Analysis of Forest Vegetal Characteristics of Akure Forest Reserve from Optical Imageries and Unmanned Aerial Vehicle Data

Forest vegetal characteristics monitoring has a long tradition records with a success rate ranging from low to medium or high depends on the  application at the hands. Details information about the indication of association of phenomena as forest indicators, such forest gap, estate and forest status, provides high spatial resolution images. The aim of this study focuses on combining unmanned Aerial Vehicles (UAVs) and satellite multispectral imaging along side by side to details forest parameter during the seasons. UAVs image at 0.15m appeared more detailed of having features such as rock, road, bare ground, riparian trees among others than that of Landsat OLI image, though the features such as rock, road, bare ground, and riparian forest were also seen on the image but it was poorly seen due to the coarse spatial resolution of 30 m. The 3-Dimensional of UAVs, relief pattern and contour from Shuttle Radar Topography Mission was also compared and this study further demonstrated on the advantages of Unmanned Aerial Vehicle data over established remotely sensed data which includes flying blow the cloud, high spatial resolution, flexibility, inexpensive of data acquisition, time effective, using video footage to detect human activities such as tree flora, burning and logging.


I. INTRODUCTION
Early work on forest plantation in Nigeria commences at the beginning of the 20th century especially in the southwest which practically involved on the economical important indigenous tree species [7], ever since Nigeria settlement after independent, over half of nation`s forests and woodlands have been progressively cleared subsistence agriculture [17].Despite recognition of the factors associated with their clearance, deforestation rates accelerated in the uncontrollable manner [17].Much of the clearance in south-west occurred in the more productive forest ecosystems.Reference [2]  which earmarked at the beginning of the 20th century.References [11] and [19] estimated 285 hectares as the average annual rate of deforestation in Nigeria between 1976 and1980, increasing into an estimated 400 hectares by the year 2000.Reference [9] reported Nigeria has lost 55.7% of its primary forest to logging, subsistence agriculture, collection of fuel wood and other agents between 2000 and 2005.
The same patterns had been experienced in the tropics and sub-tropics Africa.For instance, the East African region lost about 10% of its forest cover to deforestation between 1990 and 2000, with Uganda recorded the highest rate [8].In the humid tropical rainforest region of Cameroon, about 200,000 hectares of forest reported to be degraded annually due to high rate of exploitation.Such clearance has been observed and documented from almost of four decades through land cover change detection based on Landsat-1-4 MSS, Landsat-5 Thematic Mapper (TM), Landsat-7 Enhanced TM (ETM+) and Landsat -8 OLI data and has resulted in extensive losses of forest and such discoveries in assessing deforestation has generated a lot of questions on the validity of data.Among the technical issues in question, the most challenging is that there was no consensus in the literature on the rate of deforestation in most of existing forest reserves globally and regionally often because of coarse resolution of the optical remote sensors.
Recently, technologies such as GPS, miniaturized drones (UAVs) were initially developed for military use, but are increasingly being deployed in civilian applications including mapping, monitoring and managing habitats and natural resources [14].Although miniaturized drones are not used widely in environmental applications yet, their use is likely to increase rapidly as their prices decrease and the technology becomes easier to use [5].Although [12] cited [16], [5] in their reports that some initial attempts were made to employ small drones in environmental research in the 1990s and early 2000s, researchers have begun serious investigation on the use of drones over the last seven to eight years.The development of environmental remote sensing technologies and aerial drone has been closely related to the study of forests [10].Although, the bulk of academic research into the use of miniaturized drones has been greatly geared toward precision agriculture [21] and [18].

Methodology employs various techniques and approaches
to integrate this study.Such techniques and approaches focus on data acquisition, data processing and data presentation.It commences with database design, conceptual, logical and schema.[4] describes conceptual design as a process by which real world entities and their relationships are modeled to achieve maximum output while utilizing minimum amount of data.The views of reality in this study were Roads, rivers, settlements and forest.Vector and raster data model were employed to represent the spatial entities as points, lines and polygons [13] these were translated to logical and physical design which represented data model designed to reflect the recording of the data to be computerized using relational database management system (RDBMS) and digital image processing involved.Two major sources of data were used in this study; namely, primary data: x,y prominent settlements, imageries from Unmanned Aerial Vehicles and secondary data: Shuttle Radar Topography Mission (SRTM) 30 m spatial resolution(2017) and Landsat 2017 OLI.

