The 2015-16 ENSO event resulted in a period of significant drought across S.E. Asia. Droughts in tropical forests have previously resulted in changes to vegetation phenology, which have been observable in the field and in satellite images. However, whilst much of our understanding comes from undisturbed forests, the impact of droughts on logged and degraded forests remains unclear. What is clear is that the impact is likely to be site-specific, and related to the previous disturbance regime and possibly post-disturbance management. Here we report on the results of the STEED project (Spatio-TEmporal Dynamics of Forest Response to ENSO Drought), research that was undertaken as part of the UK NERC Understanding the Impacts of the recent El Nino programme. By combining observations from hyperspectral in situ remote sensing, high spatial resolution drone and satellite imagery, and LiDAR data, the project sought to provide evidence of the impact of the recent ENSO climate perturbation on both forest structure and functioning across a disturbance gradient, to assess the resilience of previously disturbed forests i.e. their capacity to respond to short-term perturbations by resisting damage and recovering quickly.
Coupled with an extensive field campaign, we first report the results of using a ground-based spectroradiometer to assess the changes in spectral response of leaves from trees that had experienced drought conditions. By acquiring data both at the start of the drought period and subsequently when more typical climatic conditions resumed, we note distinct changes in spectral response associated with changing leaf traits. By analysing 15 different leaf traits (including concentrations of N, C, SLA, P, tannins, lignin, carbohydrate and water content) we relate changes in spectral response to changing leaf biochemistry, and demonstrate that changing traits in response to drought can be detected by hyperspectral remote sensing, and thus pave the way for scaling up observations of tropical forest response and resilience to landscape scales. We also note that the drought stimulated new leaf growth but with reduced SLA.
Additionally, we will report on the use of multi-temporal high spatial resolution satellite imagery to detect tree mortality across the logged and unlogged forest. With imagery acquired during and after the drought, we note no significant evidence of widespread drought-induced mortality for the 2015-16 event, although recommend that further observations are made to account for a lagged response that has been noted elsewhere.
Finally, by acquiring high spatial resolution drone-based imagery a significant increase in liana infestation has been recorded through the drought period. Coupled with LiDAR-derived structure metrics, we note differences in liana infestation in the logged versus unlogged forest, suggesting that prior disturbance can lead to a variable response to drought.
Thus, by combining information from hyperspectral data, high spatial resolution imagery and LiDAR, and validated using ground-based observations, we demonstrate the utility of remotely sensed observations for understanding landscape- to regional-scale changes to forest function and structure in response to short-term climate perturbations. Mapping forest resilience at these scales is vital as forest managers and policy makers still lack the fundamental understanding and practical advice on managing logged tropical forest ecosystems post-disturbance to ensure long-term timber harvest, maintain biodiversity and build carbon stocks for climate change mitigation.

Recent work suggests that episodes of drought and heat can bring forests across climate zones to a threshold for massive tree mortality (Hirota et al., 2011). Across complex systems, the vicinity of a tipping point tends to come with a loss of resilience as reflected in declining recovery rates from perturbations (Flores et al., 2017; Veraart et al., 2011). Trees may be no exception, as at the verge of drought induced death, trees are found to be weakened in multiple ways affecting their ability to recover from stress (McDowell et al., 2018). Using world-wide time series of dense satellite image time series of MODIS and passive-microwave radiometer measurements, we have shown that the tropical forest recovery rates decrease as mean annual precipitation declines to levels known to be critical for tropical forests (Verbesselt et al., 2016) (Fig. 1). This implied independent support for the idea that such forests may have a tipping point for collapse at drying conditions. Moreover, the demonstration that slowing down of recovery rates after disturbances may be detected from satellite data suggests a novel way to monitor resilience of tropical forests, as well as other ecosystems known to be vulnerable to collapse.

We demonstrate preliminary results on how the recovery capacity of tropical forests decreases during the recent 2015/2016 extreme drought in the Brazilian Amazon(Smith et al., 2014). We describe and discuss the implications of these results and the need to validate findings with in-situ measures of tropical forest recovery capacity.

With Sentinel-1 and 2 satellites dense radar and optical time series data at high spatial resolution are available globally. We illustrate new methodological approaches, discuss preliminary results, and related challenges using these data sets to derive accurate and near operational forest resilience measures by directly quantifying disturbance and recovery dynamics (DeVries et al., 2015).


