Soil water content interpolation using spatio-temporal kriging with external drift

Size: px
Start display at page:

Download "Soil water content interpolation using spatio-temporal kriging with external drift"

Transcription

1 Geoderma 112 (2003) Soil water content interpolation using spatio-temporal kriging with external drift J.J.J.C. Snepvangers*, G.B.M. Heuvelink, J.A. Huisman Institute for Biodiversity and Ecosystem Dynamics (IBED), Centre for Geo-Ecological Research (ICG), Universiteit van Amsterdam, Nieuwe Achtergracht 166, 1018 WV Amsterdam, The Netherlands Received 19 November 2001; accepted 14 October 2002 Abstract In this study, two techniques for spatio-temporal (ST) kriging of soil water content are compared. The first technique, spatio-temporal ordinary kriging, is the simplest of the two, and uses only information about soil water content. The second technique, spatio-temporal kriging with external drift, uses also the relationship between soil water content and net-precipitation to aid the interpolation. It is shown that the behaviour of the soil water content predictions is physically more realistic when using spatio-temporal kriging with external drift. Also, the prediction uncertainties are slightly smaller. The data used in this study consist of Time Domain Reflectometry (TDR) measurements from a 30-day irrigation experiment on a m grassland in the Netherlands. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Geostatistics; Space time interpolation; Soil hydrology 1. Introduction Sampling and monitoring often put a heavy load on the budget of environmental studies. Techniques that can increase the insight in the spatio-temporal (ST) distribution of an environmental variable, without increasing the measurement effort, are therefore valuable. Geostatistics offers a variety of techniques to make optimal use of measurement information for interpolating variables in space (S). However, many branches within the earth and environmental sciences deal with variables that vary not only in space but also in * Corresponding author. Present address: Netherlands Institute for Applied Geosciences, P.O. Box 80015, 3508 TA Utrecht, The Netherlands. addresses: j.snepvangers@nitg.tno.nl (J.J.J.C. Snepvangers), g.b.m.heuvelink@science.uva.nl (G.B.M. Heuvelink), s.huisman@science.uva.nl (J.A. Huisman) /02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S (02)

2 254 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) time (T). For instance, within soil science, there are numerous dynamic spatial attributes, such as soil water content, infiltration rate, water pressure and solute concentration. Kyriakidis and Journel (1999) presented a thorough review of the status of ST geostatistical techniques. Most of the techniques mentioned in their review only make use of measurements of the variable of interest itself, so-called primary information. In many studies, however, secondary information is available as well. For example, other variables are measured that show a (strong) relationship with the variable under study. Using such relationships in interpolation, by co-kriging or kriging with external drift, may decrease the prediction uncertainties (Goovaerts, 1997). Interpolation in the full ST domain offers new possibilities for these techniques, as dynamic relationships can also be taken into account. The aim of this study is to show how a dynamic physical relationship between soil water content (h) and net-precipitation can be used to improve h interpolations. h depends on the amount of water leaving and entering the soil and on soil hydraulic properties. Netprecipitation is an important characteristic in this system as it largely determines the soil water fluxes at the top of the system. As compared to the other fluxes in the system (the fluxes to the groundwater and the horizontal fluxes), the net-precipitation flux is often relatively large. Furthermore, net-precipitation can be measured relatively easy. We compare ST ordinary kriging (ST-OK), which ignores secondary information, with ST kriging with external drift (ST-KED), which employs net-precipitation as secondary information. We demonstrate how KED can be used in the ST domain and analyze the advantages and disadvantages of ST-KED compared to ST-OK. The data for this study were obtained from an irrigation experiment on a m grassland in the south of the Netherlands, which we will refer to as the Molenschot dataset. 2. Spatio-temporal geostatistics Extending S interpolation techniques to the ST domain is not simply adding another dimension, as there are some fundamental differences between the space and time domain (Christakos and Vyas, 1998; Kyriakidis and Journel, 1999; Rouhani and Myers, 1990). Space represents a state of coexistence, in which there can be multiple dimensions (or directions) and interpolation is usually of main interest. Time on the contrary represents a state of successive existence, a clear ordering (nonreversible) in only one dimension is present and extrapolation is usually of main interest. Moreover, the origin of the S and T variation can be different. For example, in the case of h, one can imagine that the T behaviour is dominated by net-precipitation and drainage, whereas S variation in h depends more on soil texture, soil physical properties and vegetation. The difference in origin of variation can lead to strong anisotropic behaviour, both geometric and zonal. In recent years, progress has been made in building ST geostatistical models in several scientific disciplines, for instance, in environmental science (e.g. Buxton and Pate, 1996; De Cesare et al., 1996, 2001a,b; Angulo et al., 1998; Christakos and Vyas, 1998; Kyriakidis and Journel, 2001), agronomy (e.g. Stein et al., 1994; Hoosbeek, 1998), meteorology (e.g. Handcock and Wallis, 1994; Bogaert and Christakos, 1997a; Cressie and Huang, 1999; Bechini et al., 2000), hydrology (e.g. Rouhani and Myers, 1990;

3 Bogaert and Christakos, 1997b) and soil science (e.g. Comegna and Vitale, 1993; Heuvelink et al., 1996). Basically, the aim of these studies is the same, namely to predict an attribute z={z(s, t)jses, tet} defined on a geographical domain SoR 2 and a time interval ToR, at a space time point (s 0, t 0 ), where z was not measured. The prediction is to be based on n measurements at n ST points (s i, t i ), i =1,..., n. To predict z(s 0, t 0 ), it is assumed that z is a realization of a ST random function (ST-RF) Z, and Z(s 0, t 0 ) is predicted conditional on the measurements z(s i, t i ). The ST-RF model Z consists of a trend component representing some average behaviour of the ST process (m) and a zero-mean residual component (e): Zðs i ; t i Þ¼mðs i ; t i Þþeðs i ; t i Þ i ¼ 1;...; n: ð1þ 2.1. The trend component J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) The simplest way to model the trend component m(s, t) is to assume an unknown constant mean. Interpolation can then be carried out using ST ordinary kriging (ST-OK). When the assumption of a constant mean is not realistic, a trend must be taken into account. A simple option is to detrend the data beforehand, after which ST-Simple Kriging can be used for interpolation (Angulo et al., 1998; De Cesare et al., 2001b). However, uncertainties in the detrending procedure are not taken into account in further analysis. This causes the interpolation uncertainty to appear lower than it is. It is also possible to model the trend component as a linear trend function, consisting of the sum of products of some known base-functions and some unknown coefficients: mðs; tþ ¼ Xp i¼1 f i ðs; tþb i : ð2þ In the simplest case, the base-functions are the coordinates (x,y,t) (universal kriging). When secondary information is available, then this may also be used to define the basefunctions (kriging with external drift). Bogaert and Christakos (1997a), for example, use altitude as secondary information in their ST study of thermometric data The residual component The residual in Eq. (1) can be characterised by the ST semivariogram, c(s i, s j, t i, t j ): cðs i ; s j ; t i ; t j Þ¼ 1 2 E ðeðs i ; t i Þ eðs j ; t j ÞÞ 2 : ð3þ Under appropriate stationarity assumptions, an estimate of the ST variogram may be obtained from the measurements by computing the experimental semivariogram ĉ(h S, h T ): ĉðh S ; h T Þ¼ 1 2Nðh S ; h T Þ Nðh X S ;h T Þ i¼1 ½eðs; tþ eðs þ h S ; t þ h T ÞŠ 2 ð4þ

