Analysis of Biomass Change Using Remote Sensing Technique

Introduction

A big part of the population in Rwanda lays in farming communities where agriculture is the main source of income and livelihood hood. This may lead to severe land degradation due agriculture activities and the high demand for firewood. Apart from that, large number of parcels of land are cleared.

Biomass plays an important role by providing different ecosystem services which may help to adapt and mitigate the global climate change. Spectral vegetation index data are used to investigate the relationship between climate and vegetation at the landscapes level, to assist land management and sustainable utilization of forest and other vegetation resources and also to investigate climate change impacts and carbon sequestration by different vegetation types (Nurhussen, 2016). Currently, the global vegetation cover is decreased due to human induced activities mainly through deforestation for other different land uses in need.

Remote sensing is broadly defined as a science of collecting and interpreting information about a target without being in physical contact with the target (Sabins, 1997). The remote sensing process mainly consist in the analysis and interpretation of data collected by a sensor. Remote sensing approaches provide useful information about the subject under investigation from different perspective of view , it include diverse techniques , ranging from traditional methods of visual interpretation to digital information extraction methods using sophisticated computer process.

This study aimed to analyze the biomass change seasonally using satellites remote sensed data, Musanze District, Rwanda, with following specific objectives; to Map existing and current status of biomass in Kimonyi sector, to produce NDVI maps and biomass class’s trends across the study area, to seasonally assess vegetation condition, to seasonally detect the changes in vegetation biomass. Therefore, the study reveals that Remote Sensing is powerful tools to monitor biomass change in a given period, hence directly contribute to any planning activities especially in environment degradation and agriculture domains.

Methodology

Kimonyi Sector is one of fifteen Sectors of Musanze District in Northern Province of Rwanda. Kimonyi sector has 4 cells including Birira, Buramira, Mbizi and KivumuI. It has fifteen thousand, five hundred and eighty-nine (15,589) of population (NISR, 2012). Kimonyi sector relay in volcanic plain part with an average altitude of 1860 m and it has an area of 21, 60 km2 (Luis & Byizigiro, 2012). There are four seasons in the study area, namely long rain season that starts from march to May, long dry season which starts from June to mid-September, short rain season which starts from October to November, and lastly short rain season which starts from December to February.

To perform this study, four Landsat 8 OLI/TIRS images of May 2016, august 2016, November 2016, and January 2017 were downloaded from USGS web platform (www.earthexplorer.usgs.gov/.). Landsat satellites sensors provides data in 11 spectral bands with 30m spatial resolution for multispectral band and 15m for panchromatic band.

Image Pre-processing and Analysis

The geometric image registration was performed in order to minimize all geometric distortions inherent to the image. The Land sat L8 OLI images was registered to a common Universal Transverse Mercator (UTM) projection, 35 Zone with WSG84 as Datum, Thereby removing a large amount of the geometric errors in the raw data.

This step was performed to combine separated bands into one multispectral image. The combined bands for Landsat 8 OLI are band2, band3, band4, band5, band6, and band7(Barsi et al., 2014).This allow research to extract and analyze vegetation cover using Landsat imagery, bands 3,4 and 5 are most suitable for this analysis as they combine the most important spectral reflectance aspects of vegetation

Resolution merge was performed, where multispectral bands were combined with panchromatic band in order to get an image of 15 meter resolution, to enhance and increase the visibility of images (Johnson et al., 2012).

The Landsat tile is much larger than a project study area. In this case it is beneficial to subset the downloaded image to remain with the area of interest only (JARS, 1993). The study area shapefile was re-projected to be given theprojection similar to the one of satellite image. Then it was used to subset that satellite image using Erdas imagine 2014 software.

Image interpretation is the process of manually and digitally examining a digital remote sensing image to extract necessary information or to identify features in that image. Image characteristics (also called image attributes) are made of seven elements that are used to derive information about objects in an image

Normalized difference vegetation index (NDVI) was computed using the model maker tool in ERDAS Imagine, this is normally the ratio between measured reflectivity in the red (band4) and near infrared (band5) portions of the electromagnetic spectrum. NDVI values range from -1 to 1. NDVI was computed using the following formula (Richardson & Everitt, 1992):

NDVI= (NIR(band5)-RED(band4))/(NIR(band5)+ RED(band4) (1)

VCI quantifies the weather component. The weather-related NDVI envelope is linearly scaled to 0 for minimum NDVI and 100 for the maximum for each grid cell and week (Parrinaz et al., 2008). It is defined asVCI = (NDVI-NDVImin)/(NDVIMAX-NDVI Min) (2)

Where NDVI max and NDVI min are the maximum and minimum value of that NDVI image. VCI changes from 0 to 100, corresponding to changes in vegetation condition from-to extremely unfavorable to optimal.

Calculated NDVI was reclassified in ArcGIS 10.3 reclassification is preformed to assign values of preference, sensitivity, priority, or some similar criteria to a raster. For the present study NDVI was reclassified into three main classes.