Periodic Reporting for period 1 - Emerald (Early warning systeM for soil dEgRadation: a statistical physics Approach to cLimate change aDaptation)
Berichtszeitraum: 2022-06-01 bis 2024-03-31
Climate change is an urgent challenge and impacts ecosystems, society and economic activities. Here we focus on the problem of loss of fertile soil (soil degradation), a process that is induced and accelerated by extreme weather and climate change and affects, among other sectors, agriculture, land use planning and food security. Current tools to monitor soil health for commercial purposes or policymaking are either limited in scale due to high cost, or insufficiently accurate due to the lack of underlying robust models. The EMERALD Proof of Concept aims to test and further develop a novel method to detect soil stress founded in the statistical physics of disordered systems. The use of earth observation data ensures a scalable solution to tackle the problem of desertification globally.
The spatial distribution of vegetation has been analysed through a numerical stochastic model and through the analysis of land data from satellite images. The model is known to exhibit three phases, namely showing contiguous, fragmented, and degraded vegetation, corresponding to increasing environmental stress. Considering samples of 30km in size, we looked at the vegetation fraction and the fraction of contiguous vegetation in each sample with the aim of identifying these processes in real land data.
We have shown that the model's description of the vegetation dynamics can be qualitatively compared to the land data even with relatively low resolution data and with some necessary simplification in the model. The values of the vegetation fraction and the largest cluster size allowed us to offer a qualitative categorization of each sample between the three scenarios of contiguous, fragmented and degraded vegetation.
We have shown that the model's description of the vegetation dynamics can be qualitatively compared to the land data even with relatively low resolution data and with some necessary simplification in the model. The values of the vegetation fraction and the largest cluster size allowed us to offer a qualitative categorization of each sample between the three scenarios of contiguous, fragmented and degraded vegetation.
The method developed in this Proof of Concept has been developed and tested on large scale Earth observation images spanning several European countries. The potential impacts involve the increased ability to monitor soil quality, which will eventually directly benefit farmers, governmental agencies for land use, food and emergency response, agricultural organizations and insurance companies, as well as indirectly having an impact on environmental aspects (land conservation) and social policies (food security and peoples’ well-being).