Full Paper
7th Australian Stream Management Conference – Short Communication
Tracking Waterhole Persistence in Dry Tropical Rivers 1
2
Kinsey-Henderson, Anne , Steve Marvanek , David McJannet
3
1. CSIRO Land and Water, ATSIP Building 145, James Cook University, Douglas, QLD 4811 Australia, [email protected] 2. CSIRO Land and Water, Gate 5, Waite Road, Urrbrae, SA, 5064 Australia 3. CSIRO Land and Water, Ecosiences Precinct, 41 Boggo Road, Dutton Park, QLD 4102 Australia
Key Points • Landsat imagery was used to identify and track changes to waterhole area over widespread and often inaccessible areas • Waterholes were of greater area and more persistent in the Gilbert than the Flinders catchment • Relationships between waterhole area and time since cease to flow provided a means to predict the potential impacts of changed flow conditions on waterhole persistence
Keywords Landsat, northern Australia, waterhole, refugia, persistence, climate change
Introduction During the dry season (May-October) the Flinders and Gilbert rivers (northern Australia) dry to a series of in-stream waterholes, the persistence of which is important for providing water for native fauna and stock and refugial habitat for terrestrial and aquatic biota. The catchments for both of these rivers have been identified as potential areas for further agricultural development. Therefore it is important to consider the likely impact of changed flow conditions on waterhole persistence. As part of the CSIRO Flinders and Gilbert Agricultural Resource Assessment (FGARA www.csiro.au/fgara), Landsat time series imagery was used to quantify the intra-annual changes in waterhole area from 2003 to 2010 and to then relate these changes to streamflow (or time since cease to flow). This work is important for understanding the impacts development and future climate may have on these habitats, particularly when considered in conjunction with water quality (e.g. Faggotter et al. 2011,; Waltham et al., 2013). This short communication provides a brief summary of our work on in-stream waterhole persistence in the Flinders and Gilbert catchments. For more comprehensive information, readers are referred to (McJannet et al., 2013; McJannet et al., Accepted).
Study Area and Methods 2
The Flinders and Gilbert catchments (109,000 km and 46,354 km2 respectively) are located in the tropics and have -1 -1 highly seasonal and variable rainfall averaging 492 mmy and 775 mmy respectively. The dominant land use is grazing and the region is generally remote with poor on-ground access. Seven river reaches (three in the Flinders and four in the Gilbert), were identified as focus areas for this work based on potential development areas identified as part of the Flinders and Gilbert Agricultural Resource Assessment. The analysis was restricted to areas within the stream network in the selected reaches. The Statewide Landcover and Trees Study (SLATS) Landsat water index (Muir & Danaher, 2008) was used for determining the areas of waterholes along each reach. Calibration of the index threshold (water vs non-water) was undertaken using waterhole area from multi-date high resolution Google Earth imagery as a reference set. A total of 400 cloud-free scenes spanning 6 Landsat scene areas were ultimately combined to generate a time series of waterhole area for each reach for an eight year period (2003-2010). Key refugia were identified as those areas within the river that persisted as water for at least 90% of the time-series (100% persistence was not always reliably detected in the 30 m resolution water indices), although long-term persistence alone did not imply healthy, functional or resilient ecosystems (Waltham et al., 2013). This time series of waterhole area was then compared to stream flow data to develop relationships between waterhole area and stream flow characteristics. Such relationships are vital for predicting changes in available waterhole habitat following changed flow conditions under future climate and development.
