The analysis indicated that significant inroads can be made into the required sediment yield reductions required to meet the interim water quality targets through prioritising the top ~ 260 large alluvial gullies . However , the results showed that the Reef 2050 WQIP target of a 50 % reduction in fine sediment supply ( for the Burdekin Basin ) will not be reached by targeting the single high yielding gullies alone , and that remediation of apparently less cost-effective gullies further along the cost curve would have to be undertaken . The most cost-effective way to remediate these less cost-effective gullies is to target clusters which include high yielding single gullies , treating whole clusters ( including lower yielding gullies ) at the outset . This would appear to make these sites less cost-effective in the short term , but it would be significantly more cost-effective than attempting to treat lower yielding gullies later when mobilisation costs cannot be managed . The model was highly sensitive to the effectiveness of low intensity treatments , which also have a higher risk of failure . There is currently considerable uncertainty as to the remediation effectiveness ratios associated with this type of gully treatment .
1 . The five sites for which detailed site analyses have already been completed have all been prioritised within the top 73 clusters across the catchment . These sites could all proceed to the design phase for on-ground works as soon as possible .
2 . More detailed analysis should be undertaken at one of the sites ( from the five detailed sites outlined in Recommendation 1 ). The broad analysis indicates 2-3 clusters with potential for intensive rehabilitation , but the detailed report suggests across the site lower intensity treatments are more appropriate for most gullies .
3 . Many of the mapped gullies have not been ground-truthed . Therefore , it is highly recommended that field data is fed back into the analysis as part of the site evaluation process for developing detailed site rehabilitation designs . This process will rigorously evaluate whether on-ground observations match the predictions ( in terms of gully activity , sediment yield predictions , the priority for rehabilitation , costs estimates , logistics , soil materials , etc .). The gully classification developed through NESP and the Queensland Water Monitoring Network would be an appropriate framework for systematically gathering these data .
4 . Hillslope gully yield estimates may be overestimated . An unavoidable limitation of the analysis was a reliance on available data to train the modelled sediment yields , which had a higher proportion of alluvial gullies . Further efforts are required to validate predictions and quantify hillslope gully sediment yields , with the potential to revise sediment and cost estimates from these gullies .
5 . Detailed site analysis should proceed on the next set of priority clusters with the number of sites determined by the potential available funding . The economic analysis would suggest that an initial phase of 20 cluster sites should be the focus for detailed site assessment , followed by a further 20 in a second phase .
6 . It should be recognised that the sediment yields assigned to the 22,311 mapped gullies in the study area are empirically modelled estimates , albeit based on derived highresolution datasets ( e . g . LiDAR DEM of Difference ( DoD ), which is a comparison between DEMs acquired at different times , derived sediment yields from a training dataset of ~ 1,200 gullies , LiDAR and satellite image derived gully metrics , soil materials extrapolated from 90 field sample sites and satellite image derived gully soil reflectance ). Therefore , it was recommended that each of these empirical datasets should be augmented and refined to improve the reliability of the empirical modelling . In particular :
• The value of repeat LiDAR cannot be underestimated – any opportunity to capture repeat LiDAR should be progressed , particularly focused on alluvial areas and gently undulating hillslopes . Such
90