Estimating terrestrial gross primary productivity in water limited ecosystems across Africa using the Southampton Carbon Flux (SCARF) model.

The amount of carbon uptake by vegetation is an important component to understand the functioning of ecosystem processes and their response/feedback to climate. Recently, a new diagnostic model called the Southampton Carbon Flux (SCARF) Model driven by remote sensing data was developed to predict terrestrial gross primary productivity (GPP) and successfully applied in temperate regions. The model is based on the concept of quantum yield of plants and improves on the previous diagnostic models by (i) using the fraction of photosynthetic active radiation absorbed by the photosynthetic pigment (FAPARps) and (ii) using direct quantum yield by classifying the vegetation into C3 or C4 classes. In this paper, we calibrated and applied the model to evaluate GPP across various ecosystems in Africa. The performance of the model was evaluated using data from seven eddy covariance flux tower sites. Overall, the modelled GPP values showed good correlation (R>0.59, p<0.0001) with estimated flux tower GPP at most sites (except at a tropical rainforest site, R=0.38, p=0.02) in terms of their seasonality and absolute values. Mean daily GPP across the investigated period varied significantly across sites depending on the vegetation types from a minimum of 0.44gCm-2day-1 at the semi-arid and sub-humid savanna grassland sites to a maximum of 9.86gCm-2day-1 at the woodland and tropical rain forest sites. Generally, strong correlation is observed in savanna woodlands and grasslands where vegetation follows a prescribed seasonal cycle as determined by changes in canopy chlorophyll content and leaf area index. Finally, the mean annual GPP value for Africa predicted by the model was 35.25PgCyr-1. The good performance of the SCARF model in water-limited ecosystems across Africa extends its potential for global application.

Combining LiDAR and IKONOS data for eco-hydrological classification of an ombrotrophic peatland.

Remote sensing techniques have potential for peatland monitoring, but most previous work has focused on spectral approaches that often result in poor discrimination of cover types and neglect structural information. Peatlands contain structural "microtopes" (e.g., hummocks and hollows) which are linked to hydrology, biodiversity and carbon sequestration, and information on surface structure is thus a useful proxy for peatland condition. The objective of this work was to develop and test a new eco-hydrological mapping technique for ombrotrophic (rain-fed) peatlands using a combined spectral-structural remote sensing approach. The study site was Wedholme Flow, Cumbria, UK. Airborne light dectection and ranging (LiDAR) data were used with IKONOS data in a combined multispectral-structural approach for mapping peatland condition classes. LiDAR data were preprocessed so that spatial estimates of minimum and maximum land surface height, variance and semi-variance (from semi-variogram analysis) were extracted. These were assimilated alongside IKONOS data into a maximum likelihood classification procedure, and thematic outputs were compared. Ecological survey data were used to validate the results. Considerable improvements in thematic separation of peatland classes were achieved when spatially-distributed measurements of LiDAR variance or semi-variance were included. Specifically, the classification accuracy improved from 71.8% (IKONOS data only) to 88.0% when a LiDAR semi-variance product was used. Of note was the improved delineation of management classes (including Eriophorum bog, active raised bog and degraded raised bog). The application of a combined textural-optical approach can improve land cover mapping in areas where reliance on purely spectral discrimination approaches would otherwise result in considerable thematic uncertainty.

Linking remote sensing, land cover and disease.

Land cover is a critical variable in epidemiology and can be characterized remotely. A framework is used to describe both the links between land cover and radiation recorded in a remotely sensed image, and the links between land cover and the disease carried by vectors. The framework is then used to explore the issues involved when moving from remotely sensed imagery to land cover and then to vector density/disease risk. This exploration highlights the role of land cover; the need to develop a sound knowledge of each link in the predictive sequence; the problematic mismatch between the spatial units of the remotely sensed and epidemiological data and the challenges and opportunities posed by adding a temporal mismatch between the remotely sensed and epidemiological data. The paper concludes with a call for both greater understanding of the physical components of the proposed framework and the utilization of optimized statistical tools as prerequisites to progress in this field.