or reflected electromagnetic radiations in the form of digital number (DN) for each pixel. These digital numbers can be transformed to more meaningful intensity units like radiance or reflectance. This process of conversion of digital numbers to radiance or reflectance is known as radiometric calibration. Sensor specific information is required to conduct radiometric calibration.

The satellite data is available in raster format. Raster data consists of two – dimensional array of pixels or grids which stores the values representing features. All the pixels are of same size and shape with their particular coordinate location. A raster data format is shown below.

2.1. Open source satellite data

The satellite images have been invaluable information sources for Natural Resource Management. They provide synoptic view of the Earth periodically helping in inventory, surveying, and monitoring in agriculture, hydrology, geology, mineralogy, land cover, land use and environment. Free global satellite image data of spatial resolutions of 1m to 1 km and spectral resolution of single band to hyperspectral are available from an increasing number of sources (Table 2.1).

1.1 R and R studio

R has become the world-wide language for statistics, predictive analytics, and data visualization. R is not limited to statistics; it has a wide range of tools for spatial analysis and digital image analysis and can handle any process needed for image analysis flexibly. It offers a wide range of methodologies for digital image processing, from the most basic to the most complex. As an open source project, it’s freely available for a range of platforms, including Windows, Mac OS X, and Linux. An integrated development environment (IDE) for R, R Studio, allows interactive execution of R functions with user interface. RStudio is available in two formats - RStudio Desktop and RStudio Server - both available in free and fee-based (commercial) editions.

1.1.1. Download and install R

The R software setup files can be downloaded from any of the widely distributed Comprehensive R Archive Network (CRAN). To download R, please choose your preferred CRAN mirror, a location close to you. Follow the following steps:

In conventional breeding for genetic upgradation of yield and other desirable attributes of crop plants, favourable assemblages of genes from two or more parents are made to develop a commercial variety. The important steps of this exercise are: locating the right kind of parents that possess the desired characteristics; making crosses between or among them; handling the segregating generations in an appropriate manner so as to identify the required recombinants; stabilizing the recombinants and finally evaluating their performance in a series of trials. This approach has been very successful in the sense that it has provided increments in productivity continuously. In an analysis done by the American Society of Agronomy on the genetic contributions to yield gains of five major crop plants, it was concluded that sorghum yields increased due to genetic improvements at the rate of 1.2 per cent per year between 1950 and 1980. For corn, the corresponding yield increase was 90 kg/ha/year between 1930 and 1980, which is equivalent to a rate of increase of 1 per cent per year. It was also estimated that 70 per cent of the total gains in corn yields are attributable to genetic improvement and 30 per cent of the yield gains are due to non-seed inputs like fertilizers, pesticides and cultural practices.

In India also, the improvement of productivity of the major food, fibre and several cash crops has been very impressive. The success is due to intensive research in plant breeding resulting in the evolution of varieties with very high genetic yield potentials and the concurrent support from other disciplines that has provided the package of practices which help in the realization of the created potential. In all crop plants, where the breeding efforts have been intensive, the phenomenon of plateauing of yields has become apparent. Wheat provides, perhaps, the best example of the situation. Since the 1970s, when the gains from genetic restructuring of the plant type had been established, major yield increases have not occurred. Since the demands of the future will be undoubtedly much larger, and additional cultivable land will not be available, ways will have to be found for a sizable increase in productivity per unit area in the immediate future. This requirement calls for the development of strategies and procedures that will make dramatic yield gains possible. For meeting the challenge, a continuous search is being made for innovative methods which will give yield increments of a much higher order than those provided by the conventional approach. In this chapter, an attempt has been made to introduce some such promising approaches.

Genetic resources deal with germplasm of a given species that provides the necessary genetic variation for the improvement of that species in terms of particular products, increased efficiency and higher levels of productivity. Genetic resources are, in this sense, the building blocks, fundamental not only to a crop improvement programme, but also for the very survival of the species in time and space (Swaminathan, 1983).

As agriculture is becoming more and more intensive and location-specific, plant breeding objectives are becoming more and more complex. Besides high yield, a modern variety is often expected to combine specific duration, nutritional attributes, adaptation to varying soil and water regimes, and resistance to ever-multiplying pests and diseases—to mention only a few of the character combinations. To meet these objectives, a wide range of germplasm must exist, and the plant breeder should have an easy access to the genetic resources.

