About project 7

Despite the past years’ continuous developments and advancements of earth observation sensors and missions, of classification algorithms, and of computing capacities for big data processing, the currently available global datasets of water extent and snow cover dynamics do not yet fulfill the requirements to be optimally integrated into the C/DA approach. One major aspect is that available datasets are not available at a sufficient spatial and temporal resolution or do only cover short time periods. Many global datasets are static, i.e. representing only the average situation of water extent during a certain time period, e.g. one year. Dynamic datasets are either very coarse in their original spatial resolution, resulting in low data quality, they cover only short time periods, or are not available consistently in space and time (data gaps). Furthermore, while global average accuracies are available for most of the datasets, none of the currently available state of the art products on water extent and snow cover provide uncertainty estimates at pixel level. While the DLR’s Global WaterPack and the Global SnowPack have optimal temporal and spatial resolutions and coverages to be flexibly aggregated and thus to be used at various spatial and temporal scales for integration and validation of C/DA and WaterGAP, challenges remain to improve the detection of particular types of water (frozen, sediment-rich, polluted etc.), and to quantify uncertainty in a way that is suitable for integration with radar altimetry, and for the integration into the C/DA approach.

The DLR’s Global WaterPack (GWP) has been presented by Klein et al. in 2017. This MODIS-based global, dynamic, multiannual product at 250 m spatial and near-daily temporal resolution exploits the full 250 m archive of daily MODIS time series, and is of unprecedented temporal and spatial resolution and spatio-temporal consistency.

Most approaches of surface water detection from optical data take advantage of the characteristics of water to strongly absorb at visible, near-infrared (NIR), and shortwave-infrared (SWIR) wavelengths (Feyisa et al. 2014; Ji et al. 2009; Li et al. 2013; Sun et al. 2012). Deep, clear water bodies have very distinct spectral signatures and can be easily delineated from the surrounding land surface. However, the situation is more complex for shallow water bodies and for water influenced by algal blooms, submerged vegetation, pollution, sediment-load or turbidity (Verpoorter et al. 2012). Shallow or sediment-rich inland water bodies which feature for example similar spectral characteristics as soils generally cause higher misclassification rates (Jain et al. 2006; Fichtelmann and Borg, 2012). Furthermore, confusions can occur with other land cover types that have similar spectral characteristics as water. Among those are shadows of clouds and mountains, burned areas, or bare surfaces of dark geology and soils.

The DLR’s SnowPack is a considerably enhanced version of the MODIS snow product capturing snow cover patterns at a spatial resolution of 500 m. This global daily dataset on snow cover and annual snow cover statistics (e.g. snow cover duration) has been presented by Dietz et al. (2015).

Snow cover extents can be derived from optical and radar types of satellite remote sensing data. The high spectral reflectivity of snow at visible wavelengths and low reflection in the region of the near-infrared can be used for snow cover detection from optical data. The scattering characteristics of snow depend on many different factors such as snow grain size and shape, or liquid water content, and can be also used for snow cover detection from passive microwaves or SAR systems. The major challenge in mapping snow cover dynamics from optical sensors is the unavailability of land surface reflectance information during periods of cloud cover, and the complexity of discriminating between snow and clouds, where clouds with ice content are particularly challenging. The typical decline of snow reflectance towards the shortwave-infrared can be used for an accurate discrimination between snow and clouds. Based on these spectral characteristics, the Normalized Difference Snow Index (NDSI) in combination with other indices such as NDVI has been widely used to map snow extent from optical data. A combination of optical and SAR data for snow cover delineation can be used to improve spatial coverage and resolution, and to reduce data gaps due to the cloud independency of radar data.

The central hypothesis of this individual project is that improving the quality and the uncertainty description of satellite-based observational datasets of water extent and snow cover can considerably contribute to a better understanding of our global freshwater system. Further enhancement of delineation algorithms will also advance the reliability and validity of earth observation products. The development and provision of sound uncertainty information will enable the integration of these products by combination with radar altimetry and via C/DA approaches into the hydrological model WaterGAP. We hypothesize that in this way both model outputs and earth observation based datasets will be improved, and thus contribute to a better understanding of the global freshwater system.


Dietz A. J., Kuenzer, C., Dech, S. (2015) Global SnowPack: a new set of snow cover parameters for studying status and dynamics of the planetary snow cover extent, Remote Sensing Letters, 6-11.

Feyisa, G. L., Meilby, H., Fensholt, R., Proud, S. R. (2014): Automated Water Extraction Index: A new technique for surface water mapping using Landsat imagery, Remote Sensing of Environment, 140: 23-35.

Fichtelmann, B., Borg, E. (2012): A New Self-Learning Algorithm for Dynamic Classification of Water Bodies. In: Murgante, B., Gervasi, O., Misra, S., Nedjah, N., Rocha, A.M.A.C., Taniar, D., Apduhan, B.O. (Eds.), Computational Science and Its Applications – ICCSA 2012, Part III, Salvador de Bahia, Brazil: Springer: 457-470.

Ji, L., Zhang, L., Wylie, B. (2009): Analysis of Dynamic Thresholds for the Normalized Difference Water Index, 75(11), 1307-1317.

Jain, S.K., Saraf, A.K., Goswami, A., Ahmad, T. (2006): Flood inundation mapping using NOAA AVHRR data, Water Resources Management, 20(6): 949-959.

Klein, I., Gessner, U, Dietz A. J., Kuenzer C. (2017): Global WaterPack – A 250 m resolution dataset revealing the daily dynamics of global inland water bodies. Remote Sensing of Environment, 198: 345-362.

Li, S., Sun, D., Goldberg, M., Stefanidis, A. (2013): Derivation of 30-m-resolution water maps from TERRA/MODIS and SRTM, Remote Sensing of Environment, 134: 417-430.

Sun, F., Sun, W., Chen, J., Gong, P. (2012): Comparison and improvement of methods for identifying waterbodies in remotely sensed imagery, International Journal of Remote Sensing, 33(21): 6854-6875.

Taniar, D., Apduhan, B.O. (Eds.), Computational Science and Its Applications – ICCSA 2012, Part III, Salvador de Bahia, Brazil: Springer: 457-470.

Verpoorter, C., Kutser, T., Tranvik, L. (2012): Automated mapping of water bodies using Landsat multispectral data. Limnology and Oceanography: Methods, 10: 1037-1050.