Evaluating spatial and temporal variations of aerosol optical depth climate factors human activities and biomass burning over Southeast Asia using satellite data

Abstract

Major cities in Southeast Asia (SEA) are faced with severe air quality problems including dust smog and haze pollution which are mainly caused by atmospheric aerosols (smoke) from biomass burning Technological advances in monitoring atmospheric aerosol and biomass burning have been fostered by a series of new space based satellite instruments and data products In this study a variety of satellite product maps of aerosol optical depth (AOD) precipitation wind city light burned area (BA) and active fire were collected and processed to evaluate the spatial and temporal variations among atmospheric aerosol climate factors human activities and biomass burning in SEA during 2002-2011 Satellite data applied in this study includes: 1) the Moderate Resolution Imaging Spectroradiometer (MODIS) derived AOD; 2) three MODIS BA products including the BA derived from vegetation change and land-cover classification (MCD45A1) the BA derived from active-fire (GFED4 0) and the combination of GFED4 0 and BA caused by small-scale fires (GFED4 0s); 3) the MODIS active fire data (MCD14ML); 4) the National Oceanic and Atmospheric Administration (NOAA) surface wind data; 5) the MODIS International Geosphere-Biosphere Programme (IGBP) classes land cover dataset (MCD12Q1); 6) the Global Precipitation Climatology Project (GPCP) monthly precipitation dataset; and 7) the DMSP-OLS nighttime light representing the strength of human activities All satellite data was converted visualized summarized and analyzed using the spatial analyst tool within ESRI ArcGIS? 10 2 To better understand the cause and effect relationships between various causative factors and atmospheric aerosols the results were organized into five sections First the spatial and temporal variations of aerosol optical depth in SEA during 2002 to 2011 were examined High aerosol areas (HAA) located in the northern and southern intertropical zone are identified respectively from the monthly AOD distribution maps The northern HAA consists of Myanmar Vietnam Laos Thailand and Cambodia with the peak AOD months are from November to March The southern HAA includes Malaysia Sumatra Java and Kalimantan with the peak AOD months are from May to October Generally the peak AOD months are consistent with the dry season in each region which provides evidence that the temporal AOD distribution in SEA is partly related to biomass burning Second the recently released BA product (GFED4 0s) shows that Myanmar has the largest annual BA in north intertropical zone followed by Cambodia and Thailand Burned areas in south intertropical zone are mainly distributed in Indonesia The peak burning months are also consistent with the dry months in each region Noted that the burning area in the northern intertropical zone is ten times higher than that found in southern intertropical zone However the level of annual average AOD in the southern HAA is very similar with that in the northern HAA It is evidence that biomass burning in peatlands results in a higher emission factor of particulate matter Third the correlations between AOD and climate factors were assessed The level of AOD is generally inversely proportional to precipitation which is partly related to less biomass burning occurring during the wet seasons The monthly average wind climatology can partly explain the large scale movement of aerosol plumes in the northern HAA during the burning months (November to next April) For the southern HAA there is no significant correlation between wind and the spatial distribution of AOD Fourth the level of AOD is generally high in urban and metropolitan areas however there is no significant temporal correlation between AOD and the strength of human activity Finally to seek a quantifiable linkage between AOD and biomass burning the study area focuses on HAAs only and different products representing biomass burning are applied Among the three BA products applied (MCD45A1 GFED4 0 and GFED4 0s) GFED4 0s considers both the BA identified by GFED4 0 and BA caused by small-scale fires and can better explain the temporal and spatial distributions of AOD in HAAs (R=0 5 and 0 85 for northern and southern HAA respectively) The correlation between commonly used MCD45A1 BA and AOD is not significant (R=0 25 and 0 58 for north and south HAA respectively) Compared to other BA or active fire products it was found that the MCD45A1 BA has the lowest correlation to AOD and it is suspected that the BA derived from vegetation-change may seriously underestimate the area of burning in SEA To better quantify the relationship between AOD and biomass burning this study develops two simple regression models for the estimation of monthly AOD from remotely sensed burning products in HAAs The regression model developed for northern HAA uses MCD14ML active fire data as the independent variable and obtained a R2 value of 0 57 The model developed for southern HAA uses GFED4 0s BA data as the independent variable and obtained a R2 value of 0 76 Generally the empirical models can explain well the temporal trends of AOD in HAAs

Date of Award	2015 Aug 28
Original language	English
Supervisor	Chih-Hua Chang (Supervisor)