Using Remote Sensing in Environmental Impact Assessment of Agricultural Areas: A Case of Kikonge Dam and Irrigation Project in Tanzania
Translated title
Using Remote Sensing in Environmental Impact Assessment of Agricultural Areas
Author
Thimm, Rasmus
Term
4. Term
Publication year
2019
Submitted on
2019-06-03
Pages
55
Abstract
Dette speciale undersøger, hvordan satellitbaseret fjernmåling kan støtte miljøkonsekvensvurderinger (MKV) med et landbrugsområde i det sydvestlige Tanzania som case. Analyserne er udført i Google Earth Engine, en skybaseret platform til behandling af satellitbilleder. Fire opgaver blev gennemført: (1) kortlægning af arealanvendelse og arealdække; (2) udarbejdelse af en tidsserie for NDVI (normaliseret differens-vegetationsindeks), et satellitbaseret mål for planters grønhed; (3) estimering af omfanget af afgrøder i den tørre sæson; og (4) sammenligning af NDVI på tværs af forskellige dele af landbrugsarealerne. Resultaterne viser, at omkring halvdelen af studieområdet er opdyrket. Kortlægningen var meget præcis for de fleste klasser, men mindre præcis for bebyggede områder. NDVI-tidsserien var påvirket af skydække, hvilket peger på behovet for bedre skysortering for tydeligere at se årlige vækstmønstre (fenologi). Afgrøder i den tørre sæson blev estimeret til ca. 20% af arealet—omtrent en tredjedel af, hvad andre kilder angiver—hvilket kræver nærmere undersøgelse. Disse arealer overlapper med steder med højt årligt NDVI, hvilket kan tyde på bedre vækstbetingelser og/eller dyrkningspraksisser. Der blev ikke gennemført feltundersøgelser af logistiske og økonomiske grunde, og det anbefales derfor at validere resultaterne i felten. Overordnet kan fjernmåling støtte MKV’er, men kan ikke løse alle udfordringer, især dem der skyldes fattigdom, svage institutioner eller korruption. Den bør bruges sammen med andre tilgange for at styrke MKV’er og miljøbeskyttelse.
This thesis examines how satellite-based remote sensing can support Environmental Impact Assessments (EIAs), using an agricultural area in southwestern Tanzania as a case study. Analyses were run in Google Earth Engine, a cloud platform for processing satellite images. Four tasks were carried out: (1) mapping land use and land cover; (2) creating a time series of the normalized difference vegetation index (NDVI), a satellite measure of plant greenness; (3) estimating how much of the area is planted with dry-season crops; and (4) comparing NDVI across different parts of the farmland. Results show that about half of the study area is farmland. The land cover map was very accurate for most classes, but performed less well for built-up areas. The NDVI time series was affected by clouds, suggesting that better cloud filtering is needed to see annual plant growth patterns (phenology) more clearly. Fields with dry-season crops were estimated to cover about 20% of the area—roughly one third of what other sources report—highlighting a gap that needs further investigation. These areas overlap with places that have high annual NDVI, which may point to better growing conditions and/or farming practices. No field surveys were possible because of logistical and financial constraints, so the study recommends validating these findings on the ground. More broadly, remote sensing can aid EIAs but cannot solve all challenges, especially those rooted in poverty, weak institutions, or corruption. It should be used alongside other approaches to strengthen EIAs and environmental protection.
[This abstract was generated with the help of AI]
Keywords
Documents