The spatio-temporal ndvi analysis for two different Australian catchments


Kumari N., Yetemen Ö., Srivastava A., Rodriguez J. F., Saco P. M.

23rd International Congress on Modelling and Simulation - Supporting Evidence-Based Decision Making: The Role of Modelling and Simulation, MODSIM 2019, Canberra, Avustralya, 1 - 06 Aralık 2019, ss.958-964 identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.36334/modsim.2019.k3.kumari
  • Basıldığı Şehir: Canberra
  • Basıldığı Ülke: Avustralya
  • Sayfa Sayıları: ss.958-964
  • Anahtar Kelimeler: Aspect, Insolation, Ndvi, Remote sensing
  • İstanbul Teknik Üniversitesi Adresli: Evet

Özet

Contrasts in insolation lead to the development of aspect-controlled ecosystems characterized by heterogeneity in vegetation type and density in semi-arid ecosystems. The aspect-controlled solar radiation creates variation in the type and amount of vegetation across the two opposite facings of the hillslopes. In the Southern Hemisphere (SH), the north-facing slopes (NFS) have an abundance of paleotropical xeric biota, whereas the south-facing slopes (SFS) have higher densities of mesic temperate species. The reverse patterns are mostly observed in the Northern Hemisphere (NH). In the SH, SFS are dominated by the evergreen sclerophyllous woodland, while open scrub vegetation with spiny shrubs, sub-shrubs, and small trees exist on the NFS. This general vegetation pattern creates differences in erosion control and resistance on different slopes, and thus the underlying landscapes evolve differently. Although many previous studies have focused on aspect-controlled vegetation growth in the NH, there have been limited studies in the SH, especially in Australia. Remote sensing provides one of the best options to capture the long-term biomass changes over the large spatial and temporal coverage. The normalized difference vegetation index (NDVI) is based on the relationship between the reflectance of the red and near-infrared bands of multispectral sensors, and it can be used due to its computational simplicity and easy accessibility. In this study, we considered two catchments, Mount Wilson, South Australia and Risdon Hills in Tasmania to study the long-term spatial and temporal variation in NDVI at these catchments. Both sites are unaffected or minimally affected from anthropogenic activities upon visual inspection through Google EarthTM, in addition to reviewing both sites from the literature. We also explored how the precipitation and potential evapotranspiration patterns at these sites affect the vegetation growth during the year. In this study, we extracted NDVI values derived from Landsat 5, 7, and 8 (obtained from Google Earth Engine) for a 18-year period (2000-2017) for both catchments. Thereafter, we used 30-m SRTM DEM to calculate the aspect and slope datasets for two locations. With the aspect data classified, the vegetation index NDVI is computed for each slope, NSF and SFS. We compared and contrasted the inter-annual variability in NDVI at the two sites to capture the temporal variation in NDVI. We have also introduced NDVIdiff as the difference between NDVI at NFS to SFS, where NDVIdiff > 0 states that NDVI is higher on NFS than SFS and vice-versa. The spatial NDVI is extracted for the summer and winter months, November and June, respectively, to see the seasonal NDVI at each catchment. The results show that the Mount Wilson site (~35°S) has higher NDVI values than the Risdon Hill site throughout the year though receiving similar annual precipitation. It is observed that the Mount Wilson site shows approximately similar NDVI on NFS and SFS in the austral summer period. However, in the winter season when seasonal total precipitation exceeds total PET demand, the NDVI on NFS is comparatively higher than on SFS, which is attributed to differences in vegetation phenology on opposing hillslopes and relatively more incoming solar radiation on NFS than SFS. On the other hand, the site at Risdon Hills (~42°S) has relatively lower range of NDVI at both NFS and SFS, and NDVI at NFS and SFS does not vary noticeably. Further, the spatial NDVI patterns at both locations also illustrate similar behaviour, following the temporal patterns at both locations.