Seasonal precipitation interpolation at the Valencia region with multivariate methods using geographic and topographic information

Abstract

ABSTRACT:The spatial pattern of precipitation is a complex variable that strongly depends on other geographic and topographic factors. As precipitation is usually known only at certain locations, interpolation procedures are needed in order to predict this variable in other regions. The use of multivariate interpolation methods is usually preferred, as secondary variables -generally derived using GIS tools -correlated with precipitation can be included. In this paper, a comparative study on different univariate and multivariate interpolation methodologies is presented. Our study area is centred in the region of Valencia, located to the eastern Spanish Mediterranean coast. The followed methodology can be divided in three steps. First, secondary variables having significant correlations with the precipitation were derived, where the hillsides were used as influence areas of certain variables. Secondly, precipitation was interpolated with different methodologies. Finally, the derived models were compared in terms of predicted errors. Models were achieved for seasonal scales, considering a total of 179 raingauges; data of another 45 raingauges were also used to predict errors. Results prove that there is no ideal method for all the cases but it will depend on one hand, on the number of geographical factors that influence the rainfall and, on the other hand, on the major or minor spatial correlation within the rainfall.

Batch orographic interpolation of monthly precipitation based on free-of-charge geostatistical tools

“…It is known that the precipitation totals measured at points spread over a mountainous area correlate with elevation, but the correlation is higher if the elevation representative of a wider vicinity of a rain gauge is studied instead of the elevation corresponding to the place where the rain gauge is situated (see e.g. [20]). Therefore, also the 'raster' package (and its function 'focal()'; see [21]) had to be applied in order to compute mean elevation values representative of places up to 6 and 12 km away from the rain gauges.…”

“…intercept and different slopes specified by the subscripts), ε is the residual component and the superscripts indicate powers whose inclusion has been shown beneficial for several mountainous areas (see e.g. [20]). The estimated parameters can in turn be used for the derivation of precipitation estimates ̂ for every cell of the DEM where also the coordinates are known.…”

Change trend of monthly precipitation in China with an improved surface modeling method

“…Many studies show that the mixed spatial interpolation method, which combines regression method and interpolators, usually produces better results than other models in climate variables simulation (Burrough and McDonnell 1998;Ninyerola et al 2000;Liu et al 2004;He et al 2005;Portales et al 2010;Yue et al 2013). Regression with residual correction can modify any local overestimation or underestimation.…”

Climatological modeling of monthly air temperature and precipitation in Egypt through GIS techniques

“…Given that many studies employed multivariate regression for spatial interpolation of climatic variables, particularly temperature and precipitation (e.g. Holdaway 1996, Agnew & Palutikof 2000, Ninyerola et al 2000, 2007a,b, Brown & Comrie 2002, Portales et al 2009), the regression-based techniques seem to be superior to the classical interpolation models (Brown & Comrie 2002). Multivariate regression models employ measurements from a network of stations to fit the best equation as a function of a range of topographical and environmental predictors.…”