- Geological and Bioregional Assessment Program
- GBA Roadshow
- Roadshow 3.3 Gas extraction and vegetation condition: identifying management-driven dynamics in vegetation cover
Gas extraction and vegetation condition: identifying management-driven dynamics in vegetation cover
Click here to open transcript
Hi, my name is Randall Donohue. I'm a research scientist at CSIRO Land and Water. I use satellites to study vegetation. And so for GBA in the Cooper I conducted an assessment of whether gas extraction activities have an impact on vegetation and whether that can be detected from satellite.
One of the challenges for doing this is that the region is very data-poor regions. So there's not a lot of field-based observations to build a model off so that the analysis technique had to be fairly remote. And satellites are excellent for that. The other challenge was that to assess the impact of human activities on vegetation, you first got to account for the natural variability in vegetation. And I'm going to present a method we came up with for achieving that. To examine the impacts of vegetation, we used a variable called cover. So that's the total cover. It's a measure of how much vegetation is covering the ground of any sort, whether it's dead or alive. And the reciprocal of that is bare ground. The value of this is that it's very effectively observed from satellite. But it's also a good proxy of landscape condition in terms of grazing intensity and disturbances generally, in terms of how much soil is exposed to waves and erosion. So we examined the region to see if there was an observable impact of mining activities, extraction activities on cover.
You can see here that there's a gradient in the left map going from the southwest to the northeast, and that's rainfall-driven gradient. So assessing the impact of any human activity on vegetation, you first got to account for natural variability and patterns. So that you're not, for example, comparing one site that’s a dry location with another site that's quite wet because you inherently get different answers. So a key idea in this assessment is that we are only comparing sites that have the same growing conditions, the same biophysical attributes.
The method we use is a benchmarking framework called Compere. It's a very general and flexible framework for doing peer-based comparisons that are aimed at teasing apart the natural variability in some variable from the human-induced or management effects in that variable. So the key idea is that you can identify equivalent locations across the region equivalent being locations that have the same biophysical or growing properties. And so, for a target location, if you can identify all the locations around it that are equivalent, then the values of those locations in our case cover all the cover values. Any difference in those values across those equivalent locations should be due to management.
So in other words, the benchmarking framework takes out the effects of natural processes. This is done for every location across a study region, and also for every year through the study period. So you end up with maps and trends of the effects of management on cover. The variables we used to define biophysical equivalents for the Cooper were precipitation because that is the key driver of vegetation. We used slope and this is because the redistribution of surface water is also a critical process for vegetation. And we also use soil colour because soil types and landforms and the different vegetation types that go with them are reasonably well explained using soil colour as a proxy for soil type. The advantage of using soil colour is that it's remotely sensed. And it's higher quality data in terms of realistic boundaries than the available soil maps.
So if you do this benchmarking comparison for every point across the whole region, so you get your set of equivalent cover values, we ranked them each time. And so if the target location is ranked at the 50th percentile, then it's on average, and it's, you know, in between in terms of cover. Whereas if it ranked at the 100th percentile, we know that it's the highest cover, whatever the management is doing at that location, it is achieving the highest cover possible for those biophysical equivalence properties. And because we've did this through time, we can get a trend as well. So cover ranking can be interpreted directly as a condition. Cover condition 100 is the highest condition present, zero is the lowest condition present.
So on the right, you can see trends, no overall big patterns of change, it's fairly speckly, there are a few patches of increase, few patches of decrease. But when you dive in and look at some of the details, that's where the useful information comes. So we looked at well locations to use this to study the impact of gas extraction on vegetation. So for all the recorded well locations across the whole area, we looked at the cover condition at the well. And then we looked to look at the effect of the concentration or the spatial density of wells. We also analysed the cover for a given number of wells per grid cells. So these grid cells are 500 metres.
So you can see the graph on the right, the cover rank, which is our proxy for condition for cells without wells was about 45. But as soon as you start looking at cells that have one or more well within them, the cover rank is lower, so the condition is lower and statistically significantly lower. But it doesn't seem to matter how many wells you have in a cell, it doesn't matter what the concentration of cells is.
And if you look at this, if you split it in two parts of the landscape that receive run on so the floodplains, that's what we call the channels versus the dry land where the only water they receive is rainfall. You get a slightly different result. So wells within channels seem to have a larger impact on cover than wells in dryland regions. We can also look at the temporal effects of the presence of wells. So this graph establishes the rig date so the time where it starts operating when looking at cover one year before rig date, the year of rig date, and then the years following rig date. And you can see that the first year after, up to the fourth year after, the rig date condition, the cover ranking is low statistically significantly lower in those first four years than it was prior to well establishment. But then a funny thing happens is at year five, it recovers to the same level as prior to the well and then it actually gets better after year six, year seven. The cover condition in the immediate vicinity of the well increases. We're not entirely sure why. But we suspect this has got something to do with a change in management of the region immediately surrounding wells, in terms of restriction of grazing and suppression of fire.
Finally, we... Because fire is such an important part of the landscape and the vegetation out there, we compared the effects of wells with the effects of fires. And you can see these graphs the fire has a much greater and longer-term impact on cover ranking and cover condition, than the presence of a well. So in both the dry land and channel locations, cover condition drops a lot, especially in the dry land locations immediately following a fire, and then it takes quite a while perhaps a decade or more to recover to the original levels than before the fire. So the take-home from this is that the biggest impact that gas extraction activities are likely to have on vegetation in this region is through any impacts they might have on natural fire regimes.
14. Gas extraction and vegetation condition
This investigation undertook analysis to determine if existing gas production activities have had a measurable impact on the condition of vegetation cover across the Cooper GBA region.
About the presenter
Dr Randall Donohue
Randall’s main interest is understanding and studying the dynamic, functional nature of vegetation – how plants grow, how they respond to changes in growing conditions and climate, what their role is in the carbon and water cycles. He predominantly uses satellite imagery to undertake his research.
- Bioregional Assessment Program
- Lake Eyre Basin bioregion
- Northern Inland Catchments bioregion
- Clarence-Moreton bioregion
- Northern Sydney Basin bioregion
- Sydney Basin bioregion
- Gippsland Basin bioregion
- Indigenous assets
- Bioregional assessment methodology
- Compiling water-dependent assets
- Assigning receptors to water-dependent assets
- Developing a coal resource development pathway
- Developing the conceptual model of causal pathways
- Surface water modelling
- Groundwater modelling
- Receptor impact modelling
- Propagating uncertainty through models
- Impacts and risks
- Systematic analysis of water-related hazards associated with coal resource development
- Assessment components
- Component 1: Contextual information
- Component 2: Model-data analysis
- Components 3 and 4: Impact and risk analysis
- Component 5: Outcome synthesis
- Metadata and datasets
- Geological and Bioregional Assessment Program