Results and Discussion For waterholes in the Flinders and Gilbert, water index thresholds of 77 and 80 respectively provided the best match to the area of the waterholes digitised from Google Earth. Estimation of waterhole area from water index thresholds either side of the selected thresholds for each catchment resulted in over- or under-predictions of between 19% and 44%, lending confidence to the selected values. The necessity for catchment-specific thresholds warrants further Kinsley-Henderson, A., Marvenek, S. & McJannet, D. (2014). Tracking Waterhole Persistence in Dry Tropical Rivers, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 138-140. 138
7ASM Short Communication Kinsey-Henderson et.al. – Tracking waterhole persistence in dry tropical rivers investigation, although it seems likely to relate to the factors such as waterhole clarity. Waterholes in the Gilbert tend to be larger, and clearer than waterholes in the Flinders Lymburner and Burrows (2008, 2009) and Waltham, et al. (2013). Key refugia were much more numerous in the Gilbert catchment than the Flinders (Figure 1a). The river channels in the Flinders are wide, flat, and complex, resulting in only 6 key refugia, of which only one Is larger than 5 ha (Dalgonally Waterhole). In the Gilbert there were numerous key refugia, particularly in the Einasleigh River sections which have well defined river channels, higher rainfall and potential for groundwater inflows (Jolly et al., 2013). The low number of key refugia in the Flinders catchment makes this area particularly sensitive to changed flow conditions.
Figure 1: a) Location of key aquatic refugia identified in the Flinders and Gilbert catchments (small = 0.36 to 2.5 ha, medium = 2.5 to 5.0 ha, large = 5.0 to 7.5 ha, very large = > 7.5 ha) b) Typical relationship between time since cease to flow and total waterhole area (in this case for the Mid-Flinders river reach). Yellow points represent outliers. Good relationships were found between time since cease to flow and area of waterholes for a majority of the river reaches. Time since cease to flow was chosen as the criteria as the highly seasonal rainfall resulted in extended periods of zero flow of up to 332 days (90% of the year). A typical relationship is shown in Figure 1b. Outliers tended to be of two types, firstly where streamflow was occurring in the upper part of the river reach but had not yet made it to the gauging point (yellow points in Fig 1b) and secondly as a consequence of small flows during the dry season (not shown). Relationships of a similar form were found for a number of reaches. Combining these relationships enabled derivation of a generalized relationship which could be used to predict changes to waterhole area under changed flow conditions for other regions without detailed Landsat analysis. For analysis of the potential impact of climate change on waterhole area in the Flinders and Gilbert, readers are referred to McJannet et al. (2013) and McJannet et al. (Accepted).
Conclusions This study showed that Statewide Landcover and Trees Study (SLATS) Landsat water index can be used to detect development and persistence of in-stream waterholes in ephemeral river systems in northern Australia. Catchmentspecific calibration of the water index was essential and was likely due to differences in waterhole clarity and shape between the two catchments. We found that key refugia are numerous in the Gilbert due to higher rainfall, better defined channels and potential groundwater influx. Waterholes were much less common in the Flinders catchment, making in-stream ecosystems in this area potentially more vulnerable to development pressures and climate change. We found good relationships between waterhole area and time since cease to flow which will enable prediction of waterhole area under different flow conditions.
Kinsley-Henderson, A., Marvenek, S. & McJannet, D. (2014). Tracking Waterhole Persistence in Dry Tropical Rivers, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 138-140. 139
7ASM Short Communication Kinsey-Henderson et.al. – Tracking waterhole persistence in dry tropical rivers Acknowledgments The research would not have been possible without the processed Landsat imagery provided by members of the SLATS program including Giri Khanal, Neil Flood, Nicholas Goodwin, Lisa Collette and Christian Witte. Funding for this research was provided by the Australian Government Office of Northern Australia and CSIRO Water for a Healthy Country.