4 256 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) where h S and h T are the S and T lags and where N(h S, h T ) is the number of pairs in the ST lag. Fitting a model to the ST experimental semivariogram has some additional problems over conventional semivariogram modelling, due to the distinct differences between the S and T variation. One way of coping with these problems is to use completely separate S and T semivariance structures and to model the total ST semivariance as the sum of these structures. Although this approach facilitates the structural analysis, it has some important drawbacks that are caused by the strict separation. For instance, the ST separation means that knowing the attribute value at three corners of a rectangle in the ST domain completely determines the attribute value at the fourth corner (Heuvelink et al., 1996). This implies that the S behaviour must be the same for all time points and that the T behaviour must be the same for all space points. However, this is not what we see in practice, where different spatial patterns emerge at different times and where time series at different locations show different behaviour. Complete ST separation is therefore unrealistic from a physical perspective. Furthermore, when measurements at all four corners are collected, a singularity problem will be encountered (Rouhani and Myers, 1990). Separate product structures, such as suggested by Rodriguez-Iturbe and Mejia (1974), Bogaert (1996) and De Cesare et al. (1996), may overcome the singularity problem. However, they still do not model space time interaction. Consequently, these structures are severely limited in their ability to fit the data well (Cressie and Huang, 1999). From a mathematical statistical standpoint, a variety of more advanced permissible nonseparate semivariance structures, which do not suffer from the above drawbacks, have been proposed (Cressie and Huang, 1999; De Cesare et al., 2001a; De Iaco et al., 2002). Although these structures are mathematically correct, they often lack physical support and are somewhat artificial. Therefore, environmental scientists often do not feel comfortable with them. Bilonick (1988) presented a simple nonseparate model form. He proposed an extension of the separate-sum models using geometric and zonal anisotropy to solve the problems arising from the differences in S and T variability. In the Bilonick model (Eq. (5)), the residual component is divided in three parts. These are an S part e S (s), a T part e T (t) and an ST part e ST (s, t) that only comprises geometric anisotropy and no zonal anisotropy: eðs; tþ ¼e S ðsþþe T ðtþþe ST ðs; tþ: ð5þ Assuming these three parts to be second-order stationary and mutually independent, the semivariogram of e(s, t) is the sum of three components: cðh S ; h T Þ¼c S ðh S Þþc T ðh T Þþc ST ðh ST Þ: ð6þ ThepST lag h ST is obtained by introducing a geometric anisotropy ratio a: h ST ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi h 2 S þ ah2 T. The advantage of the Bilonick model is that it has S, T and ST components that can be fairly easily interpreted in a physical sense (Heuvelink et al., 1996). The disadvantage is that estimation of the model parameters is not easy. Also, by introducing the space time anisotropy ratio a, it is assumed that distances in space and time can be reduced to a single space time distance. This may not be very realistic in all practical situations. Model

5 building is all about providing a sufficiently realistic description of the real world while still being able to identify the model parameters and apply the model. In balancing these two concerns, the level of complexity of the Bilonick model seems appropriate for the case study investigated here Spatio-temporal kriging J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) When models for the trend and the residual are obtained, ST kriging can be carried out; either ST-OK or ST-KED. The equations for kriging in the ST domain are exactly the same as the standard S kriging equations. One should be aware, though, of the consequences of kriging in the T domain. Future measurements influence present predictions just as much as past measurements, because they are both weighed using the same semivariogram. This may lead to physically unrealistic results, especially when sudden inputs in the system occur. One may choose to use only past measurements for interpolation, but this causes a loss of information. ST-KED can reduce the unrealistic effects without ignoring information because the sudden inputs can be incorporated in the base-functions of the linear trend. For the OK and KED equations and their derivation, we refer to geostatistical handbooks, for example, Goovaerts (1997). 3. Molenschot dataset In the summer of 2000, we carried out an irrigation experiment on a grassland (60 60 m) located in Molenschot, the Netherlands (51j35VNand4j52VE). The soil was classified as a Plaggept on sandy loam (US soil taxonomy; USDA, 1975). Sprinklers with different ranges and intensities created an S pattern of h on the grassland on two occasions in a 30-day monitoring period (August 16 September 14, 2000). Drying and re-wetting by natural precipitation caused the S pattern to change over time. We chose to do this type of irrigation experiment with a distinct irrigation pattern on a relatively small field, as we needed a strong S and T structure in h to make semivariogram modelling worthwhile and to obtain a good insight in the net-precipitation information used by ST-KED. We monitored h with vertically installed 10-cm Time Domain Reflectometry (TDR) probes. TDR measures the propagation velocity of an electromagnetic wave along parallel metallic rods inserted in the soil. The velocity depends on the permittivity (K a ) of the soil. Permittivity can be translated to h, as the permittivity of water ( F 80 at 20j) is much larger than that of the other soil constituents (air: 1, solids: F 4 8) (Topp et al., 1980).We used a site specific p calibration equation to convert the permittivity to h. The calibration curve, h ¼ 0:1116 ffiffiffiffiffi K a 0:1543, has an R 2 of 98.7% and a standard error of m 3 /m 3. We collected TDR measurements both manually and automatically. For the manual measurements, a probe was manually installed at each of 229 locations at every measurement time (Fig. 1). It is impossible to reinstall a probe at exactly the same location, and therefore, we chose to reinstall within 15 cm of the exact measurement

6 258 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) Fig. 1. Locations of the TDR and meteorological measurements. Shades of grey represent the irrigation pattern. location. The displacements were taken into account in the geostatistical analysis. In total, there were 19 manual measurement rounds (Fig. 2). The probes for automatic measurements remained installed throughout the whole experimental period at 34 locations (Fig. 1). They were connected to two computercontrolled measurement systems (Heimovaara and Bouten, 1990). The two systems, A and B, caused clustering of the automatic probes as the quality of the TDR measurements decreases with cable length (Heimovaara, 1993). Automatic measurements were carried out every 15 min. The measurements started at August 12 for system A and at August 18 for system B. There is a large difference between the number of automatic and manual measurements (Table 1). To balance the number of automatic and manual measurements and their distribution over the T and S domain, we drew at random 6% of the automatic measure-

7 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) Fig. 2. Course of the daily average NP natural during the experiment, displayed with the irrigation and manual TDR measurement times. ments for further analysis and omitted the other 94%. Table 1 shows that the data reduction has a negligible effect on the summary statistics. We also carried out meteorological measurements to gain information on the netprecipitation. We distinguished between natural net-precipitation, NP natural, and total netprecipitation, NP total. NP natural is the input to the topsoil from precipitation P(t) minus the output from the topsoil through actual evapotranspiration ET a (t): NP natural ðtþ ¼PðtÞ ET a ðtþ: ð7þ We assumed that NP natural was constant over space. We further assumed that the potential evapotranspiration (ET p ) calculated with the Penman equation was a satisfactory estimate of ET a, since during the monitoring period, no water shortage occurred in the field. For obtaining ET p, we measured air temperature, relative humidity, wind velocity, and net-radiation in the southwest corner of the field (Fig. 1). Table 1 Statistical summary of the TDR measurements N Minimum Maximum Mean Standard deviation All Manual Automatic System A Automatic System B All Automatic All % Automatic + All Manual