Figure 1. Slowness of dynamics of intact tropical forest as reflected by temporal autocorrelation of NDVI (MODIS 2000-2011) as a function of mean annual precipitation (TRMM MAP, 2000-2011) in South America, Africa, Asia and Australia (Verbesselt et al., 2016). The black solid lines represent the mean curve from an ensemble of six curves (i.e., three detrending methods and two indicators of temporal autocorrelation, see supplements). The gray shaded area highlights the range of the six individual ensemble members. The distribution of the available MAP observations is indicated by the density of the vertical lines above the x-axis.


DeVries, B., Decuyper, M., Verbesselt, J., Zeileis, A., Herold, M., Joseph, S., 2015. Tracking disturbance-regrowth dynamics in tropical forests using structural change detection and Landsat time series. Remote Sens. Environ. https://doi.org/10.1016/j.rse.2015.08.020

Flores, B.M., Holmgren, M., Xu, C., van Nes, E.H., Jakovac, C.C., Mesquita, R.C.G., Scheffer, M., 2017. Floodplains as an Achilles’ heel of Amazonian forest resilience. Proc. Natl. Acad. Sci. U. S. A. 114, 4442–4446.

Hirota, M., Holmgren, M., Van Nes, E.H., Scheffer, M., 2011. Global Resilience of Tropical Forest and Savanna to Critical Transitions. Science (80-. ). 334, 232–235.

McDowell, N., Allen, C.D., Anderson-Teixeira, K., Brando, P., Brienen, R., Chambers, J., Christoffersen, B., Davies, S., Doughty, C., Duque, A., Espirito-Santo, F., Fisher, R., Fontes, C.G., Galbraith, D., Goodsman, D., Grossiord, C., Hartmann, H., Holm, J., Johnson, D.J., Kassim, A.R., Keller, M., Koven, C., Kueppers, L., Kumagai, T., Malhi, Y., McMahon, S.M., Mencuccini, M., Meir, P., Moorcroft, P., Muller-Landau, H.C., Phillips, O.L., Powell, T., Sierra, C.A., Sperry, J., Warren, J., Xu, C., Xu, X., 2018. Drivers and mechanisms of tree mortality in moist tropical forests. New Phytol. https://doi.org/10.1111/nph.15027

Smith, A.M.S., Kolden, C.A., Tinkham, W.T., Talhelm, A.F., Marshall, J.D., Hudak, A.T., Boschetti, L., Falkowski, M.J., Greenberg, J.A., Anderson, J.W., Kliskey, A., Alessa, L., Keefe, R.F., Gosz, J.R., 2014. Remote sensing the vulnerability of vegetation in natural terrestrial ecosystems. Remote Sens. Environ. 154, 322–337.

Veraart, A.J., Faassen, E.J., Dakos, V., Van Nes, E.H., Lürling, M., Scheffer, M., 2011. Recovery rates reflect distance to a tipping point in a living system. Nature 481, 7–10. https://doi.org/10.1038/nature10723

Verbesselt, J., Umlauf, N., Hirota, M., Holmgren, M., Van Nes, E.H., Herold, M., Zeileis, A., Scheffer, M., 2016. Remotely sensed resilience of tropical forests. Nat. Clim. Chang. 6, 1028–1031. https://doi.org/10.1038/nclimate3108

The abundance and dominance of lianas is increasing in tropical forests as a result of elevated CO2 and forest disturbance. Despite of this, little is known of their impact on critical Essential Climatic Variables (ECVs), such as the fraction of Photosynthetic Active Radiation (fPAR). Recent liana removal experiments have provided indirect inferences on the impact of these non-structural elements via observations derived from changes on Leaf Area Index (LAI). Unfortunately, these inferences fall short on the understanding of their impact on fPAR. As such, this work tries to elucidate the impact that lianas may have on fPAR by looking at its temporal dynamics at Tropical Dry Forest (TDFs) sites with and without liana infestation.
This work involved the comparison of light fields in liana and non-liana TDFs forests during the growing and dry season measured using a spatially distributed Wireless Sensor Network (WSN). WSNs are unique technological tools that allow for nano-second synchronization and measurements ECV in the forest floor every 15-min. Node sensor information was collected and send to a node aggregator which in turn send the data to a cyber-infrastructure systems for data quality control, curation, processing and archiving (http://enviro-net.org).
Results suggest that there fPAR is not affected by the presence or absence of lianas during the growing season confirming that fPAR driven by LAI saturation eliminates their effect. However, fPAR is affected during the dry season suggesting that non-photosynthetic materials (e.g. woody component) are the controlling factors of fPAR in tropical dry forests infested with lianas. This effect seems to be also systematic across other TDFs across the Americas, where “darker” forest seems to be present even with the absence of full leave development. The former makers Woody Area Index a key controlling factor not previously documents in the fPAR literature.
Our findings represent a significant contribution to the detection of liana infestation in tropical environments and their impact on ECV such as fPAR, suggesting that the extent of infestation may be easier to detect using remote sensing approaches in TDF than rainforests where a well defined dry season is present.

