Predicting hot spots of aquatic plant biomass in a large floodplain river catchment in the Australian wet-dry tropics
Introduction
Numerous ecological and socio-economic functions supported by wetlands and freshwater habitats such as regulating nutrient cycle, maintaining fishery production and the role of aquatic plants in enhancing water clarity, among other functions have been widely reported (e.g., Chen et al., 2014, Zhao et al., 2012, Tockner et al., 2010, Midwood and Chow-Fraser, 2010, Keddy et al., 2009, Gidley, 2009, Cazzanelli et al., 2008, Ozesmi and Bauer, 2002). The large-scale assessment of wetlands and floodplains is therefore required to improve our contemporary understanding of wetland functions and their response to the degrading impacts of anthropogenic actions and strong fluctuations in climate. This is particularly the case for floodplains that are currently unaltered but under pressure from development that would impact their ecological function (Kingsford, 2000). However, physical access to several wetlands and freshwater habitats can be restricted due to difficult topography. Inaccessibility during times of inundation and extreme floods and technological limitations associated with field surveys are other key factors limiting the large-scale assessments of floodplain wetlands. While inaccessibility makes in-situ sampling challenging, aquatic primary producers such as macrophytes often thrive in freshwater habitats and floodplains that are extremely difficult to reach and perhaps unsafe (e.g., Zhao et al., 2012, Davranche et al., 2010), thus making traditional field surveys not only expensive but nearly impossible and risky. The ongoing development of satellite technologies are helping to overcome these challenges.
Several improvements in satellite remote sensing optical systems that have occurred in recent years have resulted in increased use of these technologies for the assessment of floodplain inundation, inventorying of wetlands, mapping the distribution of floods and aquatic plants, assessing planform changes, monitoring and extraction of surface water in extensive floodplains are growing (see, e.g., Tulbure and Broich, 2019, Normandin et al., 2018, Khandelwal et al., 2017, Dewan et al., 2017, Ward et al., 2016, Feyisa et al., 2014, Syvitski et al., 2012, Tockner et al., 2010, Sakamoto et al., 2007, Midwood and Chow-Fraser, 2010, Ozesmi and Bauer, 2002). The general isolation of the Australian wet-dry tropics, coupled with increasing pressures for water resource development on floodplain ecosystems and subsequent threats to ecological assets (e.g., Ward et al., 2014) has necessitated the use of remote sensing methods to assess floodplains on a large-scale. Such assessment is vital, not only to improve contemporary knowledge on drivers of inundation and aquatic primary production but to highlight the critical water needs for ecosystem health that will foster relevant policy and aid sustainability.
Tropical floodplain ecosystems are typically characterised by strong seasonality in rainfall, which in most cases results in considerable river flows and floodplain inundation during the wet seasons (e.g., Ward et al., 2014, Ward et al., 2013). Consistent with this pattern, the Flinders River catchment in the wet-dry tropics of northern Australia, experiences large fluctuations in rainfall resulting in long dry spells and seasonal flooding. Characterizing their seasonal spatial patterns and variability is critical to assess the water requirements for primary production and aquatic food webs. Whereas floodplain inundation mapping has been undertaken for some river basins in Australia, though at much smaller scales (e.g., Tulbure et al., 2016, Fisher et al., 2016, Ward et al., 2013, Ward et al., 2014), the characteristics (e.g., duration and extent) of large-scale seasonal inundation for the Flinders catchment, especially during major summer floods are largely undocumented.
These characteristics underpin knowledge on the impacts of climate change and/or anthropogenic influence on the distribution of primary production in aquatic ecosystems within wet-dry tropical systems. Isolated and discreet water holes within these systems have been identified as highly productive centers in terms of phytoplankton biomass production, macrophyte, periphyton and other primary producers that generate biomass for secondary aquatic producers (e.g., Burford et al., 2016, Ward et al., 2016, Waltham et al., 2013, Faggotter et al., 2013). While inundation extent in saline supratidal mudflats in wet-dry tropical regions of Australia was found to be an important indicator for the rates of primary productivity (see, Burford et al., 2016), the vulnerability of freshwater systems and aquatic habitats to climate influence and changes in human-water policies and management are strong motivations warranting large-scale assessments. In addition to this, the variability in reflectance values of aquatic vegetation caused by poor discrimination of aquatic plants from water signals in Lakes, especially during flood periods, and the limitations of the near-infrared wavelength in aquatic plant biomass prediction (e.g., Chen et al., 2018, Davranche et al., 2010, Cho et al., 2008, Silva et al., 2008, Malthus and George, 1997) have been reported. While these are pertinent issues, necessitating further research, the aforementioned reports were site-specific studies conducted within small Lakes, reservoirs, and wetlands/catchments.