8 260 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) Table 2 Number, type and characteristics of sprinklers used for the S irrigation pattern Shape Size Number Estimated area a [m 2 ] Average intensity a [mm/h] Number of measurement cups Round Large Round Medium Round Small Square Medium With number of raingauges, we mean the number of measurement cups used to estimate the S pattern of a sprinkler type. The total net-precipitation NP total is NP natural plus irrigation I(s, t): NP total ðs; tþ ¼NP natural ðtþþiðs; tþ: ð8þ Due to the irrigation, NP total varied not only in time but also in space. We irrigated approximately one third of the field with 52 sprinklers of four different types (Table 2) at two dates: August 17 (day 230) and September 1 (day 245). At both days, we started the irrigation early in the morning, respectively, at 5.42 and 4.00 AM, to prevent evaporation during irrigation. Both irrigations lasted 4 hours. To obtain the S distribution of the irrigated water, we assumed that all sprinklers of one type had the same irrigation characteristics. This allowed us to measure the distribution of irrigated water around one sprinkler per sprinkler type and translate this to a total irrigation pattern. In Fig. 1, the irrigation patterns for the two irrigation dates are shown. In Fig. 2, the course of the daily natural net-precipitation is shown together with the irrigation dates and the manual TDR rounds. 4. Application of spatio-temporal interpolation to the Molenschot dataset 4.1. Spatio-temporal ordinary kriging The first step in the analysis was to carry out ST-OK. The only prerequisite for ST-OK is a model of the ST semivariance structure The ST semivariogram Fig. 3 (top graph) shows the experimental semivariogram for the ST-OK case. There are clear differences in the behaviour of the semivariance in the space and time directions. In the S direction, there is a strong increase in semivariance until 5 m and a less pronounced increase up to 10 m. In the T direction, a periodicity with a period of approximately 15 days stands out. This periodicity can be attributed to the fact that both the irrigation days and the heavy rainstorms occurred with intervals of approximately 15 days (Fig. 2). The largest semivariance in the time direction can be found at lags of about 9 days. In the marginal semivariograms for ST-OK (Fig. 4), the differences in the semivariance behaviour in the space and time directions are more clearly visible. In addition to the

9 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) Fig. 3. Experimental semivariograms for ST-OK (a), ST-KED (linear) (b) and ST-KED (logarithmic) (c).

10 262 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) Fig. 4. Marginal experimental semivariograms for ST-OK, ST-KED (linear) and ST-KED (logarithmic): S: c(h S,0) (a) and T: c(0, h T ) (b). already mentioned differences, it can be seen that the T behaviour is much smoother than the S behaviour. Besides this, a substantial nugget effect is present in the space direction, whereas it is absent in the time direction. This is caused mainly by small-scale S variation due to texture, vegetation differences, and animal activity (among others molehills). The marginal semivariograms were used to obtain the model forms of the S and T model parts of the Bilonick model. The S part was modelled with a nugget model plus an exponential model. The T part was modelled with solely an exponential model. The T periodicity was not taken into account as it hardly influences the interpolation, because the period is large in relation to the observation density. An idea of how the ST part should be modelled cannot be obtained using marginal semivariograms, but because the other two components showed exponential behaviour, we decided to model the ST part with an exponential model as well. This resulted in a total semivariance model for the ST-OK case: cðh S ; h T Þ¼p S Nugð0Þþq S expðr S Þþq T expðr T Þþq ST expðr ST Þ ð9þ with p S being the S nugget value, q S, q T and q ST being the sill parameters of the S, T, and ST model parts and r S, r T and r ST being the range parameters of these parts, respectively. Recall that the ST semivariance model also contains the ST-anisotropy parameter a.

11 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) Fitting the model to the experimental data is difficult due to the fact that eight parameters must be estimated. We used a weighted least squares method minimizing: nrxof lags i¼1 w i ðĉðh Si ; h Ti Þ cðh Si ; h Ti ÞÞ 2 ð10þ where the weighing factor w i is the quotient pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi of the number of pairs in the lag N(h Si,h Ti ) and the square root of the semivariance ( ĉðh Si ; h Ti Þ). To prevent the iteration to get stuck in a local minimum due to the initial choice of parameters, parameters were fitted with 500 randomly chosen initial parameter sets. The optimization algorithm we used is a standard Matlab subspace trust region algorithm based on the interior-reflective Newton method (Coleman and Li, 1996). Mean values and standard deviation bands for the eight parameters are visualized in Fig. 5. All parameters, except for the temporal range, are estimated rather accurately, suggesting that the influence of local minima is small. The large standard deviation in the temporal range is explained by the periodicity in the marginal temporal semivariogram (see Fig. 4b). This causes the optimized ranges to follow a bimodal distribution. In some cases, the optimization algorithm yields a range of about 8 days, in other cases, a range of about 18 days Spatio-temporal interpolation We examined the behaviour of the ST-OK h predictions at different times around the irrigation times. We used S-block/T-point kriging with 11 m S-blocks, as we were not interested in the variability of h at a smaller S support than 1 m 2, but we were interested in interpolation at exact time points. The top row of Fig. 6 gives ST-OK interpolations at Fig. 5. Parameters for the Bilonick model for ST-OK, ST-KED (linear) and ST-KED (logarithmic) (displayed for each parameter in that order; y: average, -: one standard deviation bands).

12 264 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) Fig. 6. h Interpolations using ST-OK, ST-KED (linear) and ST-KED (logarithmic) for three time points on day 230 (August 17, 2000): 2.30 AM, AM and 8.00 PM. three time points on August 17. Due to local dryer and wetter areas, caused by small-scale spatial variation, all three maps show some spotting. The overall pattern is clear, though. Most striking is that the pattern of irrigation (see Fig. 1) is already visible (as a darker area in the north-east) at 2.30 AM in the morning (left), three hours before the irrigation started. This is because future measurements influence predictions as much as past measurements in kriging, as was mentioned before. At AM (middle) and 8.00 PM (right), it can be seen that the irrigation caused a strong increase in h in a large part of the field and that this area stayed wet. The dry corner in the north-east of the field is caused by a large tree, 5 m outside our study area. To examine the T behaviour, a test location (x = 31.55, y = 19.00; Fig. 1) was selected to test how well measured h time series could be reproduced by ST-OK. The automated TDR measurements of system B were not included in the interpolation as the coverage in the T domain is very high around automated TDR locations, leaving little challenge for interpolation. As the comparison here was against point measurements, we used ST point kriging. In Fig. 7 (top graph), measured and predicted h at the test location are displayed. At times without sudden precipitation, measurements and ST-OK predictions are within the uncertainty limits of plus or minus one standard deviation, but close to irrigation times and heavy rainstorms, the measurements are much higher than the predictions. This is due to the smoothing effect of kriging. In Fig. 8, the kriging standard

13 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) Fig. 7. Measured versus predicted h time series at location (x,y)=(31.55,19.00) for ST-OK (a), ST-KED (linear) (b) and ST-KED (logarithmic) (c). The dashed lines show the + or one standard deviation bands. deviation is displayed at the test location. Close to the manual measurement times the ST-OK kriging standard deviation is small, but further away, the uncertainty quickly increases.