Since genetic variability is essential for any crop improvement programme, the creation and management of genetic variability becomes central to crop breeding. Experimentally induced mutations provide an important source of variability. An advantage of such created variability is that the starting point can be an agronomically accepted cultivar rather than a genetic stock of no direct commercial value with many undesirable attributes. It should, however, be realized that none of the imposed treatments in a plant system has so far been able to direct a specific change. As a result, following mutagenic treatments, a mixed bag of induced variants is found and many of these may not be of any value. Also, many events take place concurrently in a treated cell to produce variation at several loci within a genome. Because of these reasons it will be unrealistic to expect miracles out of a mutation breeding programme. The desired results can be achieved only when the objectives are clearly defined and the experiments are specifically designed to achieve those objectives. When a mutant of the required type is obtained, we only reach the step of identification of a desired recombinant in a conventional breeding programme and all the succeeding steps of evaluation of the mutant are in no way different from those in the conventional breeding procedures.

Breeding methods for vegetable crops are basically similar to those adopted in others like cereals, oilseeds or pulses. Certain aspects, peculiar to vegetables, however, require specific attention: (i) In many cases, the edible part of the plant is not the seed (as cauliflower, cabbage, carrot and onion) and hence while breeding for improved plant part that is used as vegetable, its seed production capacity needs due consideration, (ii) Vegetables like tomato, garden pea and onion are consumed in different forms (fresh or processed) and the end use of the product (like processing quality) has to be taken into account in breeding programmes, (iii) Consumer preferences (as for example in eggplant) size, colour and shape of the vegetable has to be given due emphasis, (iv) Being perishable in nature, keeping quality and suitability for long distance transportation are important attributes in vegetable crops unlike grain crops where bulk handling and haulage can be mechanised, (v) Another peculiarity with vegetable crops is that they vary widely in modes of reproduction necessitating appropriate choice of breeding methods. In addition to vegetatively propagated crops like potato, sweet potato, cassava (tapioca), garlic, pointed gourd or “Parwal (Trichosanthes dioica L.) and ivy gourd or ‘Kundru’ (Coccinia cordifolia L.), there are many seed propagated vegetable crops which are auto- or allogamous or intermediate (Table 1).

Within these broad groups, variations in the extent of cross- or self-pollination, namely, 15-20 per cent natural cross-pollination in chillies and 20-25 per cent natural self-pollination in muskmelon have been recorded.

Forage crops are plant species used to feed domestic animals (such as cattle, sheep, goat, pig, and poultry). Both cultivated and wild plants can serve as forage. The domestic animals feed on forages either by grazing or stall feeding (cut fodder eaten in stalls).

Grazing, that is pasturing of animals, is cheaper and allows production of both livestock and dairy produce at a lower cost because no expense is involved in growing, cutting, transporting or feeding. In North and South America, Australia, New Zealand, and in many other temperate countries pastures are widespread and popular because sufficient land is available. While land is cheap and plentiful, labour is scarce and expensive; hence management of natural, and in recent years sown pastures has received much attention.

Until about fifty years ago, the natural grasslands in the Western countries met the minimum needs of the farmers and grassland management was operated at a level comparable to subsistence agriculture of many developing countries. In other words, emphasis was laid on exploitation of natural productivity of the grassland, using a low level of management based on minimum use of inputs like chemical fertilizers. However, it was soon realized that productivity of native grasslands will have to be greatly improved if the high demands for meat and milk were to be adequately met.

In India, however, out of a total cultivable area of about 131.11 million ha less than 2 per cent is utilized for forage crops because feeding the large human population takes the first preference for land use to produce grain and cash crops. It is, therefore, necessary to intensify efforts, particularly in over-populated countries, for a more efficient production of high yielding forage crops to get maximum return out of minimum land.

Introduction

Tree improvement through the application of genetic principles is directed towards modifying the heredity of tree populations so that they are able to meet the needs of the forester and the wood based industry. The breeder of annual crop plants resorts to hybridisation, selection, testing of the improved strains, multiplication of improved seeds and finally their release for commercial planting in that logical sequence. But the tree breeder, whose career may be shorter than the life cycle of the tree species taken up for improvement, tends to carry out the above operations in a modified form and often some of the operations are carried out concurrently.