The ForDroughtDet project is funded by the German Fereral Agency of Agriculture and Food and aims to answer questions related to the adaptation of the forest sector to climate change. The processing chain should detect changes in physiological/morphological status of forest trees under drought stress on large areas in a highly automatic way. Site conditions are the main parameters defining forest growth regions. Especially precipitation and soil condition control the resilience of forest tree species under drought stress. Climate change, especially global warming, are assumed to increase the frequency and magnitude of drought periods in Mideurope. Forest site maps available across Bavaria are spatially not that explicit to represent all variabilities in soil cover and site conditions of a forest stand. The basic idea of the ForDroughtDet project is to identify drought stress symptoms of forest trees at a reversible stage, meaning before reaching a lethal stress level. Such, the system under development should identify forest sites with drought stress endangered trees in advance of a drop out, an information required by forest management for the preferential start of restructure measures. The approach integrates existing geo information from the forest sector with spectral, height and anisotropy information from remote sensing systems of the latest generation in a spatial-temporal context (4D approach). Anisotropy is still a fundamental challenge in remote sensing and must be considered for qualitative data evaluations. A new aspect of the ForDroughtDet project is the exploration of anisotropy features as additional, independent information source.
The known spectral signature reaction on drought stress is seen as main source of information and should be quantified on behalf of the physiologic morphologic measurements provided by the hosting Kranzberger Forest Roof Experiment (KROOF). The second hypothesis of the investigation concept is that water deficit will result in a change of leaf and needle position (shrinking, rolling, leaf angle and exposure change) which might be detectable by exploring anisotropy information from appropriated air- or spaceborne sensors.
The research framework is covering the scale range from on site, air to space level, includes basic data base generation, upscaling, system fusion, reflectance model adaptation and inversion for prognosis provision. At the ‘at site’ scale field gonio-spectrometric measurements are performed with the GonioTree device across the vegetation period providing the BRDF models for beech and spruce trees under artificial drought stress (400–1000 nm), as well as phenologic spectral libraries of forest trees (400-2500nm)(TUM). The simultaneously taken stereo images of the crown segment of the GonioTree measurements are analyzed to derive biometric parameter like leaf area distribution (leaf angle and exposure, size) as well as descriptive parameter assumed to control the directional reflectance (shade and sun lit fraction, etc.)(DLR). These information are employed for reflection model adaptation (UTr). The study site is imaged twice the year by hyperspectral HySpex systems (DLR, UTr) which are operated together with stereo data imaging systems (3-K, EOS-M3 low cost system). Stereo data is delivering singletree crown delineation and shade/sun lit fractions. Hyperspectral imaging is used for tree species identification and status assessment. By inversion of the developed drought stress symptom sensitive reflection models, the detection of drought stress prone patches within larger forest areas should be possible.
The system development chain relies on a combination of (hyper-) spectral and multidirectional data sets from research airplanes (HySpex and 3K). The envisaged operational version should be based on satellite data of the Copernicus Program of the European Space Agency (ESA) (Sentinel-2, Sentinel-3) supported by national missions (Landsat 8, Spot, EnMAP, TanDEM-X, Zi Yuan 3 a,b twins).
The results deliver empirical support for the “stress-gradient hypothesis” (SGH) in environmental science and serve to generate second-generation forest risk maps.