14 266 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) Fig. 8. Kriging standard deviation at location (x,y)=(31.55,19.00) for ST-OK, ST-KED (linear) and ST-KED (logarithmic) Spatio-temporal kriging with external drift The next step in our analysis was performing ST-KED. This requires a linear trend model to incorporate the dynamic secondary information and a semivariance model of the residuals The linear trend model We assumed net-precipitation, here NP total, to be the major influence on the inputs and outputs of water from the soil and therefore on h. It was difficult to decide though, how long in the past net-precipitation information is still informative for the h state at any given point in time t 0. We therefore calculated the cumulative amount of NP total over several intervals with different lengths back in time: NP p ðt 0 Þ¼ Z t0 p t 0 NP total ðtþdt ð11þ with time interval length pa(0.5, 1, 2, 3, 6, 9, 12, 18, 24, 36, 48, 72, 96, 120, 144) in hours. After calculation of the individual NP p s, stepwise multiple regression with the h data being dependent on the NP p s was carried out. In this way, the most informative netprecipitation delay periods were selected. The rule we used for selection is a significant (95%) increase of at least 1% in explained variance (R 2 ). Selection of several netprecipitation delay periods makes it possible to obtain a weighted NP total function with recent NP total weighing stronger than less recent NP total. By doing so, we assume that the relationship between h and NP p is linear. When we look at an example scatter plot of NP 96 versus h (Fig. 9), it becomes clear that assuming linearity is unrealistic because the natural maximum and minimum h values, respectively, the saturated h and the residual h, are ignored when assuming a linear model.

15 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) Fig. 9. Example of the cumulative net-precipitation (NP 96 ) h relationship and three models describing this relationship: linear, sigmoidal and logarithmic. A Bolzmann sigmoidal curve may resemble the relationship of NP p with h more closely. This model has four parameters: mðs; tþ ¼A þðb AÞ 1 : ð12þ 1 exp C NPpðs;tÞ D Parameter A represents a minimum value, B represents a maximum, C represents a horizontal shift and D represents the slope between the levels A and B. The model can be rewritten in a linear trend function with two base-functions with f 1 =1, b 1 = A, 1 f 2 ðs; tþ ¼ 1 exp C NPp D and b 2 = B A. A disadvantage of this model is that it is hard to estimate parameter A, the residual h, as this is far outside the measurement range. Furthermore, the model cannot be written in a linear form where all four parameters are to be estimated, as parameters C and D are integrated in base-function f 2, which should be a known factor. An alternative way to model the nonlinear relationship between NP p and h is to use a logarithmic form: mðs; tþ ¼E þ F lnðnp p ðs; tþþgþ ð13þ where E, F and G represent the vertical shift, the steepness and the horizontal shift of the model, respectively. This model can also be rewritten in linear form with two basefunctions f 1 = 1 and f 2 (s, t)=(ln(np p + G)), and coefficients b 1 = E and b 2 = F. As before, parameter G cannot be isolated as it is integrated in base-function f 2. However, this is less problematic than with the sigmoidal model because the horizontal shift is not a very sensitive parameter, in contrast to the combination of slope and horizontal shift in Eq. (12). The horizontal shift can be estimated beforehand based on the minimum NP p. Problems do occur with the logarithmic model at low h values, where minus infinity can be reached. For the range of h values in this study (Table 1), this is of minor importance.

16 268 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) In Fig. 9, the NP 96 h relationship and the three fitted models are given. Clearly, the A parameter of the sigmoidal curve cannot be estimated very well. Based on this and the fact that the logarithmic model can more easily be rewritten in linear form without loss of important parameters, we chose to use the logarithmic model to describe the nonlinear relationship between h and net-precipitation. The G parameter was chosen to be the minimum of the NP p data per delay period rounded down to the closest integer. As with the linear model, stepwise multiple regression with the same selection rule was used to determine which net-precipitation delay periods best explained h. In order to judge whether using the nonlinear model really improves the interpolations, both models were used in further analysis. The former being referred to as ST-KED (linear) and the latter as ST-KED (logarithmic). The results of the multiple regression of ST-KED (linear) and ST-KED (logarithmic) are given in Table 3. It is remarkable that the selected net-precipitation delay periods are rather long for both models. For both models, the shorter periods were just outside the selection range. This shows that h in our study area has a long memory regarding netprecipitation The spatio-temporal semivariogram The residuals from the multiple regression were used in the semivariance analysis. First the experimental ST semivariograms were calculated (Fig. 3). As with the ST-OK case, we chose exponential models for the S, T and ST parts of the semivariogram model. The ST-KED model parameters and their standard deviations are again visualized in Fig. 5. Comparing the ST-OK case with the ST-KED cases (Figs. 3 5) makes clear that there is a strong decrease in T and ST sills, which is caused by the detrending procedure. For the S direction, the effect of detrending is hardly visible. For the S nugget, this can easily be explained because no information of the small-scale variability was used in the ST trend. With regard to the S sill, it must be concluded that the contribution of the irrigation pattern to spatial variation in h was not very strong. Apparently there are more important sources of spatial variation in h such as soil texture, soil physical properties and vegetation. However, since these were not known in a spatially exhaustive manner, they could not be incorporated in the trend. Table 3 Results of the stepwise multiple regression for ST-KED (linear) and ST-KED (logarithmic); order of included parameters with explained variance R 2 and increase in R 2 per step ST-KED (linear) R 2 R 2 increase ST-KED (logarithmic) R 2 R 2 increase NP NP NP NP NP NP NP NP NP NP The parameters included in the final linear trend models are in bold.

17 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) Spatio-temporal interpolation As with ST-OK, the application of ST-KED is shown by mapping in space and time. The middle and bottom rows of Fig. 6 give the results for ST-KED (linear) and ST-KED (logarithmic) interpolation, respectively. Most striking is the clear imprint of the irrigation pattern in the time points following irrigation. Before irrigation, at 2.30 h, the h map shows a fairly even h picture, this in contrast to the ST-OK map, but in accordance with measurements. In the middle and bottom graphs of Fig. 7, the predicted ST-KED (linear) and ST-KED (logarithmic) time series at the test location are shown. As the h measurements used for the interpolation are the same, the average behaviour of both ST-KED predictions closely resembles the ST-OK predictions, with a slight underestimation of the h at the test location. It is striking though, that ST-KED predictions show more h variations in between manual measurement rounds than the ST-OK predictions. At some time points, even the daily cycle of h becomes clear. The occurrence of this small-scale variation is the result of the small T measurement support (15-min) of the net-precipitation measurements. Fig. 7 also shows that the ST-KED predictions follow the sudden increases in h much better than the ST-OK predictions. When comparing the two ST-KED variants, we see that the ST-KED (logarithmic) predictions behaves best at the sudden increases. This is related to the incorporation of the nonlinearity in the ST-KED (logarithmic) model. By levelling off the relationship between h and net-precipitation close to the saturated h, with the ST- KED (logarithmic) model, a strong increase in h is possible at lower net-precipitation amounts than with the ST-KED (linear) model. The prediction standard deviation at the test location in Fig. 8 shows a different pattern for ST-KED and ST-OK. The main cause for this is that ST-KED considers both the uncertainty due to the ST semivariogram and the uncertainty in estimating the trend parameters. Therefore, the strong decrease in uncertainty close to ST measurement points disappears and the strong increase away from measurement points is lowered because netprecipitation information is available at all points. The strong increase in uncertainty of ST-KED (logarithmic) around day 240 is a local excess, probably due to a local lack of data to adequately fit the linear trend model. 5. Discussion and conclusions In this paper, we showed how the relationship between net-precipitation and h can be used in ST-KED. It can be concluded that this kriging technique, which uses dynamic secondary information, has some clear advantages over ST-OK. Even though the h measurement coverage over the ST domain was high, ST-KED resulted in a decrease in prediction uncertainty. Another improvement was the physically more realistic behaviour of the ST-KED predictions. Especially the ST-KED (logarithmic) variant, which takes the nonlinearity of the relationship between h and net-precipitation into account, better described the sudden changes and the daily cycle in h. Some less positive remarks have to be made here as well. First of all, the reduction of the prediction uncertainty was less pronounced than we expected at first. This is attributed to the crude assumptions we made to keep the trend model simple and manageable. To

