APPLICATION OF SATELLITE BASED REMOTE SENSING TO THE ESTIMATION AND MONITORING OF CROP HEALTH

ABSTRACT

The effect of loss in the availability of farm products has a lot of negative impact on the society. The decrease in crops production has created disparity between the food demand of world population and the global agricultural output. Crop production faces alot of challenges, some of which includes water scarcity, bad soils, unsuitable temperatures, pests, diseases and weeds which attacks the crops. Ground based or manual agricultural approach to detecting pest invasion and rapidly curbing it is not only time consuming and laborious, it is not also a real-time option especially for large scale farmlands.   Remote sensing provides a rapid, intrusive and a more viable option for the collection and analysis of spectral properties of earth surfaces from various distances, ranging from satellites to ground-based policy. This study is aimed at assessing the applicability of remote sensing instrumentation and its techniques in evaluating and estimating crop health. This is to boycott the long process, time consuming and expensive biological laboratory tests always carried out by agricultural scientists to estimate the same attribute of crops. Sentinel-2A images and Landsat 8 images were acquired for use in this study. The total area covered for this research is101 Hectares. These images were preprocessed using ArcGIS software in order to remove effects of atmospheric conditions on the reflectance properties of the image channels. The images were processed to produce several representations of vegetation indices, soil indices and tasseled cap indices. Correlation, regression and analysis of variance (ANOVA) statistical tools were employed to assess the agreement of these vegetation and soil indices and their correlation with that of the tasseled cap indices. On the other hand, laboratory tests were performed to assess sampled crops and estimate their health status. A PCA model was developed to convert the  laboratory test  results  to  an  equivalence  of  the  remote  sensing  NDVI,  a  de-facto vegetation index for assessing crop health. Statistical analysis was performed to evaluate the relationship between the outputs from the remote sensing approach and the agricultural approach of crop health estimation, and the result shows a very weak correlation (0.16) between the two techniques. This implies that on general consideration of crop health estimation, there are just 1.6% (approximately negligible) chances that the result from the remote sensing technique will give equivalence to that of the laboratory result in agriculture. Statistical and graphical analysis performed based on each crop species reveal that cassava gave a 48.8% similarity with that of laboratory, 50.2% for groundnut and 63.8% for maize similarity respectively for two techniques, for rice health status, the study found out that the remote sensing technique could give 23.9% similar to that of laboratory. This makes it unreliable for such approach to be used in estimating maize health analysis. The result also found out that when satellite images are employed for estimating soya beans health, the output will be negatively correlated with that of the laboratory. For yam specie, the results show that there is no any correlation between the results from the laboratory and that of the remote sensing (0.021). Therefore, attention should be focused on the northern region of the study area for cultivation. Also, result obtained from this study should be further validated in order to establish a more valid PCA model for the study variables.

CHAPTER ONE

1.0    INTRODUCTION

1.1      Background to the Study

The relevance of food production has been recognized by Sustainable Development Goals (SDGs) and this is as a result of the strong relationship that exists between food and insecurity. A lot of measures have been put in place to curb the challenges of low food production, especially in developing countries. Furthermore, despite the fact that agricultural production has been decreasing in most developing countries, most of these developing countries rely so much on agriculture for survival and as sources of income (Adesugba and Mavrotas, 2016). This underscores the importance of the development of automated technologies, such as satellite based remote sensing, for estimating and monitoring crops health and for the quick identification of pest invasion on plants, before they wreck and cause irreversible havoc on the plants thus, negatively influencing the rate of food production.

Standard means for estimating and monitoring crop health and identifying crops diseases are major issues in agricultural sectors (Savary et al., 2015). Most farmers previously rely on manual checking of their crops for signs or symptoms that are detectable within human limitation, this process of disease identification leans on the variety of crop, size and area where the crops were planted, and in the case for most commercial farms, the farm land are habitually very large, this approach of checking crop condition is prolonged and very stressful (Oerke, 2006).Manual inspection strongly depends on the categories of infection or stress showing clearly detectable symptoms, which in the mid age regularly occurs to latter stages of infection. In order to determine the causal agent of such ailment in crop, it is achievable through either manual assessment or laboratory examination. Due to advancement in the agricultural practice, which is a major cause why the largely manual process needs to be replaced with a more sophisticated, precise, and sensitive approaches (Mahlein, 2012).

