Chlorophyll Estimation in Mangrove Forest as a Potential Indicator of Vegetation Condition.
Wandera, Loise1; Verhoef, Wout2; Fauzi, Anas2; Schlerf, Martin1
1CRP Gabriel Lippmann, LUXEMBOURG; 2University of Twente, ITC, NETHERLANDS

Mangroves are forest communities predominantly found in habitat zones characterised by limited nutrient supply and high salinity levels i.e. tropical/subtropical coastal environment and estuarine tidal or intertidal areas. In spite of the unfavourable conditions, the mangroves are able to attain and maintain optimum productivity that compares well to other tropical forest stands. However it has been reported that the current rates of human activities taking place within mangrove ecosystem has increased tremendously and the results are threatening environmental stressors to mangrove vegetation. Studies have demonstrated that, altering mangrove ecosystem destabilises their survival strategies and increases their vulnerability to variation from the norm in any abiotic factors like rainfall seasonality, temperature patterns and salinity fluctuations. With this revelation stakeholders are obliged to explore practical and effective methods of monitoring the physiological status of these forests at a given time to enable formulating and enforcing laws that will enhance sustainable use of the forest ecosystem resources by the communities and safeguarding its functionality in the long run. In mangrove forest studies, remote sensing has been a commonly used technique however, with more focus on species identification, forest delineation and a handful studies related to biophysical and biochemical attribute retrieval. This situation creates a gap for the development of effective management and conservation strategies. The missing link is reliable information that can be linked to physiological status of the vegetation and at the same time on prevailing rates of ecosystem processes e.g. nutrients and carbon cycling.

Chlorophyll (Chl) is a leaf biochemical element that on its own accord has been used in many disciplines other than remote sensing as an indicator of plant primary productivity and a proxy of predicting biogeochemical processes in ecosystems like agricultural fields, and forests. Plant Chl has been found to significantly vary when exposed to different natural and anthropogenic conditions, and for this reason a good parameter choice for ecosystem studies. Researchers have used various approaches to estimate Chl in the past, however, remote sensing images have become a lucrative option since they cater for issues of scale, costs, flexibility on method and most importantly Chl distribution dynamics in a canopy can be well observed in the visible domain of optical image reflectance .There are existing algorithms that can be used to link variations in spectral reflectance pattern displayed in an image to Chl in plants which either rely on establishing statistical linkage between spectral reflectance and vegetation attribute using regression equations or physical approach that adhere to laws of energy transfer within a medium.

This study uses a physical model or radiative transfer model (RTM) referred to as the Soil Leaf Canopy Model (SLC). SLC is composed of three sub models for Soil (4soil) for the leaves (Prospect) and for the canopy (4SAIL2) in addition to sun and sensor angles geometry. An outstanding advantage of the SLC as an RTM is that it incooperate structural and architectural designs in a vegetation canopy when simulating the reflectance e.g. clumping, leaf angle, brown materials, vertical distribution of green and brown leaves, hot spot and soil moisture that makes representation of canopy characteristics from image data to ground observation more realistic. The model was parameterized to suit mangrove canopy characteristics and sensitivity analysis was conducted to ensure that the model was responsive enough to slight variations in canopy variable, a desirable attribute for any modelling set up. The mangrove spectral reflectance was then simulated followed by an inversion by use of Look up Table (LUT) applied to hyperspectral image and Chl distribution map was generated. The Chl obtained from model inversion was compared to field measurements to assess the quality of the model derived Chl estimates.

The objective of this work was to assess the feasibility of using remote sensing techniques as an alternative to ground survey methods that is normally constrained by accesibilty to retrieve vital vegetation parameter, in this case Chl which is a potential indicator of plant and ecosystem condition.