18 270 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) reduce the prediction uncertainties further, one can try to add more information (e.g. soil texture, soil physical properties, vegetation) in the trend model. This would require an additional measurement effort, though. Also, including more and more additional information such as physical laws will at some point become extremely difficult due to the rigid structure of ST-KED. For incorporation of more advanced physical process information, a Kalman filtering approach will be much more convenient (e.g. Or and Hanks, 1992; Heuvelink and Webster, 2001; Bierkens et al., 2001), although it should be kept in mind that this requires a major investment in development time. Second, ST-OK also has some important advantages over ST-KED: it needs less data and it is a simpler technique. Although this may seem trivial, it is something one should keep in mind when selecting an interpolation technique. After all, the best technique is not necessarily the technically most advanced one. Acknowledgements NWO-ALW grants (J.J.J.C. Snepvangers) and (J.A. Huisman) financially supported this study. We thank B. Jansen, K. Raat, M. Van Der Velde, A. Visser, P. De Willigen and especially L. De Lange for assistance during the fieldwork period. The comments of P. Bogaert and an anonymous reviewer substantially improved the paper. References Angulo, J.M., Gonzalez-Manteiga, W., Febrero-Bande, M., Alonso, F.J., Semi-parametric statistical approaches for space time process prediction. Environmental and Ecological Statistics 5, Bechini, L., Ducco, G., Donatelli, M., Stein, A., Modelling, interpolation and stochastic simulation in space and time of global solar radiation. Agriculture, Ecosystems & Environment 81, Bierkens, M.F.P., Knotters, M., Hoogland, T., Space time modeling of water table depth using a regionalized time series model and the Kalman filter. Water Resources Research 37, Bilonick, R.A., Monthly hydrogen ion deposition maps for the northeastern U.S. from July 1982 to September Atmospheric Environment 22, Bogaert, P., Comparison of kriging techniques in a space time context. Mathematical Geology 28, Bogaert, P., Christakos, G., 1997a. Spatiotemporal analysis and processing of thermometric data over Belgium. Journal of Geophysical Research 102 (D22), Bogaert, P., Christakos, G., 1997b. Stochastic analysis of spatiotemporal solute content measurement using a regression model. Stochastic Hydrology and Hydraulics 11, Buxton, B.E., Pate, A.D., Estimation of joint temporal spatial semivariograms. In: Baafi, E.Y., Schofield, N.A. (Eds.), Geostatistics Wollongong. Kluwer Academic Publishing, Dordrecht, pp Christakos, G., Vyas, V.M., A composite spatiotemporal study of ozone distribution over eastern United States. Atmospheric Environment 32, Coleman, T.F., Li, Y., An interior, trust region approach for nonlinear minimization subject to bounds. SIAM Journal on Optimization 6, Comegna, V., Vitale, C., Space time analysis of water status in a volcanic Vesuvian soil. Geoderma 60, Cressie, N., Huang, H.C., Classes of nonseperable, spatio-temporal stationary covariance functions. Journal of the American Statistical Association 94, De Cesare, L., Myers, D.E., Posa, D., Spatio-temporal modelling of SO 2 in Milan district. In: Baafi, E.Y., Schofield, N.A. (Eds.), Geostatistics Wollongong. Kluwer Academic Publishing, Dordrecht, pp

19 J.J.J.C. Snepvangers et al. / Geoderma 112 (2003) De Cesare, L., Myers, D.E., Posa, D., 2001a. Estimating and modeling space time correlation structures. Statistics & Probability Letters 51, De Cesare, L., Myers, D.E., Posa, D., 2001b. Product sum covariance for space time modeling: an environmental application. Environmetrics 12, De Iaco, S., Myers, D.E., Posa, D., Nonseperable space time covariance models: some parametric families. Mathematical Geology 34, Goovaerts, P., Geostatistics for natural resources evaluation. Applied Geostatistics Series. Oxford Univ. Press, Oxford. 483 pp. Handcock, M.S., Wallis, J.R., An approach to statistical spatial temporal modeling of meteorological fields. Journal of the American Statistical Association 89, Heimovaara, T., Design of triple-wire time domain reflectometry probes in practice and theory. Soil Science Society of America Journal 57, Heimovaara, T., Bouten, W., A computer-controlled 36-channel time domain reflectometry system for monitoring soil water contents. Water Resources Research 26, Heuvelink, G.B.M., Webster, R., Modelling soil variation: past, present, and future. Geoderma 100, Heuvelink, G.B.M., Musters, P., Pebesma, E.J., Spatio-temporal kriging of soil water content. In: Baafi, E.Y., Schofield, N.A. (Eds.), Geostatistics Wollongong. Kluwer Academic Publishing, Dordrecht, pp Hoosbeek, M.R., Incorporating scale into spatio-temporal variability: applications to soil quality and yield data. Geoderma 85, Kyriakidis, P.C., Journel, A.G., Geostatistical space time models: a review. Mathematical Geology 31, Kyriakidis, P.C., Journel, A.G., Stochastic modeling of atmospheric pollution: a spatial time-series framework: Part I. Methodology. Atmospheric Environment 35, Or, D., Hanks, R.J., Spatial and temporal soil water estimation considering soil variability and evapotranspiration uncertainty. Water Resources Research 28, Rodriguez-Iturbe, I., Mejia, J.M., The design of rainfall networks in time and space. Water Resources Research 10, Rouhani, S., Myers, D.E., Problems in space time kriging of geohydrological data. Mathematical Geology 22, Stein, A., Kocks, C.G., Zadocks, J.C., Frinking, H.D., Russen, N.A., Myers, D.E., A geostatistical analysis of the spatio-temporal development of downy mildew epidemics in cabbage. Ecology and Epidemiology 84, Topp, G.C., Davis, J.L., Annan, A.P., Electromagnetic determination of soil water content: measurements in coaxial transmission lines. Water Resources Research 16, USDA, Soil taxonomy: a basic system of soil classification for making and interpreting soil surveys. Agric. Handbook, vol USDA (United States Department of Agriculture) Soil Conservation Service, Washington, DC, USA.

Space-time analysis using a general product-sum model

Space-time analysis using a general product-sum model Space-time analysis using a general product-sum model De Iaco S., Myers D. E. 2 and Posa D. 3,4 Università di Chieti, Pescara - ITALY; sdeiaco@tiscalinet.it 2 University of Arizona, Tucson AZ - USA; myers@math.arizona.edu

More information

GEOINFORMATICS Vol. II - Stochastic Modelling of Spatio-Temporal Phenomena in Earth Sciences - Soares, A.

