Kakadu National Park Landscape Symposia Series 2007–2009
|
|  Yüklə 5.42 Mb.
|
6.3 Optimising the quality of weed mapsWith the aim to produce accurate vegetation maps, there are a number of factors that need to be considered in the design and implementation of any remotes sensing monitoring program. Key issues are vegetation phenology, inundation (water in the landscape), cloud cover , fire, smoke, and the spectral resolution of the sensor. The view angle of the satellite sensor can also influence image quality and should be kept as minimal as possible (eg 0–15º off NADIR13). Seasonal variability must be accounted for in any mapping exercise, where changes in plant phenology, and the degree of inundation on floodplains influence the spectral response of plants and other surface features. Image captures should to be planned for times when target weeds and habitats are most clearly resolved (ie when the diagnostic optical differences between targets are optimal). In this regard, eriss has established a spectral database for major weeds and native vegetation in the region. However there remain information gaps with regard to acquiring quantitative spectral profiles of some key aquatic weed species (and native aquatic plants). More comprehensive hyperspectral profiles acquired from ground targets across ranges of seasonal variability in plants will assist in improving knowledge based mapping techniques. For the above reasons remotely sensed mapping of vegetation should occur in conjunction with coincident field reference and validation surveys. Appropriately designed surveys allow the quality of map products to be assessed. Quality control of maps produced from RS is important for providing confidence in the reliability of maps of weed or native vegetation distribution. Image quality and capture times are also influenced by atmospheric conditions and ground disturbance effects. For example, excessive cloud cover during the Wet season necessitates that capture times for optical imagery are limited to the Dry season period. Dry season smoke haze can also effect image quality and higher incidence of fire on wetlands is in the late dry season means that these times should be avoided when wetland weed mapping is the aim. In general it is recommended that capture times be limited to the early to mid-Dry season (April to June) period, when water levels have receded substantially, but remain high enough to allow some access by airboat for field surveys. If airboat access is not possible, a helicopter is a suitable alternative for field validation surveys. It should be noted that those with field (and aerial survey) based experience in wetland environments and weed identification can also provide useful advice on the best times for image capture. Given the variability of wetlands between years, a reconnaissance field survey (undertaken in the late-wet/early-dry season) may be useful in anticipating optimal image capture times in any one year.
6.4 Applications of mapping informationVHR is perhaps the most effective technology at detecting small ‘satellite’ weed colonies. Early detection and eradication of small ‘satellite’ weed incursions14 is important to minimise costs of weed management. Early detection of Olive Hymenachne outbreaks, that have tended to occur in relatively isolated and unpredictable locations within KNP, would likely be facilitated by applying VHR. In situations where large weed infestations occur, VHR also can provide information for rehabilitation of key habitats where the effectiveness of weed control strategies aimed at reducing the density or extent of weeds may be evaluated. Some of the potential applications of VHR and the status of these in terms of development towards operational monitoring in KNP are provided in Table 1. Table 1 The authors’ assessment of potential applications of remote sensing for monitoring key wetland habitats (landscape management units), the impacts of weeds and other environmental pressures; and the current status in terms of development towards an operational monitoring framework.
Note: R & D = Research and Development |