@techreport{RN166,
   author = {Fröberg, Mats and Löfgren, Stefan},
   title = {Förbättrad skattning av typhalterav N och P från skogs- ochmyrmark i södra Sverige införPLC6 – kan modeller baserade påkNN-data användas?},
   number = {141},
   abstract = {In PLC5, the type concentrations for nitrogen (N) and phosphorus (P) indischarge from forest land in northern Sweden were estimated from theirrelations to altitude, while the inorganic fractions were handled as constants.For southern Sweden, the type concentrations were based on median valuesobtained from measurements in a number of streams with forests andwetlands as dominating land cover in the catchment. The only forestoperation taken into account was clear-felling. In southern Sweden, the Ntype concentration from clear-felling was based on the relation between Ndeposition and the N concentration in soil water. In northern Sweden the Ntype concentration was obtained by multiplying the forest typeconcentration with a factor of 2. For P, the type concentration from clearfellingwas obtained in a similar way by multiplying the forest typeconcentration with a factor.In earlier SMED projects (SMED report no 52:2011, SMED report no100:2011, SMED report no 109:2012), nitrogen and phosphorusconcentrations in runoff from forest land and wetlands have been studied bymeasurements in stream water in about 200 randomly selected small forestand wetland dominated catchments. In these earlier projects forest status(increment, biomass, tree species etc. was classified by the ProbabilisticClassifier method and data from satellite images and the national forestinventory (NFI). Thereafter, models have been created to estimate the N andP type concentrations. The explanatory power of these models was muchhigher compared with those used for PLC5. Additionally, the data from therandomly selected streams showed that the PLC5 type concentrationstangibly underestimated the N and P concentrations in southwest Sweden.For PLC6, there is a need of improved N and P type concentrations fromforest land and wetland, especially in southern Sweden.Forest data based on the Probilistic Classifier method do not exist fromlarge areas of Sweden, which implies that improved PLC6 typeconcentrations cannot be developed for these regions. In this project wehave therefore used nationally covering geographic information from kNNSverigeand Lantmäteriet to describe forest status. kNN-Sverige is anothermethod for characterizing forest status and it is also based on satellite andNFI data.These publically available forest status data were used to model nitrogenand phosphorus concentrations in runoff from forest land and wetland in8southern Sweden south of Lake Siljan during four seasons. The models werevalidated against independent data from 22 streams in southern Sweden.This data originate from national and regional monitoring and the SwedishForest Agency forest soil liming project (SKOKAL).It was found that forest status based on kNN-data did not add any significantinformation to the models. The most important explanatory variables wereinstead geographic coordinates (longitude, latitude, altitude) and theproportion of forests and wetlands in the catchment.Based on models without kNN-data as explanatory variables, the modelingwas relatively successful for total N with R2 between 0.22 and 0.46 for therandomly selected streams and R2 between 0.11 and 0.51 for the 17 streamswith data on total-N in the test dataset. For total P modeling was lesssuccessful with R2 between 0.04 and 0.27 for the calibration dataset and R2close to zero for the test dataset. For the strams in the test dataset, the newmodels overestimated the N and P concentrations in many cases. This maybe explained by longer water residence time in these systems, which createsprerequisites for larger N and retention.Based on models without kNN-data as explanatory variables, calculations ofnew N and P type concentrations were made for the PLC5 sub-catchments.On average, the new estimates of concentrations are higher than those usedin PLC5, especially in low elevation areas in southernmost Sweden. Thenew N concentrations are better than those used in PLC5 and can be usedfor southern Sweden (south of lake Siljan) in PLC6. In southeast, themodeled N concentrations are less certain due to lack of water chemic l datafrom randomly selected streams. Such studies should be initiated in order tovalidate the model estimates. The modeled P concentrations should not beused for PLC6 due to low accuracy in the estimates.},
   keywords = {Natural Sciences
Earth And Related Environmental Sciences
Environmental Sciences
Naturvetenskap
Geovetenskap Och Miljövetenskap
Miljövetenskap
Smed (Svenska Miljöemissionsdata)
Vatten},
   url = {http://urn.kb.se/resolve?urn=urn:nbn:se:naturvardsverket:diva-1666},
   year = {2014},
   type = {Report}
}