GEOINFORMATICS Vol. II - Stochastic Modelling of Spatio-Temporal Phenomena in Earth Sciences - Soares, A. STOCHASTIC MODELLING OF SPATIOTEMPORAL PHENOMENA IN EARTH SCIENCES Soares, A. CMRP Instituto Superior Técnico, University of Lisbon. Portugal Keywords Spacetime models, geostatistics, stochastic simulation

More information

7 Geostatistics. Figure 7.1 Focus of geostatistics

7 Geostatistics. Figure 7.1 Focus of geostatistics 7 Geostatistics 7.1 Introduction Geostatistics is the part of statistics that is concerned with geo-referenced data, i.e. data that are linked to spatial coordinates. To describe the spatial variation

More information

FORTRAN programs for space-time modeling $

FORTRAN programs for space-time modeling $ Computers & Geosciences 28 (2002) 205 212 FORTRAN programs for space-time modeling $ L. De Cesare a,b, D.E. Myers c, D. Posa a,d, * a Facolt "a di Economia, Dipartimento di Scienze Economiche, e Matematico-Statistiche,

More information

Spatiotemporal Analysis of Solar Radiation for Sustainable Research in the Presence of Uncertain Measurements

Spatiotemporal Analysis of Solar Radiation for Sustainable Research in the Presence of Uncertain Measurements Spatiotemporal Analysis of Solar Radiation for Sustainable Research in the Presence of Uncertain Measurements Alexander Kolovos SAS Institute, Inc. alexander.kolovos@sas.com Abstract. The study of incoming

More information

PRODUCING PROBABILITY MAPS TO ASSESS RISK OF EXCEEDING CRITICAL THRESHOLD VALUE OF SOIL EC USING GEOSTATISTICAL APPROACH

PRODUCING PROBABILITY MAPS TO ASSESS RISK OF EXCEEDING CRITICAL THRESHOLD VALUE OF SOIL EC USING GEOSTATISTICAL APPROACH PRODUCING PROBABILITY MAPS TO ASSESS RISK OF EXCEEDING CRITICAL THRESHOLD VALUE OF SOIL EC USING GEOSTATISTICAL APPROACH SURESH TRIPATHI Geostatistical Society of India Assumptions and Geostatistical Variogram

More information

Mapping Precipitation in Switzerland with Ordinary and Indicator Kriging

Mapping Precipitation in Switzerland with Ordinary and Indicator Kriging Journal of Geographic Information and Decision Analysis, vol. 2, no. 2, pp. 65-76, 1998 Mapping Precipitation in Switzerland with Ordinary and Indicator Kriging Peter M. Atkinson Department of Geography,

More information

Spatio-temporal statistical models for river monitoring networks

Spatio-temporal statistical models for river monitoring networks Spatio-temporal statistical models for river monitoring networks L. Clement, O. Thas, P.A. Vanrolleghem and J.P. Ottoy Department of Applied Mathematics, Biometrics and Process Control, Ghent University,

More information

Introduction. Semivariogram Cloud

Introduction. Semivariogram Cloud Introduction Data: set of n attribute measurements {z(s i ), i = 1,, n}, available at n sample locations {s i, i = 1,, n} Objectives: Slide 1 quantify spatial auto-correlation, or attribute dissimilarity

More information

What s for today. Introduction to Space-time models. c Mikyoung Jun (Texas A&M) Stat647 Lecture 14 October 16, / 19

What s for today. Introduction to Space-time models. c Mikyoung Jun (Texas A&M) Stat647 Lecture 14 October 16, / 19 What s for today Introduction to Space-time models c Mikyoung Jun (Texas A&M) Stat647 Lecture 14 October 16, 2012 1 / 19 Space-time Data So far we looked at the data that vary over space Now we add another

More information

ARTICLE IN PRESS. Computers & Geosciences

ARTICLE IN PRESS. Computers & Geosciences Computers & Geosciences 36 (2010) 636 646 Contents lists available at ScienceDirect Computers & Geosciences journal homepage: www.elsevier.com/locate/cageo FORTRAN programs for space time multivariate

More information

4th HR-HU and 15th HU geomathematical congress Geomathematics as Geoscience Reliability enhancement of groundwater estimations

4th HR-HU and 15th HU geomathematical congress Geomathematics as Geoscience Reliability enhancement of groundwater estimations Reliability enhancement of groundwater estimations Zoltán Zsolt Fehér 1,2, János Rakonczai 1, 1 Institute of Geoscience, University of Szeged, H-6722 Szeged, Hungary, 2 e-mail: zzfeher@geo.u-szeged.hu

More information

Flux Tower Data Quality Analysis in the North American Monsoon Region

Flux Tower Data Quality Analysis in the North American Monsoon Region Flux Tower Data Quality Analysis in the North American Monsoon Region 1. Motivation The area of focus in this study is mainly Arizona, due to data richness and availability. Monsoon rains in Arizona usually

More information

Investigation of Monthly Pan Evaporation in Turkey with Geostatistical Technique

Investigation of Monthly Pan Evaporation in Turkey with Geostatistical Technique Investigation of Monthly Pan Evaporation in Turkey with Geostatistical Technique Hatice Çitakoğlu 1, Murat Çobaner 1, Tefaruk Haktanir 1, 1 Department of Civil Engineering, Erciyes University, Kayseri,

More information

Modeling of Atmospheric Effects on InSAR Measurements With the Method of Stochastic Simulation

Modeling of Atmospheric Effects on InSAR Measurements With the Method of Stochastic Simulation Modeling of Atmospheric Effects on InSAR Measurements With the Method of Stochastic Simulation Z. W. LI, X. L. DING Department of Land Surveying and Geo-Informatics, Hong Kong Polytechnic University, Hung

More information

Gridding of precipitation and air temperature observations in Belgium. Michel Journée Royal Meteorological Institute of Belgium (RMI)

Gridding of precipitation and air temperature observations in Belgium. Michel Journée Royal Meteorological Institute of Belgium (RMI) Gridding of precipitation and air temperature observations in Belgium Michel Journée Royal Meteorological Institute of Belgium (RMI) Gridding of meteorological data A variety of hydrologic, ecological,

More information

Conditional Distribution Fitting of High Dimensional Stationary Data

Conditional Distribution Fitting of High Dimensional Stationary Data Conditional Distribution Fitting of High Dimensional Stationary Data Miguel Cuba and Oy Leuangthong The second order stationary assumption implies the spatial variability defined by the variogram is constant

More information

MEASURING SNOW WATER EQUIVALENT AND SNOW DENSITY USING TDR MINI-PROBES

MEASURING SNOW WATER EQUIVALENT AND SNOW DENSITY USING TDR MINI-PROBES MEASURING SNOW WATER EQUIVALENT AND SNOW DENSITY USING TDR MINI-PROBES Paper No. 05-049 M. Krishnapillai, Ph.D. Department of Biosystems Engineering University of Manitoba Winnipeg MB R3T 5V6 R. Sri Ranjan,

More information

Spatial Analysis II. Spatial data analysis Spatial analysis and inference

Spatial Analysis II. Spatial data analysis Spatial analysis and inference Spatial Analysis II Spatial data analysis Spatial analysis and inference Roadmap Spatial Analysis I Outline: What is spatial analysis? Spatial Joins Step 1: Analysis of attributes Step 2: Preparing for

