Simple and effective monitoring of historic changes in nearshore environments using the free archive of Landsat imagery

https://doi.org/10.1016/j.jag.2009.09.002Get rights and content

Abstract

The recently released archive of Landsat imagery can be used to detect historic changes in nearshore environments. We used a series of free Landsat images spanning the years from 1984 to 2009 to detect changes in the spatial extent of dominant substrate types, coral, algae, and seagrass, around Bawe and Chumbe islands in Zanzibar, and we compared the use of true-colour composites and supervised classifications. Results indicate temporal changes in the spatial extent of seagrass meadows are easily mapped with Landsat imagery, whereas temporal changes in algae cover and particularly coral cover pose greater challenges because of the similarities in spectral reflectance properties between the relevant substrate types. Supervised classification requires substantially more processing than the simple display of true-colour composites, but does not improve interpretation in our study. We suggest that historic Landsat imagery, obtained at no cost and processed minimally with free software, is the best available data source for studies of historic changes in the nearshore environments of East Africa.

Introduction

The nearshore environments of East Africa contain numerous resources crucial to the economy of the region, including hundreds of fish and invertebrate species that are either exported or consumed locally (Gössling et al., 2004, Torell et al., 2007), mangrove trees that are harvested for construction or left standing for shoreline protection, as well as coral reefs and white beaches that attract millions of tourists every year (United Nations, 2002). Pressures on these resources have intensified along with increases in the human population (McClanahan et al., 1999), making appropriate coastal planning and management crucial for sustainable development (Cohen et al., 1997, Small and Nicholls, 2003). Information on the magnitude and spatial location and extent of changes in the nearshore environment, either natural or resulting from human impacts, will therefore be essential for the development of management responses. Such information, however, can be difficult and expensive to generate, especially in the least developed regions of Africa where the nearshore environment is most crucial for local livelihoods. The time series of data necessary to monitor environmental change are usually only available for a few sites located in easily accessible locations (Gullström et al., 2002), and research funding and capacity is often too limited to expand in situ monitoring. The recently released archive of Landsat imagery is an alternative data source, one that is both temporally and spatially extensive and now freely available for download over the Internet (http://glovis.usgs.gov/). The archive contains imagery from 1972 to the present, and covers most coastal regions of the world (Arvidson et al., 2001). With very limited processing, change detection using this imagery can provide scientists and managers with the necessary information on environmental change in otherwise data-poor areas.

In this study, we investigate what changes are identifiable in the tropical nearshore environment with a time series of Landsat imagery, focusing on the three major living substrate types in this environment: corals, seagrasses and algae. We use 27 Landsat images covering the period from 1984 to 2009, along with recent field data (from 2007 to 2008), to assess long-term trends in these three substrate types around two small islands in Zanzibar. In addition, we assess the magnitude of seasonal variations in substrate reflectance and its effect on interpretations of the time series of Landsat images, and we assess whether the use of supervised classification provides information in addition to what can be obtained by visual inspection of true-colour composites.

Section snippets

Study area

Zanzibar consists of two main islands, Unguja and Pemba, with numerous smaller islands surrounding them. We focus our study on the shallow (<10 m depth) inter-tidal and sub-tidal areas around two small islands, Chumbe and Bawe, both located off the west coast of Unguja, the main island of Zanzibar (Fig. 1). The nearshore areas around Chumbe and Bawe islands represent two extremes of human exploitation. Bawe Island, located 5 km from Zanzibar Town, is a popular fishing ground, visited daily by

Data and methods

A total of 27 Landsat images were used in this study (Table 1), all freely available for download from the Landsat archive. Images were preprocessed to the 1G, 1Gt or 1T levels (USGS, 2009), technical details for Landsat imagery are provided in Table 2.

All images were geometrically corrected to the 2009 image with <20 m RMSE. In addition to our study area, the full Landsat scenes (path 166, row 64) also cover Unguja and nearby islets, the southern part of Pemba Island, and part of the Tanzanian

Assessment of seasonal changes in the nearshore environment

The 10 images of Bawe Island and the surrounding area from 2004 are shown in Fig. 2.

