Linking evaporation and vegetation characteristics: a data-driven study from plot to global scale

Anne J. Hoek van Dijke

Research output: Thesisinternal PhD, WU


Evaporation and vegetation are closely linked. Vegetation impacts the evaporation by enhancing the water and energy availability, and by increasing the aerodynamic conductance. On the other hand, vegetation can decrease evaporation through stomatal closure. This thesis addresses the statistical correlation between evaporation and the vegetation green biomass and the effect of vegetation changes on evaporation. The first objective of this thesis is: 'to study the link between ground-based observations of evaporation and satellite remote sensing observations of vegetation'. Satellite observations of the vegetation green biomass, such as the leaf area index or vegetation indices, are integrated in most remote sensing based evaporation models. It is however unknown whether the link between remote sensing observations of the vegetation green biomass and evaporation are valid over different vegetation types, climate zones, and across spatial-temporal scales. Chapter 2, 3, and 4 study the link between remote sensing observations of vegetation and ground-based observations of evaporation. The second objective of this thesis is: 'to study the effect of drought and land-cover change on evaporation over large areas'. Global change, including an increasing number of droughts (chapter 4) and land-cover change (chapter 5) will have an impact on future evaporation and the water, energy, and carbon balance. A better understanding of the impact of vegetation on evaporation under global change is therefore crucial to study future global evaporation.

Chapter 2 studies the link between daily mean tree sap velocity and Landsat derived NDVI in a catchment in Luxembourg. The studied hypothesis is 'there is a positive correlation between the NDVI and sap velocity', because forests with a higher leaf biomass are expected to have a higher sap velocity. However, we found a positive correlation between sap velocity and NDVI only in April during the phase of vegetation green-up. During the rest of spring and summer, we found a significant negative correlation for half of the studied days. During a dry summer, sap velocity was uncorrelated with NDVI, but varied with water availability and soil type. Methods using the NDVI to predict or scale (evapo)transpiration should be carefully applied in temperate forest ecosystems.

Chapter 3 studies the link between the yearly mean satellite-derived LAI and the latent heat flux, sensible heat flux, and GPP for a range of vegetation types. The ecosystem fluxes were derived from FLUXNET data and the LAI data was available from MODIS. In this chapter we show that the link between LAI and the latent and sensible heat fluxes depends on the vegetation type and aridity. Under arid conditions, the link between the LAI and water and energy fluxes was strong, but in energy-limited forests, there was no correlation between LAI and water and energy fluxes. In contrast to the water and energy fluxes, GPP was always positively correlated with LAI. For savanna and arid grassland, the LAI can be useful to model or extrapolate water fluxes, but for deciduous broadleaf forest and evergreen needleleaf forest, the LAI is of limited use.

Forest and grassland have a different drought coping strategy. Trees control their stomata to reduce water loss, while grasslands take the risk to lose their aboveground biomass. Both the stomatal control and the reduction in green biomass reduce the surface conductance (Gs) and evaporation. Chapter 4 studies how different MODIS vegetation and drought indices reflect the reduction in Gs in forest and grassland. We show that for grassland, the different optical and thermal indices were sensitive to the reduction in Gs. For the forest sites, the optical indices were not sensitive to the reduction in Gs, but the thermal indices did reflect the reduction in Gs. The results were however not uniform across all forest and grassland sites. A different strategy is required in order to monitor the effects of drought on the water, energy, and carbon cycle in grassland and forest.

Chapter 5 studies the effect of large-scale tree restoration (totalling 900 million hectares) on evaporation, precipitation, streamflow, and water availability. Large-scale tree restoration increases local evaporation, and increases (downwind) precipitation through evaporation recycling. We show that the combined effects of increasing evaporation and increasing precipitation create complex patterns of decreasing (up to 38%) or increasing (up to 6 %) water availability. The effect on large river basins is diverge: several rivers could lose 6 % of their streamflow, while for other rivers, the increased evaporation would be counterbalanced by enhanced evaporation recycling. Tree restoration significantly shifts terrestrial water fluxes and future tree restoration strategies should consider these effects.

Chapter 6 synthesises the results of the four core chapters. The main conclusions are: 1) a significant correlation between ground-based observations of evaporation and satellite remote sensing observations of vegetation is an exception rather than the rule. Water availability seems to play an important role in the slope, direction, and strength of the correlation. This has implications for remote sensing based evaporation monitoring. 2) Drought and land-cover change had an impact on evaporation, and these effects propagate further into the water, energy, and carbon balance.

Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Wageningen University
  • Herold, Martin, Promotor
  • Teuling, Ryan, Promotor
  • Mallick, K., Co-promotor, External person
  • Machwitz, M., Co-promotor, External person
Award date10 Jun 2022
Place of PublicationWageningen
Print ISBNs9789464471915
Publication statusPublished - 10 Jun 2022


Dive into the research topics of 'Linking evaporation and vegetation characteristics: a data-driven study from plot to global scale'. Together they form a unique fingerprint.

Cite this