Varasemalt põllumajanduslikus kasutuses olnud maa metsastumise kaardistamine satelliidi Landsat talvel pildistatud piltidelt
Laen...
Kuupäev
2024
Kättesaadavus
05.09.2024
Autorid
Ajakirja pealkiri
Ajakirja ISSN
Köite pealkiri
Kirjastaja
Eesti Maaülikool
Abstrakt
1990. aastate alguses seoses Eesti riigikorra muutumisega jäi märkimisväärne hulk põllumaad kasutusest välja. Tänaseks on osa põllumaast taas kasutusse võetud ja teine osa, mis kasutust ei leidnud, metsastunud. Antud töö eesmärk on kasutada pilditöötluse ja ruumianalüüsi võtted, mida on kasulik kasutada metsastunud põllumaa kaardistamisel. Metsastunud põllumaa leidmiseks kasutatakse 12. satelliidipilti, mis on pildistatud 34-aastasel perioodil veebruaris ja märtsis lausalise lumikatte oludes. Pildiandmetele lisaks kasutati Eesti Baaskaarti, Maa-ameti ortofotosid, Eesti topograafia andmekogu ja Põllumajanduse Registrite ja Informatsiooni Ameti (PRIA) põllumassiivide registri andmestikku. Talvistel satelliidipiltidel lausalise lumikatte olemasolul tekivad tingimused, mil saab metsa ümbritsevatest lagedatest aladest eristada tänu heledusväärtuste erinevusele. Klassifitseerimisviisina jagatakse piltide pikslite bimodaalne jaotus kaheks klassiks: „mets“ ja „mitte-mets“ eristusnivoo kui eristuspiiri seadmisega. Bakalaureusetöö katsealal Valga maakonnas, suurusega 900 ruutkilomeetrit on Baaskaardil esindatud põllumaale (26400 hektarit) tekkinud 34 aastaga 4484,6 hektarit metsa (17,0% Baaskaardi põllumaast). Klassifitseeritud metsast on põllumaa ühe piksli laiuses servavööndis, mis külgneb naabruses oleva metsaga, 2535,0 hektarit, ehk 56,5% klassifitseeritud metsakujundite pindalast. Ainult põllumaa ühe piksli laiuses servavööndis esinev ega sealt välja ei ulatu 951,2 hektarit (37,5% servavööndis olevate lappide pindalast) metsa. Just põllumaa piiriala külgneva metsaga on alaks, kus tegelike metsakujunditega koos ilmnevad vale-positiivsed metsana klassifitseeritud kujundid.
Following changes in Estonian state system in 1991 a significant amount of agricultural land was abandoned. Nowadays part of the abandoned agricultural land is back in use, other part of the abandoned land has grown into a forest. The aim of this thesis is to test image processing and spatial analysis techniques that are appropriate for mapping afforested former farmland. 12 winter satellite Landsat Thematic Mapper and Landsat Operational Land Imager images were used to map forested former farmland. The images covered a 34-year period starting with 1987. Supplementary map data were used as Estonian Basemap, Estonian Land Board orthophotos, Estonian Topographic Database and Republic of Estonia Agricultural Registers and Information Board register of farmlands. Winter satellite images were classifited into two classes: „forest“ and „non-forest“, with an arithmetic mean. 4484,6 hectares of forest has grown on the farmland of the experimental area in South Estonia with a size of 900 sq km in 34 years. That is 17,0 per cent of arable land (26400 hectares) in Basemap data set in early 1990s. Special emphasis is paid on narrow 1-pixel zones of arable land bordering with forests (2535,0 hectares or 56,5% of classified forest in 1-pixel zone of arable land). 951,2 of 2535,0 hectares (37,5%) exist only within 1-pixel zone. These areas are specially prone of false positive forest patches containing essentially also real forest patches.
Following changes in Estonian state system in 1991 a significant amount of agricultural land was abandoned. Nowadays part of the abandoned agricultural land is back in use, other part of the abandoned land has grown into a forest. The aim of this thesis is to test image processing and spatial analysis techniques that are appropriate for mapping afforested former farmland. 12 winter satellite Landsat Thematic Mapper and Landsat Operational Land Imager images were used to map forested former farmland. The images covered a 34-year period starting with 1987. Supplementary map data were used as Estonian Basemap, Estonian Land Board orthophotos, Estonian Topographic Database and Republic of Estonia Agricultural Registers and Information Board register of farmlands. Winter satellite images were classifited into two classes: „forest“ and „non-forest“, with an arithmetic mean. 4484,6 hectares of forest has grown on the farmland of the experimental area in South Estonia with a size of 900 sq km in 34 years. That is 17,0 per cent of arable land (26400 hectares) in Basemap data set in early 1990s. Special emphasis is paid on narrow 1-pixel zones of arable land bordering with forests (2535,0 hectares or 56,5% of classified forest in 1-pixel zone of arable land). 951,2 of 2535,0 hectares (37,5%) exist only within 1-pixel zone. These areas are specially prone of false positive forest patches containing essentially also real forest patches.
Kirjeldus
Bakalaureusetöö
Metsanduse õppekaval
Märksõnad
bakalaureusetööd, kaugseire, klassifitseerimine, metsastumine, põllumaa, talvised Landsati pildid