An automated remote sensing approach for monitoring hydrological impacts of vegetation cover changes

Landscapes are constantly changing under the pressure of human activities or climate instability. In particular, variations in vegetation cover have a significant impact on the hydrological cycle and the response of the landscape to hydrometeorological forcing. Greenery depletion caused by human activities, such as deforestation for agricultural purposes or uncontrolled urbanisation, or by climate, such as desertification, can lead to an unbalanced hydrological partitioning, exacerbating runoff production. In recent years, an increased awareness and concern about more frequent climate extremes has led scientists and decision-makers to seek solutions for restoring degraded ecosystems. Among these, targeted land use and land cover (LULC) changes, such as regreening initiatives aimed at increasing vegetation coverage and the associated ecosystem services (ES), have been shown to be effective for restoration purposes. Once planned and implemented, it is imperative to have tools to assess the effectiveness of such LULC changes and the associated impacts on the hydrological behaviour of the restored landscape.

The aim of this study is to present HydroSENS, an algorithm for the automated tracking of the spatio-temporal evolution of vegetation based on multispectral satellite imagery. Furthermore, the algorithm can be adopted for the monitoring of water-related ES concerning infiltration capacity and runoff management.  

HydroSENS allows the user to evaluate the composition of the satellite imagery, calculating the greenery fraction and properties through a spectral unmixing analysis, and retrieve hydrological parameters at the sub-pixel scale. Thus, these features make it a flexible tool for quantitatively understanding the impacts of LULC changes on hydrological processes and associated water-related ES.

The approach has been used to assess and monitor the effectiveness of a regreening project on a portion of land severely afflicted by land degradation as a result of unsustainable land management and climate changes in Tanzania. The project promoted the adoption of agroforestry practices by implementing rainwater harvesting techniques to improve localized water retention and increase the likelihood of survival of seasonal and perennial vegetation.

The site has been monitored for several years by analysing Sentinel-2 images acquired during both the wet and dry seasons. The monotonous positive trend of the Normalized Difference Vegetation Index (NDVI) and vegetation fraction over the period, in both the wet and dry seasons, indicates an increase in healthy vegetation.

The LULC alteration significantly influences the hydrological processes at the site scale, implying a high infiltration rate and an overall reduction in runoff. The benefits are twofold: first, better runoff management, reducing the consequences of flooding during heavy storms; second, larger water intake, decreasing water stress during dry spells. In addition, these regreening-induced effects improve the evapotranspiration capacity of the vegetation, maximising the crop yield as a favourable implication for the local populations.