Plants as bioindicator for temperature interpolation purposes: Analyzing spatial correlation between botany based index of thermophily and integrated temperature characteristics

Abstract

Abstract The success of interpolation techniques relies heavily on the density and regularity of field reference data points. For instance temperature interpolations in the Arctic are hampered by few and scattered meteorological stations. The major objective of this study is to analyze the spatial relationship between plants, defined in terms of an index of thermophily ( I t ) and temperature distribution. The study area is located in Kongsfjorden, northwest Spitsbergen (Svalbard). A systematic recording of floristic data covering the study area was made within quadrates of 1 km × 1 km (93 units). For each of them, the I t was calculated. I t provides a synthetic measure by which plants are taken as temperature indicators at a long time scale. Temperature values were recorded by means of 39 temperature loggers during the summer 2000. The model for spatial interpolation of temperature was developed using multiple regression of remote sensed data (Landsat TM) and topographical features derived from a digital elevation model (DEM). Continuous temperature layers were calculated at a spatial resolution of 50 m × 50 m, and aggregated to a resolution of 1 km × 1 km in order to correspond with the observed botanical units. Different maps were produced showing spatial distribution of the modelled temperature and I t . Correlations between the I t and temperature values derived from the modelled temperature layers were systematically explored. Correlation between the I t and temperatures works well as standard deviation of residues is 0.7 °C only. Highest correlations ( r ) of I t and the spatial distribution of temperature were obtained for: (a) maximum average temperature for August, excluding all areas higher than 100 m above sea level (0.75), (b) average daily maximum temperature for July–October (0.67), (c) average temperature for July and August (0.64, 0.65), and (d) when temperature range is >8 °C (0.55). Areas with low correlations between I t and temperature were mainly attributed to the fact that these measurements represent (a) different time scales and (b) different spatial scales. However, results from this study have shown that calculating I t provide a mean for restoring selected temperature parameters and thus can contribute to fill in and extend the network of field data points for temperature interpolation purposes.