On a global scale, the relationship between biophysical catchment properties (e.g., rainfall seasonality, dryness index, and topographic slope) and streamflow characteristics have been identified (Beck et al., 2013). In a study that investigated how drivers of streamflow characteristics vary at water management and continental scales in eastern Australia, dryness index, the fraction of photosynthetically active radiation from vegetation and soil properties accounted for observed variability in streamflow characteristics (Trancoso et al., 2017). Given the importance of floodplain rivers in providing essential services for society and ecosystems such as water supply, production of hydropower, flood control among others (e.g., Kennard et al., 2010, Keddy et al., 2009), process-based knowledge of the cascading impacts of hydro-meteorological fluctuations on ecological processes is crucial. This is because, among other reasons, they can directly feed into management and policy frameworks for biodiversity conservation and ecological assessments (Klein et al., 2009). Even though some tropical wetlands and catchments along the Australian Gulf rivers have being studied in recent times (e.g., Ndehedehe et al., 2020, Ward et al., 2016, Ward et al., 2014), the ability to answer key hydrological and ecological questions is still lacking and could be limited due to dissimilarity in biophysical catchment properties (e.g., soil properties, rainfall variability, evaporation rates, and predominant land cover states) on different spatial scales. Some of these questions include, what is the link between total downstream floodplain inundation patterns and flow regimes? What are the key drivers of hot spots of floodplain inundation and distribution of aquatic plants biomass along the Australian wet-dry tropics?
It is true that limited water availability resulting from large hydro-climatic fluctuations and decadal-scale droughts alters the biophysical properties of catchments in largely water deficit regions. This leads to non-stationary behaviour in the response of hydrological stores to rainfall as reported in some recent studies (e.g., Deb et al., 2019, Ndehedehe et al., 2019). In typical arid areas characterised by complex water use mechanisms during the dry season and where surface vegetation are morphologically and physiologically adapted to such environments (Guan et al., 2015, Seghieri et al., 2012, Huber et al., 2011), novel hydrological indicators to support the assessment of climatic constraints on agro-ecosystems have been explored (Ndehedehe et al., 2018). Therefore, the aforementioned questions are important and required to establish potential ecological problems that may result from considerable hydro-meteorological fluctuations (e.g., flood and drought events) and or change in government water policies or freshwater harvesting. With suitable analytical methodologies and empirical frameworks, these questions can be articulated and used as important tools in communicating environmental ecosystem health targeted at optimizing management initiatives, goals, and perspectives (Logan et al., 2019). To improve understanding on the connection between ecology and wetland hydrology of the Australian wet-dry tropics, an empirical framework based on quantitative assessment of relevant indicators is therefore required. While providing a basis for the use of remote sensing technique for large-scale assessments of remote floodplain systems, such a framework will also provide decision makers and other stakeholders with a quantitative basis for predicting the impacts of water resource development on floodplain wetlands.
The main aim of this study is to assess changes in large-scale seasonal total inundation and predict the spatial distribution of hot spots of primary producers (aquatic plant biomass) in a large floodplain wetland in the wet-dry tropics of Australia. The specific objectives are to (i) assess large-scale spatio-temporal changes in total floodplain inundation and (ii) predict the spatial distribution of hot spots of primary producers by integrating remote sensing indicators with digital elevation model in a classification tree model. The relationships of the predicted hot spots in (ii) above with total downstream floodplain inundation, rainfall and discharge are explored to understand the connection between ecology and wetland hydrology. Notably, remote sensing based inundation metrics have been used in quantifying floods and mapping floodplain inundation. But the need to assess the stability of these indicators in quantifying open water features and total floodplain inundation on complex landscapes such as the Flinders catchment is essential to provide more knowledge on the role of climate variability on changes in wetland hydrology. While this has also been undertaken in this study, an empirical framework that integrates a multivariate analysis of hydrological indicators (e.g., river flow) is developed for the first time and used for the quantitative assessment of local wetland hydrology. Another key application of this framework in this study is to assess the influence of inter-annual variability of rainfall on total floodplain inundation and spatial distribution of aquatic plant biomass accumulation. Such assessment is crucial and provides insight that underpins the prioritization of future surface water developments that are currently under consideration,1 conservation planning, and wetland management in this region.