Remote sensing is the science (with some approaches of art) of obtaining, without being in contact in the actual sense, Earth’s surface information. This is achieved through the sensing and recording of energy that is emitted or reflected and processing, analyzing, and applying that information. Remote sensing procedure encompasses interplay of interest target or object and radiation that are incident (Archana, 2015). The range of EMS commences from the shorter wavelengths (gamma and x-rays inclusive) to the longer wavelengths (radio waves for broadcasting  and  microwaves). Remote sensing solely depends  on  electromagnetic spectrum. The human eye is capable of detecting wavelength approximately ranging from 0.4 to 0.7μm. Chlorophyll is regarded as the most essential organic molecules present on Earth. Photosynthesis makes use of these molecules as its elemental pigments. Vicissitudes in the absolute consistency of foliar chlorophyll and the comparative proportions of chlorophyll ‘a’ and ‘b’ are effectuated by a multiplicity of biological stresses, leaf evolution and senescence. Such pigment adaptation interacts directly to the ratio of initial production. Furthermore, the chlorophylls proportionately contain a large measure of total leaf nitrogen. Consequently, chlorophyll concentration assessment may indirectly impart a definite result in the determination of plant nutrient. From a scientific view, the information pertaining to spatial and temporal changes of the chlorophyll properties of leaf is of a substantial relevance; especially carry out an investigative process of plant–environmental interactions, and from an applicable perception of agriculture, environmental management and forestry through the use of maps (George, 2008).

Application  of mapping and  satellite based  remote sensing in  crop production  have a significant impact in modern practice agriculture also to combat the ecological problems traceable to climatic variation amongst other non-climatic factors. Classification of crops is a source of vital information required in making diverse decisions needed for agricultural resources management. Processing of satellite imageries is capable to render the users, propitious and precise knowledge on crop class and genuine evaluation of crop yield applying sophisticated classification methods. Decision on the satellite imagery for grouping and segmentation of crops is dependent on criteria such as economic constrain, availability, crop type variability, and size of the area of study.

Open source data like Moderate Resolution Imaging Spectrora diameter (MODIS), Sentinel and Landsat images have been used in some applications for mapping and analysis regarding vegetation or vegetative zones (Zheng et al., 2015). Mix-pixel is a well-known challenge which is often encountered in MODIS due to poor coverage (250–500 m). Thus, land sat images (30m) when compared to MODIS especially for areas with small agricultural lands, provides better and more accurate results. The Sentinel-2 is a European Satellite which yields a multispectral image at average spatial resolution of 10m and good temporal resolution (five- days), and which can reduce the side-effects of its relatively rough spatial resolution (Drusch, 2012). The Sentinel-2 satellite Multi-Spectral Instrument (MSI) has about thirteen (13) bands and three spatial resolutions. The newly launched Sentinel-2 images served several purposes in the remote sensing environment (Whyte, 2018). Such as, crop health estimation, mapping changes in land cover, forest monitoring and for providing information on pollution of lakes etc. The invention of unmanned aerial vehicles such as DJI Phantom which is simple to operate has immensely contributed to the affordability of precise agriculture. The system of unmanned aerial vehicles (UAV) generally called drones, can be mounted with hyperspectral/visibility sensors to record countless images of an area that can be systematically processed adopting photogrammetric procedures to construct and orthophoto and NDVI maps (Chris, 2014).

NDVI can be described as a binary measure that makes use of the visible and near-infrared bands of the EMS, and it is often used to carry out analysis on remote sensing observations and detect whether the observed target contains green (living) vegetation or not (Gitelson et al., 1996). NDVI has been applied to several studies on vegetation, been implemented to check crop fertility, performance level of pasture, and rangeland carrying capacities among others. NDVI values are associated with vegetation, NDVI values varies between -1 to +1. Values that exceeds +0.1 represent vegetation and values which are closer to 0 represent naked soil and rock while negative values represent water, clouds and snow. NDVI value positive increase means that there is vegetation. Vegetation indices are utilized as a surrogate to estimate vegetation activity. Hence, NDVI is a much reliable component that can be used to estimate crop health status.