{"_id":"sba3J2wZwayDRsxhN","bibbaseid":"frberg-lfgren-frbttradskattningavtyphalteravnochpfrnskogsochmyrmarkisdrasverigeinfrplc6kanmodellerbaseradepknndataanvndas-2014","author_short":["Fröberg, M.","Löfgren, S."],"bibdata":{"bibtype":"techreport","type":"Report","author":[{"propositions":[],"lastnames":["Fröberg"],"firstnames":["Mats"],"suffixes":[]},{"propositions":[],"lastnames":["Löfgren"],"firstnames":["Stefan"],"suffixes":[]}],"title":"Förbättrad skattning av typhalterav N och P från skogs- ochmyrmark i södra Sverige införPLC6 – kan modeller baserade påkNN-data användas?","number":"141","abstract":"In PLC5, the type concentrations for nitrogen (N) and phosphorus (P) indischarge from forest land in northern Sweden were estimated from theirrelations to altitude, while the inorganic fractions were handled as constants.For southern Sweden, the type concentrations were based on median valuesobtained from measurements in a number of streams with forests andwetlands as dominating land cover in the catchment. The only forestoperation taken into account was clear-felling. In southern Sweden, the Ntype concentration from clear-felling was based on the relation between Ndeposition and the N concentration in soil water. In northern Sweden the Ntype concentration was obtained by multiplying the forest typeconcentration with a factor of 2. For P, the type concentration from clearfellingwas obtained in a similar way by multiplying the forest typeconcentration with a factor.In earlier SMED projects (SMED report no 52:2011, SMED report no100:2011, SMED report no 109:2012), nitrogen and phosphorusconcentrations in runoff from forest land and wetlands have been studied bymeasurements in stream water in about 200 randomly selected small forestand wetland dominated catchments. In these earlier projects forest status(increment, biomass, tree species etc. was classified by the ProbabilisticClassifier method and data from satellite images and the national forestinventory (NFI). Thereafter, models have been created to estimate the N andP type concentrations. The explanatory power of these models was muchhigher compared with those used for PLC5. Additionally, the data from therandomly selected streams showed that the PLC5 type concentrationstangibly underestimated the N and P concentrations in southwest Sweden.For PLC6, there is a need of improved N and P type concentrations fromforest land and wetland, especially in southern Sweden.Forest data based on the Probilistic Classifier method do not exist fromlarge areas of Sweden, which implies that improved PLC6 typeconcentrations cannot be developed for these regions. In this project wehave therefore used nationally covering geographic information from kNNSverigeand Lantmäteriet to describe forest status. kNN-Sverige is anothermethod for characterizing forest status and it is also based on satellite andNFI data.These publically available forest status data were used to model nitrogenand phosphorus concentrations in runoff from forest land and wetland in8southern Sweden south of Lake Siljan during four seasons. The models werevalidated against independent data from 22 streams in southern Sweden.This data originate from national and regional monitoring and the SwedishForest Agency forest soil liming project (SKOKAL).It was found that forest status based on kNN-data did not add any significantinformation to the models. The most important explanatory variables wereinstead geographic coordinates (longitude, latitude, altitude) and theproportion of forests and wetlands in the catchment.Based on models without kNN-data as explanatory variables, the modelingwas relatively successful for total N with R2 between 0.22 and 0.46 for therandomly selected streams and R2 between 0.11 and 0.51 for the 17 streamswith data on total-N in the test dataset. For total P modeling was lesssuccessful with R2 between 0.04 and 0.27 for the calibration dataset and R2close to zero for the test dataset. For the strams in the test dataset, the newmodels overestimated the N and P concentrations in many cases. This maybe explained by longer water residence time in these systems, which createsprerequisites for larger N and retention.Based on models without kNN-data as explanatory variables, calculations ofnew N and P type concentrations were made for the PLC5 sub-catchments.On average, the new estimates of concentrations are higher than those usedin PLC5, especially in low elevation areas in southernmost Sweden. Thenew N concentrations are better than those used in PLC5 and can be usedfor southern Sweden (south of lake Siljan) in PLC6. In southeast, themodeled N concentrations are less certain due to lack of water chemic l datafrom randomly selected streams. Such studies should be initiated in order tovalidate the model estimates. The modeled P concentrations should not beused for PLC6 due to low accuracy in the estimates.","keywords":"Natural Sciences Earth And Related Environmental Sciences Environmental Sciences Naturvetenskap Geovetenskap Och Miljövetenskap Miljövetenskap Smed (Svenska Miljöemissionsdata) Vatten","url":"http://urn.kb.se/resolve?urn=urn:nbn:se:naturvardsverket:diva-1666","year":"2014","bibtex":"@techreport{RN166,\r\n author = {Fröberg, Mats and Löfgren, Stefan},\r\n title = {Förbättrad skattning av typhalterav N och P från skogs- ochmyrmark i södra Sverige införPLC6 – kan modeller baserade påkNN-data användas?