Modelo de crescimento de gramíneas adaptado a áreas urbanas

Abstract

Green spaces in urban areas and its impacts on the population has been the subject of many studies, not only because of the important ecosystem services provided by these areas but also due to the direct contribution on public health. Initiatives for control of urban vegetation, regarding the optimization of the cutting/mowing process, reducing costs and potentially reducing impacts on the urban natural ecosystem have not been found in the literature. Thus, this work aims to implement a grass growth model suitable for appropriate management of urban green spaces, specifically in lawns, public parks squares, and roadsides and around waterways. The model was developed in Python and simulates the daily dynamics of leaf area index (LAI), biomass, evapotranspiration and soil water content, going under cutting processes or not, with spatialization capability which might be integrated within geographic information system (GIS) environment. However, only above-ground growth is modeled. Soil water content, temperature, and radiation stress are considered the only environmental growth limitations. The model presents two development stages for the plant, growth cycle and dormancy. The dormant period can be trigged from two different approaches: day length or the soil moisture index - SMI, which allows covering temperate and tropical areas. A case study using Bahiagrass (Paspalum notatum Flügge) as input to run the model is presented as well as the evaluation procedures of the model performance. Two different platforms (an unmanned aerial vehicle – UAV and PlanetScope imagery) were used in the data acquisition and, the vegetation indices NDVI, GNDI, and EVI2 were used to retrieve LAI from in situ measurements and from the sensors. EVI2 showed the best performance for Bahiagrass LAI retrieval and thus it was used in the evaluation of the LAI simulated in the model. To assess the performance of the model, LAI from the model (default and adjusted) and LAI retrieved from both sensors are compared using the associated determination coefficient (R2) and root mean square error (RMSE) as criteria. The results obtained in the analysis suggest that the proposed model is suitable for its purpose and its eventual application may help government administrations with the optimization of cut/mowing processes of urban green spaces (UGS). However, some adjustments in the LAI curve development and in the dormant period are suggested.