The Nature Assessment Approach
Last updated
Last updated
To measure Land Characteristics, we distinguish between dimensions, indicators, and models. The relationship between these terms is:
Dimensions, Indicators and Models
Dimensions define broad environmental categories.
Indicators provide measurable proxies for these dimensions (KPIs).
Models determine the exact datasets, methodologies, and assessment techniques used to evaluate the indicators and ultimately the dimensions.
In some cases, indicators may be applicable for multiple dimensions, such as soil carbon stocks, which falls both into the carbon and soil dimension.
Dimensions focus on critical aspects of ecosystems, including carbon, soil, water, and biodiversity. Each dimension encapsulates a broad environmental area that we aim to measure and monitor.
Indicators are the specific, measurable variables used to represent the state of each dimension. Indicators are selected based on their ability to effectively measure the dimension they represent. These indicators can range from large-scale remote sensing metrics to detailed on-site measurements.
Models are the tools used to assess these indicators, transforming raw data into meaningful evaluations. Each model specifies the datasets, methodologies, and relationships between inputs and outputs needed to accurately measure an indicator.
Models are continuously updated to improve their accuracy. As new data becomes available and methodologies evolve, new versions of models are released, enhancing our ability to assess each indicator with greater precision.
Data is retrieved through two primary sources: remote sensing (satellite observations) and measurement on the ground. These diverse datasets are integrated into our modelling processes to create ecosystem models.
Remote sensing involves gathering information about ecosystems from a distance using advanced technologies. Key remote sensing methods we utilize include:
Radar (SAR): Penetrates cloud cover and dense vegetation to provide data on surface roughness, biomass, and soil moisture.
Thermal: Captures temperature variations in the land surface, essential for understanding heat stress in plants, water bodies or evaporation.
Optical: Collects data using visible and infra-red light, making it possible to monitor vegetation health, land cover changes, and other environmental features.
Lidar: Uses laser pulses to generate 3D models of landscapes, allowing precise measurement of vegetation height, canopy structure, and terrain.
Ground data collection is optional and serves to validate and enhance the remote sensing data by providing detailed, site-specific insights. Key methods include:
Field Sampling: Direct collection of data such as soil composition, plant health, species presence, and other ecological metrics.
Bioacoustics: Uses audio recordings of animal calls and natural sounds to monitor biodiversity and wildlife presence.
eDNA: Environmental DNA sampling involves extracting genetic material from soil, water, or air to detect the presence of species in an area.
Camera Trapping: Motion-sensitive cameras capture images of wildlife, providing critical data on species presence, behavior, and population trends.
Once data is collected from both remote sensing and ground-based methods, we feed it into various modelling processes. These models are essential for interpreting raw data, simulating ecosystem processes, and making predictions about future environmental conditions:
AI/ML Models: Artificial intelligence and machine learning are employed to analyze large datasets, identify patterns, and continuously improve model predictions.
Direct Mapping: Converts remote sensing data into maps that show specific ecosystem attributes, such as vegetation cover, species distribution, or land use.
Process Models: Simulate natural processes (e.g., carbon cycling, hydrology, and nutrient flows), providing insight into ecosystem functions and dynamics. These models can often also used for projecting future scenarios.
Models are either developed internally by TLG or built externally by academic or commercial partners. Models are validated during a model building phase against reference test data to gauge their accuracy and robustness. Models built by external partners are trained and validated using data provided by TLG as well as data from other sources.