Section snippets
Study area
The study region is one of the southern gulf catchments in semi-arid tropical climate of northern Australia, the Flinders catchment (Fig. 1), and has an approximate area of 109,000 km2 (e.g., Waltham et al., 2013). The unique geological features, topography, landscape characteristics and hydrology of the funnel-shaped Flinders catchment (Fig. 1a) are interesting and, includes valuable resources, e.g., minerals, coal, groundwater aquifers and soil, among others. Apart from its rich geological
Delineation of inundated areas and floodplain inundation mapping
This section is designed to deliver spatial information on floodplain inundation dynamics through a comparative analyses of surface water extraction metrics using 2009 summer flood as a case study. To this end, the major floods in the wet season of 2009 was assessed using three different but prominent water extraction metrics (MNDWI, and ) derived from multi-band Landsat imagery. Moreover, in addition to assessing the hydrological indicators of floodplain productivity, the recent
Discussion
Given the role of climate and human activities as key indicators of processes that may affect water availability and the productivity of floodplain ecosystems, understanding such implications are important for environmental water allocation and the ecological communities supported by the wetlands and freshwater systems that generate multiple strings of economic values and services. Apart from the explicit representation of seasonal floodplain inundation dynamics, an ideal satellite-derived
Conclusion
Given the role of climate and human activities as key indicators of processes that may affect water availability and the productivity of floodplains ecosystems, understanding such implications is important for environmental water resource allocation and the ecological communities supported by the wetlands and freshwater systems that generate multiple strings of economic values and services. These values have generated increasing interest in water resource development among all levels of
CRediT authorship contribution statement
Christopher E. Ndehedehe: Writing - original draft, Writing - review & editing, Methodology, Formal analysis, Conceptualization, Investigation, Software, Data curation. Ben Stewart-Koster: Conceptualization, Investigation, Writing - review & editing. Michele. A. Burford: Conceptualization, Funding acquisition, Project administration. Stuart E. Bunn: Conceptualization, Project administration, Funding acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work is supported with funding from the Australian Government’s National Environmental Science Program. The Landsat data used in this study were retrieved from USGS data portal. The authors are grateful to the Queensland Government for the river discharge observations and precipitation data accessed from the SILO climate database. The invaluable comments of two anonymous reviewers, which helped improved the quality of this manuscript are greatly appreciated.
References (92)
- et al.
Assessing multi-satellite remote sensing, reanalysis, and land surface models’products in characterizing agricultural drought in East Africa
Remote Sens. Environ.
(2017) - et al.
Controls on phytoplankton productivity in a wet-dry tropical estuary
Estuar. Coast. Shelf Sci.
(2012) - et al.
Using satellite based soil moisture to quantify the water driven variability in NDVI: a case study over mainland Australia
Remote Sens. Environ.
(2014) - et al.
Wetland monitoring using classification trees and SPOT-5seasonal time series
Remote Sens. Environ.
(2010) - et al.
Mapping flooding regimes in Camargue wetlands using seasonal multispectral data
Remote Sens. Environ.
(2013) - et al.
Mechanisms influencing non-stationarity in rainfall-runoff relationships in southeast australia
J. Hydrol.
(2019) - et al.
Assessing channel changes of the Ganges-Padma River system in Bangladesh using Landsat and hydrological data
Geomorphology
(2017) - et al.
A generalized framework for deriving nonparametric standardized drought indicators
Adv. Water Resour.
(2015) - et al.
Suitability of NDVI and OSAVI as estimators of green biomass and coverage in a semi-arid rangeland
Ecol. Ind.
(2018) - et al.
Automated water extraction index: A new technique for surface water mapping using landsat imagery
Remote Sens. Environ.
(2014)