More information

Soil Moisture Modeling using Geostatistical Techniques at the O Neal Ecological Reserve, Idaho

Soil Moisture Modeling using Geostatistical Techniques at the O Neal Ecological Reserve, Idaho Final Report: Forecasting Rangeland Condition with GIS in Southeastern Idaho Soil Moisture Modeling using Geostatistical Techniques at the O Neal Ecological Reserve, Idaho Jacob T. Tibbitts, Idaho State

More information

Chapter 1. Summer School GEOSTAT 2014, Spatio-Temporal Geostatistics,

Chapter 1. Summer School GEOSTAT 2014, Spatio-Temporal Geostatistics, Chapter 1 Summer School GEOSTAT 2014, Geostatistics, 2014-06-19 sum- http://ifgi.de/graeler Institute for Geoinformatics University of Muenster 1.1 Spatial Data From a purely statistical perspective, spatial

More information

Advanced analysis and modelling tools for spatial environmental data. Case study: indoor radon data in Switzerland

Advanced analysis and modelling tools for spatial environmental data. Case study: indoor radon data in Switzerland EnviroInfo 2004 (Geneva) Sh@ring EnviroInfo 2004 Advanced analysis and modelling tools for spatial environmental data. Case study: indoor radon data in Switzerland Mikhail Kanevski 1, Michel Maignan 1

More information

Automatic Determination of Uncertainty versus Data Density

Automatic Determination of Uncertainty versus Data Density Automatic Determination of Uncertainty versus Data Density Brandon Wilde and Clayton V. Deutsch It is useful to know how various measures of uncertainty respond to changes in data density. Calculating

More information

Influence of parameter estimation uncertainty in Kriging: Part 2 Test and case study applications

Influence of parameter estimation uncertainty in Kriging: Part 2 Test and case study applications Hydrology and Earth System Influence Sciences, of 5(), parameter 5 3 estimation (1) uncertainty EGS in Kriging: Part Test and case study applications Influence of parameter estimation uncertainty in Kriging:

More information

Supplementary material: Methodological annex

Supplementary material: Methodological annex 1 Supplementary material: Methodological annex Correcting the spatial representation bias: the grid sample approach Our land-use time series used non-ideal data sources, which differed in spatial and thematic

More information

Toward an automatic real-time mapping system for radiation hazards

Toward an automatic real-time mapping system for radiation hazards Toward an automatic real-time mapping system for radiation hazards Paul H. Hiemstra 1, Edzer J. Pebesma 2, Chris J.W. Twenhöfel 3, Gerard B.M. Heuvelink 4 1 Faculty of Geosciences / University of Utrecht

More information

SPI: Standardized Precipitation Index

SPI: Standardized Precipitation Index PRODUCT FACT SHEET: SPI Africa Version 1 (May. 2013) SPI: Standardized Precipitation Index Type Temporal scale Spatial scale Geo. coverage Precipitation Monthly Data dependent Africa (for a range of accumulation

More information

Site-specific Prediction of Mosquito Abundance using Spatio-Temporal Geostatistics

Site-specific Prediction of Mosquito Abundance using Spatio-Temporal Geostatistics Site-specific Prediction of Mosquito Abundance using Spatio-Temporal Geostatistics E.-H. Yoo 1, D. Chen 2 and C. Russell 3 1 Department of Geography, University at Buffalo, SUNY, Buffalo, NY, USA eunhye@buffalo.edu,

More information

2006 Drought in the Netherlands (20 July 2006)

2006 Drought in the Netherlands (20 July 2006) 2006 Drought in the Netherlands (20 July 2006) Henny A.J. van Lanen, Wageningen University, the Netherlands (henny.vanlanen@wur.nl) The Netherlands is suffering from tropical heat and it is facing a meteorological

More information

Research Note COMPUTER PROGRAM FOR ESTIMATING CROP EVAPOTRANSPIRATION IN PUERTO RICO 1,2. J. Agric. Univ. P.R. 89(1-2): (2005)

Research Note COMPUTER PROGRAM FOR ESTIMATING CROP EVAPOTRANSPIRATION IN PUERTO RICO 1,2. J. Agric. Univ. P.R. 89(1-2): (2005) Research Note COMPUTER PROGRAM FOR ESTIMATING CROP EVAPOTRANSPIRATION IN PUERTO RICO 1,2 Eric W. Harmsen 3 and Antonio L. González-Pérez 4 J. Agric. Univ. P.R. 89(1-2):107-113 (2005) Estimates of crop

More information

Spatiotemporal Analysis of Environmental Radiation in Korea

Spatiotemporal Analysis of Environmental Radiation in Korea WM 0 Conference, February 25 - March, 200, Tucson, AZ Spatiotemporal Analysis of Environmental Radiation in Korea J.Y. Kim, B.C. Lee FNC Technology Co., Ltd. Main Bldg. 56, Seoul National University Research

More information

Transiogram: A spatial relationship measure for categorical data

Transiogram: A spatial relationship measure for categorical data International Journal of Geographical Information Science Vol. 20, No. 6, July 2006, 693 699 Technical Note Transiogram: A spatial relationship measure for categorical data WEIDONG LI* Department of Geography,

More information

Geog 210C Spring 2011 Lab 6. Geostatistics in ArcMap

Geog 210C Spring 2011 Lab 6. Geostatistics in ArcMap Geog 210C Spring 2011 Lab 6. Geostatistics in ArcMap Overview In this lab you will think critically about the functionality of spatial interpolation, improve your kriging skills, and learn how to use several

More information

A Short Note on the Proportional Effect and Direct Sequential Simulation

A Short Note on the Proportional Effect and Direct Sequential Simulation A Short Note on the Proportional Effect and Direct Sequential Simulation Abstract B. Oz (boz@ualberta.ca) and C. V. Deutsch (cdeutsch@ualberta.ca) University of Alberta, Edmonton, Alberta, CANADA Direct

More information

On the spatial scaling of soil moisture

On the spatial scaling of soil moisture Journal of Hydrology 217 (1999) 203 224 On the spatial scaling of soil moisture Andrew W. Western a, *,Günter Blöschl b a Centre for Environmental Applied Hydrology, Department of Civil and Environmental

More information

Types of Spatial Data

Types of Spatial Data Spatial Data Types of Spatial Data Point pattern Point referenced geostatistical Block referenced Raster / lattice / grid Vector / polygon Point Pattern Data Interested in the location of points, not their

More information

Snow Melt with the Land Climate Boundary Condition

Snow Melt with the Land Climate Boundary Condition Snow Melt with the Land Climate Boundary Condition GEO-SLOPE International Ltd. www.geo-slope.com 1200, 700-6th Ave SW, Calgary, AB, Canada T2P 0T8 Main: +1 403 269 2002 Fax: +1 888 463 2239 Introduction

More information

GEOSTATISTICAL ANALYSIS OF SPATIAL DATA. Goovaerts, P. Biomedware, Inc. and PGeostat, LLC, Ann Arbor, Michigan, USA

GEOSTATISTICAL ANALYSIS OF SPATIAL DATA. Goovaerts, P. Biomedware, Inc. and PGeostat, LLC, Ann Arbor, Michigan, USA GEOSTATISTICAL ANALYSIS OF SPATIAL DATA Goovaerts, P. Biomedware, Inc. and PGeostat, LLC, Ann Arbor, Michigan, USA Keywords: Semivariogram, kriging, spatial patterns, simulation, risk assessment Contents