A. Data Processing
This process involves restructuring the available data and creating sequence order of proceeding or cartographic model required for data analyses.Basically, raster and vector models are usually involved and they were employed.
B. Primary data X, Y locational coordinates of prominent settlements in the Akure Forest Reserve (AFR) were captured as points with the Garmin eTrex20 GPS device.The dilution of precision (DoP), geometric dilution of precision (GDoP) and datum was set to zone 31 North Hemisphere 1984.After the setting, the GPS was allowed to resolve and connect to at least minimum of four satellites before the data capture for the settlements.We gridded the Topographical map and coordinates were obtained from the edges of the map and coordinates were pre-loaded into Quadcopter drone through the designed path i.e. traverse from the origin to destination.300 m altitude was chosen to fly due to trees obstructions for the drone when it moves around, the drone speed was set at 3m/s, and 16.1 mega pixels integrated camera was onboard for the field of view (FOV) of 28.940 look angles.Images were captured in panchromatic mode of (RGB) with shutter capture speed at 1/1000s.It covered 2.463 km2 / 246.284 ha / 608.897 acres Also UAVs imageries was processed through the drone2map software by the conversion of flight lines and points into points, clouds, Poisson surface reconstruction, Ortho generate Digital Surface Model, and orthomosaic of 2-dimension (2D) and 3-dimension (3D).Then after the stacking of the image, it was imported into ArcGIS environment through add data tool on the ArcGIS interface.

C. Secondary Data
The Digital Image Processing (DIP) Techniques is necessitated by having imageries in digital format.Landsat 8 OLI-TIRS (2017) were sourced through Path 190/Row 055 and downloaded from GLCF/USGS in digital format into the computer via earth explorer window and then it was imported into ERDAS Imagine 9.2 version through classic viewer.Landsat 8, 2017 OLI-TIRS alone was sourced for the sensor that acquired image on the 23rd of March 2017 downloaded because it was exactly the times that drone image was acquired.The images noise was filtered through radiometric enhancement.The ground truthing, visual image interpretation and digital image processing were combined to Layer stacking, Sub-Map Creation of Raster Data, and ground truth were verified and Shuttle Radar Topography Mission was also employed by using the Filter tool on the ArcGIS 10.3 version to filter away the redundant data and then the creation of Triangulated Irregular Network (TIN) and contour of the study area.

D. Physical Design and Database Creation Phases
This phase is known as the implementation stage.It involves the representation of the data structure in the format of the implementation software and two implementation software involved.The spatial database for the study area was created in ArcGIS 10.3 while digital Database management systems involved data security, data integrity measures and database maintenance [4].Data security involves the measures adopted while designing the database using necessary backup or fitness model for the data from being lost.In ensuring data integrity, inconsistency between two features must be done away with in checking the correctness of the records in database and finally database maintenance checks on the quality and the fitness of the database.

A. Approach of Complementary Use of the Optical Remote Sensing Imageries and UAVs Technology
Interpretation of satellite imagery was a method of obtaining information about objects and the landscape.This has been extensively used by [20] in context of studying the geographical reality which based on the detection, identification and spatial localization of individual objects and terrain shapes captured in a satellite image records.Interpreting the image represented the deciphering of its multifaceted content from the point of view of the purpose it serves.The information that we are looking for in the images encoded in various shades and textures.The interpretation of digital images is basically possible in two ways, usually referred to as visual interpretation and computer interpretation.But visual interpretation is employed in this study to further the analyses