Although no general pattern of seasonal change is seen in Fig. 2, several areas display variations in brightness throughout 2004. The seemingly stochastic changes correspond to variations in tidal height at the time of image acquisition, which could not be adequately corrected for with the manual contrast enhancement. However, one area northeast of the island, highlighted with a red circle in Fig. 2, is

Discussion

What changes in the nearshore environment were identifiable in the Landsat imagery, and was there a significant difference between the quick and simple use of true-colour composites, and the more time-consuming use of supervised classifications?

The shifting spatial extent of seagrass meadows seen around Chumbe Island was easily identified in the true-colour composites, and also seen in the supervised classifications, although somewhat confused by misclassifications between “Seagrass” and

Conclusion

In this study, we have outlined a simple method for investigating historic changes in a tropical nearshore environment containing coral reefs and seagrass beds, using a type of dataset typically available to researchers at minimum cost: a set of (free) historic Landsat images, and a set of recent field observations. Results indicate that changes in the spatial extent of seagrass meadows are easy to identify on true-colour composites of Landsat imagery, whereas subtle changes in algal and coral

Acknowledgements

The authors would like to acknowledge the assistance provided to us by the Institute of Marine Sciences (IMS) in Zanzibar and the GEF/World Bank funded Coral Reef Targeted Research project, and acknowledge helpful comments from two anonymous reviewers.

References (32)

  • J.E. Cohen et al.

    Estimates of coastal populations

    Science

    (1997)
  • P. Dustan et al.

    Changes in the reef-coral community of Carysfort Reef, Key-Largo, Florida—1974 to 1982

    Coral Reefs

    (1987)
  • P. Dustan et al.

    Landsat thematic mapper: detection of shifts in community composition of coral reefs

    Conserv. Biol.

    (2001)
  • R. Ferguson et al.

    Monitoring spatial change in seagrass habitat with aerial photography

    Photogramm. Eng. Remote Sens.

    (1993)
  • T.A. Gardner et al.

    Long-term region-wide declines in Caribbean corals

    Science

    (2003)
  • S. Gössling et al.

    Use of molluscs, fish, and other marine taxa by tourism in Zanzibar, Tanzania

    Biodivers. Conserv.

    (2004)
  • Cited by (45)

    • Landsat historical records reveal large-scale dynamics and enduring recovery of seagrasses in an impacted seascape

      2022, Science of the Total Environment
      Citation Excerpt :

      The application of Landsat data in long-term time-series analysis has become more common after the Landsat archive became free and open in 2008, but the vast majority of studies is still restricted to change in terrestrial land cover and land use (Zhu, 2017). Investigations of seagrass dynamics are comparatively few and generally focused on shallow (< 10 m) waters (Chen et al., 2016; Hossain et al., 2015a; Hossain et al., 2019; Knudby et al., 2010; Lyons et al., 2013; Strydom et al., 2020). The available studies highlight the usefulness of Landsat imagery to map seagrass extent against bare sand but suggest classification uncertainty in the distinction of coral, algae or muddy substrates, particularly where habitat patches are smaller than the Landsat pixel.

    • Decadal increase in the ecological status of a North-Atlantic intertidal seagrass meadow observed with multi-mission satellite time-series

      2021, Ecological Indicators
      Citation Excerpt :

      The systematic use of EO in complement to field surveys would make it possible to more accurately document the status of intertidal seagrass meadows as a result of observation improvements in terms of number of sites, spatial coverage and resolution, temporal frequency length of time-series, and seagrass metric consistency. First, EO can provide an historical baseline of seagrass indicators in the absence of in situ data (Knudby et al., 2010; León-Pérez et al., 2019), and provide updated information on seagrass status in many sites worldwide (Unsworth et al., 2019). In Europe, EO long-term and updated time-series could also help to select the year used as reference for the computation of the EQR within the WFD.

    • Seagrass mapping in Greek territorial waters using Landsat-8 satellite images

      2018, International Journal of Applied Earth Observation and Geoinformation
    View all citing articles on Scopus
    1

    Tel.: +1 519 888 4567.

    2

    Tel.: +255 24 2230741.

    View full text