References Faggotter, S. J., Burford, M. A., Robson, B. J., & Webster, I. T. (2011). Nutrients and Primary Production in the Flinders River: Charles Dawrin University, Darwin. Jolly, I., Taylor, A., Rassam, D., Knight, J., Davies, P., & Harrington, G. (2013). Surface water groundwater connectivity. A technical report to the Australian Government from the CSIRO Flinders and Gilbert Agricultural Resource Assessment, part of the North Queensland Irrigated Agriculture Strategy: CSIRO Water for a Healthy Country and Sustainable Agriculture flagships, Australia. Lymburner, L., & Burrows, D. (2008). A Landsat TM Inventory of Waterbody Permanence and Clarity in the Mitchell and Gilbert catchments, north Queensland. ACTFR Report No 08/16: ACTFR. Lymburner, L., & Burrows, D. (2009). A Landsat TM Inventory of Waterbody Permanence and Clarity in Catchments of the Southern Gulf of Carpentaria, North Queensland. ACTFR Report No 09/10: ACTFR. McJannet, D., Marvanek, S., Kinsey-Henderson, A., Petheram, C., & Wallace, J. (2013). Instream waterholes. A technical report to the Australian Government from the CSIRO Flinders and Gilbert Agricultural Resource Assessment, part of the North Queensland Irrigated Agriculture Strategy: CSIRO Water for a Healthy Country and Sustainable Agriculture flagships, Australia. McJannet, D., Marvanek, S., Kinsey-Henderson, A., Petheram, C., & Wallace, J. (Accepted). Persistence of in-stream waterholes in ephemeral rivers of tropical north Australia and potential impacts of climate change. Marine and Freshwater Research. MF14035. Muir, J., & Danaher, T. (2008). Mapping water body extent in Queensland through time series analysis of Landsat imagery. Paper presented at the 14th Australasian Remote Sensing and Photogrammetry Conference, Darwin, Australia. Waltham N., B., D., Butler, B., Wallace, J., Thomas, C., James, C., Brodie, J (2013). Waterhole ecology in the Flinders and Gilbert catchments. A technical report to the Australian Government from the CSIRO Flinders and Gilbert Agricultural Resource Assessment, part of the North Queensland Irrigated Agriculture Strategy. CSIRO Water for a Healthy Country and Sustainable Agriculture flagships, Australia.
Kinsley-Henderson, A., Marvenek, S. & McJannet, D. (2014). Tracking Waterhole Persistence in Dry Tropical Rivers, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 138-140. 140
Tracking Waterhole Persistence in Dry Tropical Rivers 1
2
Kinsey-Henderson, Anne , Steve Marvanek , David McJannet
3
1. CSIRO Land and Water, ATSIP Building 145, James Cook University, Douglas, QLD 4811 Australia, [email protected] 2. CSIRO Land and Water, Gate 5, Waite Road, Urrbrae, SA, 5064 Australia 3. CSIRO Land and Water, Ecosiences Precinct, 41 Boggo Road, Dutton Park, QLD 4102 Australia
Key Points • Landsat imagery was used to identify and track changes to waterhole area over widespread and often inaccessible areas • Waterholes were of greater area and more persistent in the Gilbert than the Flinders catchment • Relationships between waterhole area and time since cease to flow provided a means to predict the potential impacts of changed flow conditions on waterhole persistence
Keywords Landsat, northern Australia, waterhole, refugia, persistence, climate change
Introduction During the dry season (May-October) the Flinders and Gilbert rivers (northern Australia) dry to a series of in-stream waterholes, the persistence of which is important for providing water for native fauna and stock and refugial habitat for terrestrial and aquatic biota. The catchments for both of these rivers have been identified as potential areas for further agricultural development. Therefore it is important to consider the likely impact of changed flow conditions on waterhole persistence. As part of the CSIRO Flinders and Gilbert Agricultural Resource Assessment (FGARA www.csiro.au/fgara), Landsat time series imagery was used to quantify the intra-annual changes in waterhole area from 2003 to 2010 and to then relate these changes to streamflow (or time since cease to flow). This work is important for understanding the impacts development and future climate may have on these habitats, particularly when considered in conjunction with water quality (e.g. Faggotter et al. 2011,; Waltham et al., 2013). This short communication provides a brief summary of our work on in-stream waterhole persistence in the Flinders and Gilbert catchments. For more comprehensive information, readers are referred to (McJannet et al., 2013; McJannet et al., Accepted).
