Woody Above-Ground Biomass Estimation on Abandoned Agriculture Land Using Sentinel-1 and Sentinel-2 Data

Abandoned agricultural land (AAL) is a European problem and phenomenon when agricultural land is gradually overgrown with shrubs and forest. This wood biomass has not yet been systematically inventoried. The aim of this study was to experimentally prove and validate the concept of the satellite-based estimation of woody above-ground biomass (AGB) on AAL in the Western Carpathian region. The analysis is based on Sentinel-1 and -2 satellite data, supported by field research and airborne laser scanning. An improved AGB estimate was achieved using radar and optical multi-temporal data and polarimetric coherence by creating integrated predictive models by multiple regression. Abandonment is represented by two basic AAL classes identified according to overgrowth by shrub formations (AAL1) and tree formations (AAL2). First, an allometric model for AAL1 estimation was derived based on empirical material obtained from blackthorn stands. AAL2 biomass was quantified by different procedures related to (1) mature trees, (2) stumps and (3) young trees. Then, three satellite-based predictive mathematical models for AGB were developed. The best model reached R-2 = 0.84 and RMSE = 41.2 t center dot ha(-1) (35.1%), parametrized for an AGB range of 4 to 350 t center dot ha(-1). In addition to 3214 hectares of forest land, we identified 992 hectares of shrub-tree formations on AAL with significantly lower wood AGB than on forest land and with simple shrub composition.

Automated summary

This study validates the concept of satellite-based estimation of woody above-ground biomass in abandoned agricultural land in the Western Carpathian region. By creating integrated predictive models through radar and optical multi-temporal data, an improved AGB estimate was achieved.