Detection of vegetation trends in the Jequitinhonha-MG river basin.
Name: ROSANE GOMES DA SILVA
Publication date: 17/02/2020
Advisor:
Name |
Role |
|---|---|
| ALEXANDRE ROSA DOS SANTOS | Advisor * |
| HENRIQUE MACHADO DIAS | Co-advisor * |
| JOSÉ EDUARDO MACEDO PEZZOPANE | Co-advisor * |
Examining board:
| Name |
Role |
|---|---|
| ALEXANDRE ROSA DOS SANTOS | Advisor * |
| DAIANI BERNARDO PIROVANI | External Examiner * |
| HENRIQUE MACHADO DIAS | Co advisor * |
| JÉFERSON LUIZ FERRARI | External Examiner * |
Summary: Monitoring and evaluating changes in the condition of vegetation is important for biodiversity and its relationship to human activities, assisting in planning, prioritizing, managing and monitoring biodiversity conservation. Remote sensing data can assist in this type of study, as they provide long continuous time series, product data availability based on different remote sensors and photosynthetic capacity indicators. The objective of this study was to study changes in vegetation, associated with climate variability and changes in land use, in the Jequitinhonha River basin, between 2001 and 2018. NDVI, precipitation and temperature images were used from 2001 to 2018 and land use images from 2001 to 2018. The following methods were used to analyze the interannual trends in vegetation vigor and the behavior of precipitation and temperature, considering a statistical significance of 5%: Mann Kendall monotonic trend, linear trend and linear correlation. Subsequently, the linear correlation between NDVI, temperature and precipitation was performed, using multiple linear regression. Finally, changes in land use in the basin were identified, and the gains and losses of area in each use were quantified, with emphasis on the most accentuated changes. The results showed that 79% of the area showed a tendency to decrease in the greenness, while 21% of the area showed a tendency to increase in the greenness. The main land uses that represented areas with a decreasing trend were pasture and forest formation. Areas with an upward trend were represented mainly by planted forest. The correlation of NDVI with temperature indicated the best response for a lag of three months, and for most of the basin, the correlation was between 0.4 and 0.6 (56%), followed by areas with a correlation between 0 and 0.4 (34%). For the correlation between NDVI and precipitation, this lag was one month and, most of the basin (46%), obtained a correlation between 0.4 and 0.6. About 30% of the area obtained a correlation between 0 and 0.4 and 23% between 0.6 and 0.8. Only 1% of the area showed a correlation less than 0. The analysis of the dynamics in land use indicated that there was suppression of vegetation in natural areas, which were replaced, mainly by areas of planted forests, pasture and agricultural crops. Areas classified as pasture were mostly replaced by Mosaic of agriculture and pasture, a class that includes pasture areas. Thus, it is possible to consider that most of this area has remained unchanged. The class Planted forest was the only one in which the gains in area were actually greater than the losses, and occupied mainly areas of savanna formation, forest formation and grassland. Other areas in which an increase in the greenness of the vegetation was observed may be associated with changes in the successional stages of native vegetation in the study area.
Keywords: Remote sensing, time series, greenness, Mann Kendall, Mapbiomas.