Study Area and Methods 2
The Flinders and Gilbert catchments (109,000 km and 46,354 km2 respectively) are located in the tropics and have -1 -1 highly seasonal and variable rainfall averaging 492 mmy and 775 mmy respectively. The dominant land use is grazing and the region is generally remote with poor on-ground access. Seven river reaches (three in the Flinders and four in the Gilbert), were identified as focus areas for this work based on potential development areas identified as part of the Flinders and Gilbert Agricultural Resource Assessment. The analysis was restricted to areas within the stream network in the selected reaches. The Statewide Landcover and Trees Study (SLATS) Landsat water index (Muir & Danaher, 2008) was used for determining the areas of waterholes along each reach. Calibration of the index threshold (water vs non-water) was undertaken using waterhole area from multi-date high resolution Google Earth imagery as a reference set. A total of 400 cloud-free scenes spanning 6 Landsat scene areas were ultimately combined to generate a time series of waterhole area for each reach for an eight year period (2003-2010). Key refugia were identified as those areas within the river that persisted as water for at least 90% of the time-series (100% persistence was not always reliably detected in the 30 m resolution water indices), although long-term persistence alone did not imply healthy, functional or resilient ecosystems (Waltham et al., 2013). This time series of waterhole area was then compared to stream flow data to develop relationships between waterhole area and stream flow characteristics. Such relationships are vital for predicting changes in available waterhole habitat following changed flow conditions under future climate and development.
Results and Discussion For waterholes in the Flinders and Gilbert, water index thresholds of 77 and 80 respectively provided the best match to the area of the waterholes digitised from Google Earth. Estimation of waterhole area from water index thresholds either side of the selected thresholds for each catchment resulted in over- or under-predictions of between 19% and 44%, lending confidence to the selected values. The necessity for catchment-specific thresholds warrants further Kinsley-Henderson, A., Marvenek, S. & McJannet, D. (2014). Tracking Waterhole Persistence in Dry Tropical Rivers, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 138-140. 138
7ASM Short Communication Kinsey-Henderson et.al. – Tracking waterhole persistence in dry tropical rivers investigation, although it seems likely to relate to the factors such as waterhole clarity. Waterholes in the Gilbert tend to be larger, and clearer than waterholes in the Flinders Lymburner and Burrows (2008, 2009) and Waltham, et al. (2013). Key refugia were much more numerous in the Gilbert catchment than the Flinders (Figure 1a). The river channels in the Flinders are wide, flat, and complex, resulting in only 6 key refugia, of which only one Is larger than 5 ha (Dalgonally Waterhole). In the Gilbert there were numerous key refugia, particularly in the Einasleigh River sections which have well defined river channels, higher rainfall and potential for groundwater inflows (Jolly et al., 2013). The low number of key refugia in the Flinders catchment makes this area particularly sensitive to changed flow conditions.
Figure 1: a) Location of key aquatic refugia identified in the Flinders and Gilbert catchments (small = 0.36 to 2.5 ha, medium = 2.5 to 5.0 ha, large = 5.0 to 7.5 ha, very large = > 7.5 ha) b) Typical relationship between time since cease to flow and total waterhole area (in this case for the Mid-Flinders river reach). Yellow points represent outliers. Good relationships were found between time since cease to flow and area of waterholes for a majority of the river reaches. Time since cease to flow was chosen as the criteria as the highly seasonal rainfall resulted in extended periods of zero flow of up to 332 days (90% of the year). A typical relationship is shown in Figure 1b. Outliers tended to be of two types, firstly where streamflow was occurring in the upper part of the river reach but had not yet made it to the gauging point (yellow points in Fig 1b) and secondly as a consequence of small flows during the dry season (not shown). Relationships of a similar form were found for a number of reaches. Combining these relationships enabled derivation of a generalized relationship which could be used to predict changes to waterhole area under changed flow conditions for other regions without detailed Landsat analysis. For analysis of the potential impact of climate change on waterhole area in the Flinders and Gilbert, readers are referred to McJannet et al. (2013) and McJannet et al. (Accepted).
