A critical view of long-term AVHRR aerosol data record in China: Retrieval frequency and heavy pollution

Highlights

• Retrieval frequency has significant influence on consistency in multiple satellite Deep Blue aerosol products.

• AVHRR Deep Blue retrievals tends to miss high-AOD and clean bright surface due to over-strict cloud screening.

• Early 1990s is a key time point that frequent haze pollution appears and increases in eastern China.

Abstract

The recent AVHRR aerosol products expended from MODIS Deep Blue (DB) algorithm provide an unprecedented long-term data record over land that dates back to 1980s. Unlike previous studies focused on performance of AVHRR retrievals, here we present a critical view into the influence of retrieval frequency on consistency of AVHRR Aerosol Optical Depth (AOD) with other satellite DB products, and application potential in characterizing the dramatic aerosol loading in China. Despite the good consistency of collocated AVHRR, SeaWiFS, and MODIS DB AOD, their different retrieval frequency leads to distinct distribution of annual or seasonal AOD. In particular, the over-strict cloud screening of AVHRR with fewer bands tends to filter out high-AOD and clean bright surface, with the overall AOD frequency only half of MODIS's. Retrieval frequency and selection criterion of AOD quality can determine representativity and consistency of conventional averaged satellite AOD, uncertainties of which should be considered according to specific application purpose. The time series variation of AVHRR AOD shows that early 1990s is a key time point when frequent haze pollution appears and increases in eastern China. By contrast, aerosol loading in 1995–1999 has reached similar level as in 2000s with notable hotspot of high-AOD (~0.8–0.9) and slight changes. Our results shows that full record of AVHRR aerosol products with corresponding improvement can greatly renew the understanding of aerosols in China.

Introduction

Atmospheric aerosol is one of the key factors in the climate system by intensive and intricate interactions with solar radiation and clouds (IPCC, 2013). Also, concentration of these tiny particles near surface has robust statistical correlation with numerous epidemic diseases (Pope et al., 2002). To clearly and accurately quantify climate and health effects of the aerosols, it's essential to know how the aerosols change over space and time. However, distribution of aerosol loading and properties is very inhomogeneous due to their diverse sources and short lifetime, which needs integrated observations from regional to global scales. In particular, a continuous and long-term aerosol data record with sufficient accuracy and temporal-spatial resolution is imperative in assessing the role of aerosols in climate change (Levy et al., 2015) and air quality (van Donkelaar et al., 2015).

Since late 1990s, a series of sophisticated satellites with specific well-calibrated sensors have been launched to make global aerosol observations. To enhance aerosol retrieval over land surface that has remarkable reflectance contribution and complex bidirectional properties, multi-spectral, multi-angle, and polarized measurements from Moderate Resolution Imaging Spectroradiometer (MODIS), Multi-angle Imaging Spectroradiometer (MISR), and POLDER (Polarization and Directionality of the Earth's Reflectances) respectively, are utilized to separate aerosol signal from backscattered radiation of the coupled Earth-atmosphere system (King et al., 1999). Despite apparent advantage of angular and polarized observations in retrieval of aerosol properties, existing satellite sensors with such capabilities are limited by their swath width, or pixel resolution, or satellite lifetime. By avoiding most of above restrictions, MODIS's operational aerosol products have been widely used due to its near daily global coverage (Levy et al., 2013).

Satellite products have greatly updated the knowledge regarding global distribution of aerosol loading (Kaufman et al., 2002), while MODIS observation since 2000 is no more than 20 years by now. There is a temporal gap of regional and global aerosol observations due to lack of consistent satellite products over land before 2000. Recently, the Deep Blue (DB) aerosol project has constructed a long-term satellite data record from several additional multi-spectral imaging instruments (Table 1), including Sea-viewing Wide Field-of-view Sensor (SeaWiFS) (Hsu et al., 2012), Advanced Very High Resolution Radiometer (AVHRR) (Hsu et al., 2017), and Visible Infrared Imaging Radiometer (VIIRS) (Hsu et al., 2019). In particular, AVHRR onboard the National Oceanic and Atmospheric Administration (NOAA) polar-orbiting meteorological series satellites provide nearly four decades of unique long-term continuous measurements since 1981. Despite big difference in sensor characteristics and stability, DB aerosol products from these satellite instruments all exhibit reliable accuracy in ground validations, and have obvious consistency in their time series variations (Hsu et al., 2019; Sayer et al., 2017, Sayer et al., 2019), giving an unprecedented opportunity to inspect long-term variation of aerosol properties. However, besides similar algorithms, application of satellite aerosol products from multiple sensors can still have considerable uncertainties due to their difference in retrieval frequency (Hsu et al., 2012; Lee et al., 2018) as well as sensitivity to heavy pollution (Tao et al., 2015).

Owing to the combined influence of economic development and emission control policies, aerosol loading and properties in China have undergone dramatic variations during the past decades (Tao et al., 2016; Zhang et al., 2012). The long-term DB data record can not only provide valuable information concerning global aerosol climatology (Hsu et al., 2019), but also can fill the data gap in air quality studies with respect to emission sources and health effects. Sogacheva et al. (2018b) analyzed spatial and seasonal variations of aerosols over China from 1995 to 2017 by combined MODIS and Along-Track Scanning Radiometer (ATSR) Dual View (ADV) products. However, the frequent heavy pollution in China has substantial influence on cloud screening and available retrieval frequency of different satellite aerosol algorithms (Tao et al., 2015), which also determine stability of satellite-derived aerosol climatology in turn (Lee et al., 2018), especially for AVHRR with only five or six bands. Despite good performance of the AVHRR DB algorithm itself, two key questions remain unclear by now: 1) what's the influence of sampling bias, cloud screening, and selection criterion of retrieval quality on consistency of multi-satellite DB aerosol data records; 2) to what extent can the current AVHRR DB products capture variations of the dramatic aerosol loading in China.

In this study, we provide a critical insight into consistency of the unique long-term AVHRR DB aerosol data record with operational MODIS products and ground observations in China by considering their retrieval frequency and sensitivity to heavy pollution. Meanwhile, we discussed application potential of AVHRR aerosol products in characterizing spatial patterns and temporal fluctuations of the complicated aerosols in China. Section 2 gives a brief introduction of DB aerosol algorithm applied in AVHRR, SeaWiFS, MODIS, and VIIRS aerosol data as well as ground measurements. General performance of AVHRR, SeaWiFS, MODIS, and ground aerosol results is shown in section 3.1. Section 3.2 analyzes the influence of retrieval frequency on consistency of AVHRR DB aerosol products with other satellites’. Moreover, applicability of the current AVHRR aerosol data to the prevalent heavy pollution in China is discussed in section 3.3. Section 4 makes a summary of the work. The main purpose of this study is giving a comprehensive reference for corresponding applications associated with long-term satellite aerosol data record in China.