},\r\n number = {141},\r\n abstract = {In PLC5, the type concentrations for nitrogen (N) and phosphorus (P) indischarge from forest land in northern Sweden were estimated from theirrelations to altitude, while the inorganic fractions were handled as constants.For southern Sweden, the type concentrations were based on median valuesobtained from measurements in a number of streams with forests andwetlands as dominating land cover in the catchment. The only forestoperation taken into account was clear-felling. In southern Sweden, the Ntype concentration from clear-felling was based on the relation between Ndeposition and the N concentration in soil water. In northern Sweden the Ntype concentration was obtained by multiplying the forest typeconcentration with a factor of 2. For P, the type concentration from clearfellingwas obtained in a similar way by multiplying the forest typeconcentration with a factor.In earlier SMED projects (SMED report no 52:2011, SMED report no100:2011, SMED report no 109:2012), nitrogen and phosphorusconcentrations in runoff from forest land and wetlands have been studied bymeasurements in stream water in about 200 randomly selected small forestand wetland dominated catchments. In these earlier projects forest status(increment, biomass, tree species etc. was classified by the ProbabilisticClassifier method and data from satellite images and the national forestinventory (NFI). Thereafter, models have been created to estimate the N andP type concentrations. The explanatory power of these models was muchhigher compared with those used for PLC5. Additionally, the data from therandomly selected streams showed that the PLC5 type concentrationstangibly underestimated the N and P concentrations in southwest Sweden.For PLC6, there is a need of improved N and P type concentrations fromforest land and wetland, especially in southern Sweden.Forest data based on the Probilistic Classifier method do not exist fromlarge areas of Sweden, which implies that improved PLC6 typeconcentrations cannot be developed for these regions. In this project wehave therefore used nationally covering geographic information from kNNSverigeand Lantmäteriet to describe forest status. kNN-Sverige is anothermethod for characterizing forest status and it is also based on satellite andNFI data.These publically available forest status data were used to model nitrogenand phosphorus concentrations in runoff from forest land and wetland in8southern Sweden south of Lake Siljan during four seasons. The models werevalidated against independent data from 22 streams in southern Sweden.This data originate from national and regional monitoring and the SwedishForest Agency forest soil liming project (SKOKAL).It was found that forest status based on kNN-data did not add any significantinformation to the models. The most important explanatory variables wereinstead geographic coordinates (longitude, latitude, altitude) and theproportion of forests and wetlands in the catchment.Based on models without kNN-data as explanatory variables, the modelingwas relatively successful for total N with R2 between 0.22 and 0.46 for therandomly selected streams and R2 between 0.11 and 0.51 for the 17 streamswith data on total-N in the test dataset. For total P modeling was lesssuccessful with R2 between 0.04 and 0.27 for the calibration dataset and R2close to zero for the test dataset. For the strams in the test dataset, the newmodels overestimated the N and P concentrations in many cases. This maybe explained by longer water residence time in these systems, which createsprerequisites for larger N and retention.Based on models without kNN-data as explanatory variables, calculations ofnew N and P type concentrations were made for the PLC5 sub-catchments.On average, the new estimates of concentrations are higher than those usedin PLC5, especially in low elevation areas in southernmost Sweden. Thenew N concentrations are better than those used in PLC5 and can be usedfor southern Sweden (south of lake Siljan) in PLC6. In southeast, themodeled N concentrations are less certain due to lack of water chemic l datafrom randomly selected streams. Such studies should be initiated in order tovalidate the model estimates. The modeled P concentrations should not beused for PLC6 due to low accuracy in the estimates.},\r\n keywords = {Natural Sciences\r\nEarth And Related Environmental Sciences\r\nEnvironmental Sciences\r\nNaturvetenskap\r\nGeovetenskap Och Miljövetenskap\r\nMiljövetenskap\r\nSmed (Svenska Miljöemissionsdata)\r\nVatten},\r\n url = {http://urn.kb.se/resolve?urn=urn:nbn:se:naturvardsverket:diva-1666},\r\n year = {2014},\r\n type = {Report}\r\n}\r\n\r\n","author_short":["Fröberg, M.","Löfgren, S."],"key":"RN166","id":"RN166","bibbaseid":"frberg-lfgren-frbttradskattningavtyphalteravnochpfrnskogsochmyrmarkisdrasverigeinfrplc6kanmodellerbaseradepknndataanvndas-2014","role":"author","urls":{"Paper":"http://urn.kb.se/resolve?urn=urn:nbn:se:naturvardsverket:diva-1666"},"keyword":["Natural Sciences Earth And Related Environmental Sciences Environmental Sciences Naturvetenskap Geovetenskap Och Miljövetenskap Miljövetenskap Smed (Svenska Miljöemissionsdata) Vatten"],"metadata":{"authorlinks":{}}},"bibtype":"techreport","biburl":"https://www.slu.se/globalassets/ew/org/centrb/rt/dokument/publikationslistor/rikskogstaxeringsprojektet_bibtex_ny.txt","dataSources":["Cac5nog9ND5ndLYhY","xTmt4jq9swAwpBHJB","LXbacBrgTRDPkh9C2","dLPsL5XH9N5Pjush7"],"keywords":["natural sciences earth and related environmental sciences environmental sciences naturvetenskap geovetenskap och miljövetenskap miljövetenskap smed (svenska miljöemissionsdata) vatten"],"search_terms":["ttrad","skattning","typhalterav","och","skogs","ochmyrmark","dra","sverige","inf","rplc6","kan","modeller","baserade","knn","data","anv","ndas","fröberg","löfgren"],"title":"Förbättrad skattning av typhalterav N och P från skogs- ochmyrmark i södra Sverige införPLC6 – kan modeller baserade påkNN-data användas?","year":2014}