Conclusions This study showed that Statewide Landcover and Trees Study (SLATS) Landsat water index can be used to detect development and persistence of in-stream waterholes in ephemeral river systems in northern Australia. Catchmentspecific calibration of the water index was essential and was likely due to differences in waterhole clarity and shape between the two catchments. We found that key refugia are numerous in the Gilbert due to higher rainfall, better defined channels and potential groundwater influx. Waterholes were much less common in the Flinders catchment, making in-stream ecosystems in this area potentially more vulnerable to development pressures and climate change. We found good relationships between waterhole area and time since cease to flow which will enable prediction of waterhole area under different flow conditions.
Kinsley-Henderson, A., Marvenek, S. & McJannet, D. (2014). Tracking Waterhole Persistence in Dry Tropical Rivers, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 138-140. 139
7ASM Short Communication Kinsey-Henderson et.al. – Tracking waterhole persistence in dry tropical rivers Acknowledgments The research would not have been possible without the processed Landsat imagery provided by members of the SLATS program including Giri Khanal, Neil Flood, Nicholas Goodwin, Lisa Collette and Christian Witte. Funding for this research was provided by the Australian Government Office of Northern Australia and CSIRO Water for a Healthy Country.
References Faggotter, S. J., Burford, M. A., Robson, B. J., & Webster, I. T. (2011). Nutrients and Primary Production in the Flinders River: Charles Dawrin University, Darwin. Jolly, I., Taylor, A., Rassam, D., Knight, J., Davies, P., & Harrington, G. (2013). Surface water groundwater connectivity. A technical report to the Australian Government from the CSIRO Flinders and Gilbert Agricultural Resource Assessment, part of the North Queensland Irrigated Agriculture Strategy: CSIRO Water for a Healthy Country and Sustainable Agriculture flagships, Australia. Lymburner, L., & Burrows, D. (2008). A Landsat TM Inventory of Waterbody Permanence and Clarity in the Mitchell and Gilbert catchments, north Queensland. ACTFR Report No 08/16: ACTFR. Lymburner, L., & Burrows, D. (2009). A Landsat TM Inventory of Waterbody Permanence and Clarity in Catchments of the Southern Gulf of Carpentaria, North Queensland. ACTFR Report No 09/10: ACTFR. McJannet, D., Marvanek, S., Kinsey-Henderson, A., Petheram, C., & Wallace, J. (2013). Instream waterholes. A technical report to the Australian Government from the CSIRO Flinders and Gilbert Agricultural Resource Assessment, part of the North Queensland Irrigated Agriculture Strategy: CSIRO Water for a Healthy Country and Sustainable Agriculture flagships, Australia. McJannet, D., Marvanek, S., Kinsey-Henderson, A., Petheram, C., & Wallace, J. (Accepted). Persistence of in-stream waterholes in ephemeral rivers of tropical north Australia and potential impacts of climate change. Marine and Freshwater Research. MF14035. Muir, J., & Danaher, T. (2008). Mapping water body extent in Queensland through time series analysis of Landsat imagery. Paper presented at the 14th Australasian Remote Sensing and Photogrammetry Conference, Darwin, Australia. Waltham N., B., D., Butler, B., Wallace, J., Thomas, C., James, C., Brodie, J (2013). Waterhole ecology in the Flinders and Gilbert catchments. A technical report to the Australian Government from the CSIRO Flinders and Gilbert Agricultural Resource Assessment, part of the North Queensland Irrigated Agriculture Strategy. CSIRO Water for a Healthy Country and Sustainable Agriculture flagships, Australia.
Kinsley-Henderson, A., Marvenek, S. & McJannet, D. (2014). Tracking Waterhole Persistence in Dry Tropical Rivers, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 138-140. 140