Many of our tree species are wild unselected populations. They are, therefore, expected to exhibit considerable natural variability for many characters and also to respond well to selection for further improvement. Further, altitudinal and latitudinal variation in many characters between populations are expected for those tree species which have a wide geographical distribution. So, the selection of desirable seed sources or provenances is an essential step that is basic to all tree improvement programmes. The right provenance is identified through well laid out provenance trials at a number of locations in the area where the improved selection is to be grown. Further improvement is then planned and executed for exploiting the naturally available intrapopulation variation. The variation normally measured is the phenotypic variation which is the result of interaction of the genotype of the individual with the environment. However, what is essential for selection to be effective is genetic variation for the character under consideration. Genetic variation is assessed by appropriate progeny trials. Since many of the economically valuable characters of tree species are polygenically controlled and exbibit continuous variation, measurements have to be in metric units. The use of biometrical techniques for assessing and analysing this variation, therefore, becomes an essential requirement.

Most varieties of fruit plants have originated as chance seedlings or through somatic mutation and have been maintained and multiplied through vegetative propagation by fruit growers. The improvement of fruit plants through breeding has not yet attracted the serious attention of the research workers. Unlike field crops, fruit plant breeding is time consuming and laborious. Studies on the mode of inheritance of important economic characters of fruit plants are difficult because of the heterozygous nature of the plants. Moreover, varying degrees of sexual incompatibility have complicated the situation. Due to a long juvenile phase, it takes a long time to identify a useful hybrid. This factor has become a serious handicap because most research workers leave or shift from the programme before tangible results are obtained.

During the last two decades, considerable useful information has been generated on breeding methodology for the improvement of some of the important fruit crops. In this chapter fruits like mango, banana and papaya have been included.

The immediate and long-term objective of modern plant breeding in all countries irrespective of the state of their advancement is increasing the availability of food to meet the calorie-protein requirement of their people. In the past, breeders have concentrated their efforts on increasing the yield of the important crop species. Better adaptation to local environmental conditions, high production and better resistance to the major pests and diseases have been among the priority goals of genetic plant improvement. Yielding ability and stability undoubtedly deserve high priority but enough attention also needs to be paid to improving nutritional characteristics of the plant produce for their maximum utilisation. Awareness of nutritional requirements and the optimal characteristics of crop products to be utilised as food and feed, therefore, becomes an important parameter of breeding effort.

The main source of food in the developing world is cereals which provides 45 to 85 per cent of the total calories and 50 to 80 per cent of the total protein requirements. The consumption of animal proteins, which are nutritionally well balanced, is low and costly and beyond the reach of the low income groups. On the other hand, plant proteins are cheaper to produce and easier to store and transport. All these factors go in favour of increasing the quantity and quality of cereal proteins in order to improve the dietary and nutritional status of the masses in the developing world.

Proteins are made up of about twenty different amino acids. For man, eight of these amino acids, valine, methionine, leucine, isoleucine, threonine, phenylalanine, lysine and tryptophan, are essential (EAA) for constituting a well balanced diet. In impioving the grain quality of cereals consideration should, therefore, be given apart from the content of carbohydrates, fats, minerals and vitamins in the grains, to improving the protein content as also the amino acid composition of the protein.

Plant diseases, in spite of the various measures adopted to control them, continue to be a major cause of crop losses. At times these losses have been so heavy that they have caused severe famines. Diseases are caused by a wide variety of organisms such as fungi, bacteria, viruses, mycoplasma and nematodes. The diseases may be partially or completely controlled by chemical, cultural and genetic methods. When no single method is effective in controlling a disease, integrated control constituting a combination of the three methods of disease control (chemical, cultural, genetic) may be effective. If host resistance alone is sufficient to check a disease, then it is to be preferred to other methods because it is the least expensive, has no adverse environmental effects and involves no specific action on the part of the farmer, even when only partial, genetic resistance supplements chemical control and makes it more effective in comparison with chemical control on a completely susceptible cultivar.

Many plant diseases have been successfully controlled by exploiting genetic resistance especially where differentiation in the pathogen with respect to resistance genes in the host is lacking. In some cases, even when pathotype differentiation occurs in the pathogen, genetic resistance in combination with other methods has given adequate control. Breeding for resistance to highly variable air-borne fungal pathogens has claimed much time and effort of plant breeders and success has often been transient. The reasons for the shortlived success have become clear from research during the last quarter of a century.