Spatial Modeling of Climatic Parameter Fluctuation Mapping Temperature Variation in the Bermejo Basin from 1901 to 2000 Vector maps showing thermal zones and isotherms for different periods, and isopleths of air temperature temporal variation for each couple of periods are derived from the classified raster maps obtained in the previous step. This is done in the GIS by applying a raster to vector conversion, followed by line smoothing and short segment removal. Vector maps are final products for results presentation and are also used for visual analysis of climatic parameter fluctuation within anthropic and (semi)natural ecosystems, by plotting vector lines onto land-use and vegetation maps. Finally, the two maps showing thermal zones were crossed and a map showing the shift of the thermal zones was developed, while the original raster maps showing temperature temporal variations were used to assess relationships with land cover types statistically. Results The first product obtained was a tabular database built up from temperature and rainfall data of 35 meteorological stations summarized for different periods from 1873 to 2000 (main stations location is shown in Figure 4). For this, the following sources were consulted: Gould (1882), Davis (1889 and 1895), SMN (1958a, 1958b, 1963, 1981 and 1992), CIRHMET (1993), Arias and Bianchi (1996), Olivares et al (1995), and preliminary decennial data from 1991-2000 were obtained from the Biblioteca Nacional de Meteorologia del Servicio Meteorologico Nacional (Buenos Aires, Argentina). A Digital Elevation Model (DEM) produced from topographic data in a previous study (Zuviria 2002) was already available. Even when tabular data are not reported here, the comparative data analysis of annual mean, maximum mean and minimum mean temperatures and annual rainfall carried out for stations presenting values recorded during 1901-1950 and 1951-2000, allowed to extract the following conclusions for this period: 1) annual mean temperature decreased (0 to 0.8°C), as reported by IPCC (1998) and in opposition to assumptions erroneously taken within the Bermejo Basin (Seoane and Moyano 1999); 2) annual maximum mean temperature decreased more than the annual mean (0.4 to 2.2°C); 3) annual minimum temperature increased in most of the area, showing a slight decrease in a small sector (-0.2 to 1.5°C); 4) thermal amplitude decreased (0.5 to 3.4°C); and 5) annual rainfall increased (5 to 26%), showing a high correlation (R2 = 0.85) with annual maximum mean temperature decreasing. The formulas obtained from the regression analysis between annual air temperature (Tx; °C) and elevation (E; meters * 100) for the periods 1901-1950 and 1951-2000 are the following: Tx 1901-1950 = 23.6 – 0.44 E Tx 1951-2000 = 22.5 – 0.38 E From the straight application of these formulas, it results that annual air temperature decreased within the area below 1,800 m.a.s.l. and increased above this limit. Lapse rates obtained from linear regression between annual mean air temperature and elevation were highly correlated with elevation (R2 > 0.94), as expected from a previous study (Zuviria and Burgos 1986). This correlation is higher within the hilly and mountainous sectors of the Upper Basin while in the flat Lower Basin temperature correlation with elevation is poor. This is because lapse rates are also associated within this area, but in a lower degree, with rainfall, cloudiness and latitude, as indicated by Bianchi et al (1994). These influences are accounted in the maps showing air temperature differences at sea level. Lapse rates were validated using data from secondary stations but, even when it was planned to take soil temperature measurements to complete this validation, as done in a previous study (Zuviria 1992), it could not be done here. 
ILWIS (Nijmeijer et al, 2001) GIS was used for the mapping process, and maps showing thermal zones (FAO 1997) and their spatial shift from 1901-1950 to 1951-2000 were produced (Figure 2), taking ranges of air temperature most probably correlated with land-use suitability, as recommended by Parry (1990). Maps presenting annual mean isotherms for the mentioned periods and isopleths of air temperature temporal variation were also produced and used to visualize air temperature fluctuation in a fragile ecosystem, the transition forest (Figure 3) and on main land-use types (Figure 4). Tabular results from the geostatistical analysis are also presented in the last figure, showing mean temperature variation (X var) and standard deviation (SD) within each land-use type from 1901-1950 to 1951-2000. 
The geo-database developed during this mapping process, consisting of tables, formulae, maps and their relationships, is built up using dependency links among objects. Consequently, when input source maps, tables or formulae are improved, the dependent output maps and/or tables can be automatically recalculated. This geo-database is currently used for the development of scenarios to represent future trends of air temperature under different local and global conditions, and to locate and analyze the dynamic variation of other climatic parameters, contributing to the study of climate change and land use planning at different scales. |