B. Comparism of UAVs Orthomosaic Image and Landsat of the Selected Parts of Akure Forest Reserve
The Comparism of optical remotely sensed image and orthomosaic image acquired from the drone is presented in the (Figure 2) and the comparison was based on the image characteristics which included: spatial, radiometric, spectral and temporal resolution.Both the two images were acquired of the same area, of the same month but with different techniques in acquiring them.They were arranged side by side for the visual interpretation because there was strongly difference in their spatial resolution.The area covered was 246.284 hectares of the study and Landsat 2017 was also masked by the exact the same boundary and it was covered the same area of 246.284 hectares.UAVs image appeared of having detailed information such as rock, road, bare ground, riparian trees among others were manifested on the image while in Landsat 2017 image the features such as rock, road, bare ground, and riparian forest were also seen on the image but it was poorly seen.They were arranged in the pixels order and these features were seen clearer in the drone image than the Landsat image due to the coarse spatial resolution.This evidence has shown as an added advantage over Landsat OLI, 2017 image because drone image was captured of 0.15m spatial resolution due to the fact it can drive at any altitude for capturing while Landsat image was 30-meter spatial resolution.The relevant entities such as bare ground, riparian forest among others were not seen clearly as it was showed on the UAVs image due to poor spatial resolution but what found on the Landsat image was patches of the forest.

C. Comparism of 3-Dimensional of UAVs, Relief Pattern and Contour from Shuttle Radar Topography Mission
The analyses below presented 3-Dimension orthomosaic of 0.15m from UAVs.This showed the capability of identifying the forest estate, tree stand, rocky, bare ground at aerial view while the Landsat ancillary data cannot portray the aerial view.This result has contributed tremendously as an added advantage over the Landsat data due to the coarse resolution and the relevance of Landsat data was only traced back to the it historical time of existence.In Figure 3  Based on the results, it was assumed that UAVs and satellite monitoring document reliable information about forest vegetal characteristics.The wider coverage, historical and readily availability of Landsat had been a credible in the context of making it useful in the application of forest but it was not usual possible to engaging the optical remote sensing imageries to monitor instant phenomena due to the coarse spatial resolution.It usual uses spectral signature to interact with phenomena under investigation which includes correct recognition and classification of individual objects, determining their properties, quantitative and qualitative characteristics, accurate spatial (positional) location of the detected objects, examination and evaluation of interrelationships and causalities between the displayed phenomena and identifying patterns characterizing having been crucial ingredients of this data.While thus, this study has demonstrated on the advantages of Unmanned Aerial Vehicle data over established remotely sensed data which includes flying blow the cloud, high spatial resolution, flexibility, inexpensive of data acquisition, time effective, using video footage to detect human activities such as tree flora, burning and logging.The main difference of their utilization is coming from their specification and technical limits.Satellite survey can be used for periodic monitoring of forest as the indicator of their spatial heterogeneity within fields, but with low resolution 30 m per pixel for OLI compared to the 0.15 m per pixel of UAVs imaging.On the other hand, UAVs represents a special campaign focuses on the mapping of high-detailed spatial inputs by deploying drone on the site at interval of time and for the proper monitoring.

Fig. 1 :
Fig. 1: Map of the study area (Source: Author, 2019 Map of Study area)

Fig. 2 :
Fig. 2: Subset images from UAVs and Landsat of part of Akure forest reserve (a) below the contour map display values change across the surface where there is little change in value, the lines are space further apart, where values rise or fall, the lines are closer together.Flat and steep distance between contour and ridges, hill and valleys (converging or diverging polylines; in Figure3(b)indicates the terrain relief pattern and graduate the surface with values and colour and shows area that is steepness, low and high while In Figure3(c) the digital surfaces model (DSM) aimed of supplying the elevation information and the display information was colour ramps where light black shows rise or fall area while dark black shows low area and this familiar with the work conducted by[6], and it also uses various colours for distinct description where colour blue and red indicates lower area, green colour indicates hilly, yellow colour indicates high while pale red indicates also the high.

Fig. 3 :
Fig. 3: Subset of contour, terrain relief pattern and digital surface model revealed the remaining status of the tropical rainforest in Nigeria at only 10% of tropical rainforest area as against 25% tropical rainforest I. A. Idoko is with Survey and Geoinformatics Department Federal School of Surveying, P. M. B. 1024, Oyo State, Nigeria (e-mail: idokoisaaca@yahoo.com).M. O. Okegbola is with Survey and Geoinformatics Department Federal School of Surveying, P. M. B. 1024, Oyo State, Nigeria (e-mail: digitalspace2012@gmail.com).L.O.Oyelakin is with Survey and Geoinformatics Department Federal School of Surveying, P. M. B. 1024, Oyo State, Nigeria (e-mail: olabiyilatifat@gmail.com).