干旱区生态环境知识资源中心

Global warming is affecting extreme events, such as floods, droughts, heat waves and bushfires, with important consequences for human communities and ecosystems. Large-scale reforestation, especially in the tropics and sub-tropics, has the potential to help mitigate this warming trend and the projected increases in extreme events. This raises an important question for Australia: what is the potential of large-scale reforestation to help mitigate projected changes in climate during the 21st Century? An important first step in addressing this question is to accurately represent key land surface characteristics in a climate model.

We evaluated land use and land cover data currently used in the CSIRO variable resolution Conformal-Cubic Atmospheric Model (CCAM) in preparation for simulating the climate impacts of large-scale reforestation in Queensland. CCAM land-use datasets are based on Boston University MODIS Collection 4 LAI and the IGBP (International Geosphere-Biosphere Programme) land use classification datasets. These datasets were adapted for use with the CSIRO Atmosphere Biosphere Land Exchange (CABLE) land surface model that is coupled to CCAM.

New land use datasets for Australia recently became available which capture land use changes more accurately than the existing datasets used in CCAM. This paper describes an updated land-use dataset for running the CABLE model within CCAM to represent land use and land cover conditions in Australia. Analysis shows that the IGBP dataset currently used in CCAM does not incorporate recent land cover and land use changes in eastern Australia. For this reason, we updated the IGBP land cover data set in the Australian region using the National Vegetation Inventory System (NVIS) dataset which includes the current extent of broad vegetation classes for the state and also a more recent cropping data layer.

The LAI (Leaf Area Index) dataset currently used in CCAM was adapted from the Boston University product, with further improvements by CSIRO to derive a prototype monthly LAI dataset. This dataset was based on the MODIS Collection 4 product (2001-2005), which has been shown to have low accuracy in semi-arid regions. This limitation prompted us to evaluate the Community Land Model (CLM) and Beijing Normal University (BNU) LAI products. We found that the BNU LAI performed better than the CCAM or CLM LAI datasets for Queensland. In addition, the BNU dataset is derived from MODIS Collection 5 and captures LAI for the period 2000-2009, thereby providing a longer-term estimation of LAI and was subsequently adopted for use in the CCAM model.

The evaluation of existing land cover, land use and LAI datasets in the CCAM model has identified more recent and improved datasets as being suitable for updating the land cover, land use and LAI in CCAM. The impact of these updated LULC and LAI datasets is currently being evaluated in the CCAM simulations, and the results from this evaluation will be presented at the conference.