The Tasseled Cap transformation was developed to map and evaluate urban and vegetation changes detected by different satellites; it converts the readings from group of frequencies into composed values. The weights are assigned to separate frequencies and the weighted sum of each frequency was taken. The light intensity of a single pixel in the scene is measured as the weighted sums. Distinct composed values are the linear combination of individual frequencies. Some of these weights are negative and some are positive. Three bands are commonly  used  in  tasseled  cap  transformation-based  analysis.  Band  one  which  is correspondence with image “brightness” gives a measure of soil brightness that used to develop brightness index. Band two is correspondence with “greenness “or photo synthetically-active vegetation to derive greenness index. The third tasseled-cap band is usually construed as the wetness index in which describes the interdependence of soil and canopy moisture. (Riffi and Fizazi 2012).

1.2      Statement of Research Problem

The reliability of detecting and identifying of the plant health and plant stress are a prevailing challenge in agriculture. Ideal approaches of detecting plant diseases or infections often rely on manual checking of crops for visible indicators of the presence of such diseases by agronomists.

According to the crop variability and the extent of the crop area, which for multitude of commercialized farms is tremendous in many cases, the standard methodology has proved to be time demanding, laborious and expensive. The manual process of disclosure is dependent upon the disease or stress clearly exhibit evident visible symptoms, that recurrently manifest from the middle stage to late stage of infection, which makes quick and adequate intervention difficult  and  sometimes,  impossible.  This  is  why there  is  a  perceptive  interest  in  the agricultural sector to provide alternative optimized approaches for real-time or near real time monitoring of crop growth and health. Also, it is needful today because of the trending precision agriculture which comprises mostly GPS and other geo-informatics techniques.

1.3      Aim and Objectives of the Study

The aim of this research is to estimate crop health using Remote Sensing approach with a view to enhancing crop yield in Garatu village

In the interest of achieving the stated aim, the objectives of this research are to:

1.   Extract vegetative information from  sentinel 2 images

2.   Evaluate crops health from the extracted vegetative information.

3.   Carry out tissue test for the validation of the extracted crop health information.

4.   Carry out (T.test and Pearson correlation) analysis for the results obtained from

Objectives (2) and (3).

1.4      Research Questions

1.   How best can vegetation indices be extracted from satellite image data

2.   How  correlated  are  the  result  obtained  from  satellite  images  and  crop  health estimation

3.   How best can you validate result obtained on crops health from satellite images

4.   Is there any statistically significant relationship between vegetation index gotten from satellite images and tissue test result obtained from the lab relationship between.

1.5      Justification of the Study

Nigeria being a fast-developing nation, having an approximate population of over 200 million citizens relies heavily on agriculture production, so as to avoid risk of food crisis. A lot of farmers in Nigeria face different challenges in monitoring their farm produce and have very limited chances of getting good farm yield; this is as a result of manually checking of crop disease and monitoring of crop health, which is both intensive and demanding. Identifying the causal agent is achieved by detecting manually or diagnostic tests. The implementation of remote sensing can be used to curtail the stress and challenges that farmers majorly face in the process of detecting diseases and stress of crops at early stage. These precise procedures can engender a diminished rate of pesticide and herbicide utilization, as well as impacts that are subsequently beneficial for the grower finances, ecosystem services, for the environment, and the end consumer. The introduction of precision farming, using remote sensing approach will help farmers understand how to study their crop growth by providing real time monitoring approach, with very little understanding of the GIS approaches. Hence, this research will help resolving some of the challenges faced by farmers in monitoring their crop health.

1.6      Study Area

The investigations were performed as a case study of Garatu village, located within geographical coordinate (214292mE, 1045668mN), (214781mE 1045819N), (214935mE 1045517mN) and(214579mE 1045259mN), under Bosso LGA area is situated at about 19.49km away from F.U.T Minna permanent site (Gidan Kwanu campus). Economically, agriculture plays a substantial role in the area, most people who reside in the area are farmers whom solely rely on agriculture for survival, several crops such as yam, maize, cassava, groundnut, soya beans, and rice, are being planted on the specified area of interest. Figure 1 shows the exact geographical representation of the study area. For this study area, Sentinel 2 images were obtained from European Space Agency (ESA) (www.copernicus.datahub.com). The total area covered for this research is101 Hectares.

₦8,000.00 – DOWNLOAD FULL PROJECT DOWNLOAD FULL PROJECT Added to cart

---

---

---

---

Related Project Topics :