Forests cover about one third on the Earth’s land surface and provide a number of ecosystem services. Among these, forest ecosystems play an essential role within the global carbon dynamics due to their continuous exchange of CO₂ with the atmosphere through photosynthesis. For this reason, understanding and quantifying their imprint on the global carbon budget is critical for interlinking the biosphere and climate systems. This requires spatially-explicit and temporally-repeated information about vegetation photosynthetic activity. Conversely to traditional Remote Sensing (RS) based on reflectance, RS of sun-induced chlorophyll fluorescence (F) is capable of providing information about the actual vegetation functioning, thus offering the possibility of accomplishing this goal.
In this contribution, we explored the potential of airborne-derived red and far-red F as indicator of plants photosynthetic activity in a mid-latitude mixed forest ecosystem. To achieve this, we exploited high spatial resolution (1 m) airborne images acquired with the imaging sensor HyPlant, designed as airborne demonstrator of the forthcoming FLuorescence EXplorer (FLEX) satellite mission of the European Space Agency. Hyplant images were elaborated in order to obtain two independent measures of plant photosynthesis. Firstly, maps of F were successfully obtained at the red and far-red peak (r²=0.74, p<0.001 and r²=0.73, p<0.001, respectively, compared to ground-based retrievals) from ultra-fine spectral resolution HyPlant radiance measurements. Secondly, a map of gross primary productivity (GPP) was derived using a modelling approach based on the use of airborne-derived maps of key forest traits (i.e., leaf chlorophyll content (LCC) and leaf area index (LAI)) to drive the process-based Breathing Earth System Simulator (BESS) ecophysiological model. LCC and LAI were estimated accurately (RMSE=5.66 μg cm-² and 0.51 m² m-², respectively, p<0.001) from HyPlant hyperspectral reflectance measurements using a physically-based radiative transfer model inversion, ensuring the accurate representation of the spatial variability of these determinants of the ecosystem functionality.
The spatial relationship between measured F and modelled GPP was then investigated in order to interpret the variability of ecosystem functioning in a forested area at regional scale. Results showed that far-red F is significantly correlated with GPP (r²=0.46, p<0.001) in the spatial domain, and that this relationship is nonlinear. Conversely, no statistically significant relationship was found between red F and GPP. The empirical relationships observed in this high-spatial resolution analysis evidenced the potential of exploiting F to obtain new insights into the forest ecosystems functioning. In addition, results highlighted the critical role of the spatial heterogeneity in controlling the carbon uptake. This entails the need of taking this aspect into account for an unambiguous interpretation of the F signal and for its operational use as indicator of the plants functional state.

The Sustainable Development Goals were adopted by the UN Member States in September 2015 and comprise the 2030 Agenda for Sustainable Development. There are 17 Goals with 169 associated targets and they are intended to be integrated and indivisible and to balance the economic, social and environmental dimensions of sustainable development. Developing an appropriate monitoring framework for such an ambitious and complex politically negotiated set of Goals and targets. At present there 232 agreed indicators, of which 93 are deemed conceptually clear, with an internationally established methodology and standards, and data are regularly produced by at least 50 per cent of countries where the indicator is relevant. For SDG 15, for example, there are nine indicators for the nine targets, not all of which are considered conceptually clear and with appropriate methodologies.

At the same time, the Parties to the Convention Biological Diversity are now considering what will replace the current Strategic Plan (with its 20 Aichi targets) and that will, therefore comprise the post-2020 biodiversity framework. The outcomes of the CBD’s negotiations in late 2020 will influence the biodiversity relevant Goals and targets in the SDG agenda between 2020 and 2030. This offers considerable opportunities to propose how this complex array of politically negotiated Goals and targets can be monitored most informatively and efficiently.

There is an emphasis on habitat and ecosystem conservation, restoration and management in the current CBD targets and in SDG 15 and others. It is clear, therefore, that a robust monitoring framework that informs understanding of the status of habitats and how they are changing, is both critical and badly needed. There is a range of reasons why a new approach to this has significant potential and includes: a) although the preamble to the SDGs and targets state that they are integrated and indivisible, there is emerging research that shows the connectedness is patchy across the whole network; b) the large number of targets and indicators that have been proposed for biodiversity present a challenge to both countries and global ‘indicator custodians’ in reporting on all of them to the extent expected; and c) there is an emerging desire to explore the potential for a small number of (potentially new) indicators that will inform policy and decision-makers on the state of biodiversity, including habitats and ecosystem services. It is therefore, very timely to consider exactly what ground and space sensors technology and modelling can contribute to monitoring politically Goals and targets in a way that present indictors cannot.