More information

Models for spatial data (cont d) Types of spatial data. Types of spatial data (cont d) Hierarchical models for spatial data

Models for spatial data (cont d) Types of spatial data. Types of spatial data (cont d) Hierarchical models for spatial data Hierarchical models for spatial data Based on the book by Banerjee, Carlin and Gelfand Hierarchical Modeling and Analysis for Spatial Data, 2004. We focus on Chapters 1, 2 and 5. Geo-referenced data arise

More information

C. BRANQUINHO Museu Laboratório Faculdade de Ciências da Universidade de Lisboa,

C. BRANQUINHO Museu Laboratório Faculdade de Ciências da Universidade de Lisboa, GEOSTATISTICAL MODELS FOR AIR POLLUTION M.J. PEREIRA, A. SOARES, J. ALMEIDA CMRP/Instituto Superior Técnico C. BRANQUINHO Museu Laboratório Faculdade de Ciências da Universidade de Lisboa, Lisbon, Portugal

More information

Umeå University Sara Sjöstedt-de Luna Time series analysis and spatial statistics

Umeå University Sara Sjöstedt-de Luna Time series analysis and spatial statistics Umeå University 01-05-5 Sara Sjöstedt-de Luna Time series analysis and spatial statistics Laboration in ArcGIS Geostatistical Analyst These exercises are aiming at helping you understand ArcGIS Geostatistical

More information

Geostatistics in Hydrology: Kriging interpolation

Geostatistics in Hydrology: Kriging interpolation Chapter Geostatistics in Hydrology: Kriging interpolation Hydrologic properties, such as rainfall, aquifer characteristics (porosity, hydraulic conductivity, transmissivity, storage coefficient, etc.),

More information

Spatial-Temporal Modeling of Active Layer Thickness

Spatial-Temporal Modeling of Active Layer Thickness Spatial-Temporal Modeling of Active Layer Thickness Qian Chen Advisor : Dr. Tatiyana Apanasovich Department of Statistics The George Washington University Abstract The objective of this study is to provide

More information

Monitoring daily evapotranspiration in the Alps exploiting Sentinel-2 and meteorological data

Monitoring daily evapotranspiration in the Alps exploiting Sentinel-2 and meteorological data Monitoring daily evapotranspiration in the Alps exploiting Sentinel-2 and meteorological data M. Castelli, S. Asam, A. Jacob, M. Zebisch, and C. Notarnicola Institute for Earth Observation, Eurac Research,

More information

Estimation of direction of increase of gold mineralisation using pair-copulas

Estimation of direction of increase of gold mineralisation using pair-copulas 22nd International Congress on Modelling and Simulation, Hobart, Tasmania, Australia, 3 to 8 December 2017 mssanz.org.au/modsim2017 Estimation of direction of increase of gold mineralisation using pair-copulas

More information

10. FIELD APPLICATION: 1D SOIL MOISTURE PROFILE ESTIMATION

10. FIELD APPLICATION: 1D SOIL MOISTURE PROFILE ESTIMATION Chapter 1 Field Application: 1D Soil Moisture Profile Estimation Page 1-1 CHAPTER TEN 1. FIELD APPLICATION: 1D SOIL MOISTURE PROFILE ESTIMATION The computationally efficient soil moisture model ABDOMEN,

More information

Kriging method for estimation of groundwater resources in a basin with scarce monitoring data

Kriging method for estimation of groundwater resources in a basin with scarce monitoring data 36 New Approaches to Hydrological Prediction in Data-sparse Regions (Proc. of Symposium HS. at the Joint IAHS & IAH Convention, Hyderabad, India, September 9). IAHS Publ. 333, 9. Kriging method for estimation

More information

Space-time data. Simple space-time analyses. PM10 in space. PM10 in time

Space-time data. Simple space-time analyses. PM10 in space. PM10 in time Space-time data Observations taken over space and over time Z(s, t): indexed by space, s, and time, t Here, consider geostatistical/time data Z(s, t) exists for all locations and all times May consider

More information

12 SWAT USER S MANUAL, VERSION 98.1

12 SWAT USER S MANUAL, VERSION 98.1 12 SWAT USER S MANUAL, VERSION 98.1 CANOPY STORAGE. Canopy storage is the water intercepted by vegetative surfaces (the canopy) where it is held and made available for evaporation. When using the curve

More information

Improving geological models using a combined ordinary indicator kriging approach

Improving geological models using a combined ordinary indicator kriging approach Engineering Geology 69 (2003) 37 45 www.elsevier.com/locate/enggeo Improving geological models using a combined ordinary indicator kriging approach O. Marinoni* Institute for Applied Earth Sciences, Georesources,

More information

Journal of Pharmacognosy and Phytochemistry 2017; 6(4): Sujitha E and Shanmugasundaram K

Journal of Pharmacognosy and Phytochemistry 2017; 6(4): Sujitha E and Shanmugasundaram K 2017; 6(4): 452-457 E-ISSN: 2278-4136 P-ISSN: 2349-8234 JPP 2017; 6(4): 452-457 Received: 01-05-2017 Accepted: 02-06-2017 Sujitha E Research Scholar, Department of Soil and Water Conservation Engineering,

More information

KEY WORDS: Palmer Meteorological Drought Index, SWAP, Kriging spatial analysis and Digital Map.

KEY WORDS: Palmer Meteorological Drought Index, SWAP, Kriging spatial analysis and Digital Map. PALMER METEOROLOGICAL DROUGHT CLASSIFICATION USING TECHNIQUES OF GEOGRAPHIC INFORMATION SYSTEM IN THAILAND S. Baimoung, W. Waranuchit, S. Prakanrat, P. Amatayakul, N. Sukhanthamat, A. Yuthaphan, A. Pyomjamsri,

More information

Problems in Space-Time Kriging of Geohydrological Data 1. Shahrokh Rouhani 2 and Donald E. Myers 3

Problems in Space-Time Kriging of Geohydrological Data 1. Shahrokh Rouhani 2 and Donald E. Myers 3 Mathematical Geology, Vol. 22, No. 5, 1990 Problems in Space-Time Kriging of Geohydrological Data 1 Shahrokh Rouhani 2 and Donald E. Myers 3 Spatiotemporal variables constitute a large class of geohydrological

More information

AN OPERATIONAL DROUGHT MONITORING SYSTEM USING SPATIAL INTERPOLATION METHODS FOR PINIOS RIVER BASIN, GREECE

AN OPERATIONAL DROUGHT MONITORING SYSTEM USING SPATIAL INTERPOLATION METHODS FOR PINIOS RIVER BASIN, GREECE Proceedings of the 13 th International Conference on Environmental Science and Technology Athens, Greece, 5-7 September 2013 AN OPERATIONAL DROUGHT MONITORING SYSTEM USING SPATIAL INTERPOLATION METHODS

More information

Space time variograms and a functional form for total air pollution measurements

Space time variograms and a functional form for total air pollution measurements Computational Statistics & Data Analysis 41 (2002) 311 328 www.elsevier.com/locate/csda Space time variograms and a functional form for total air pollution measurements S. De Iaco a, D.E. Myers b, D. Posa

More information

A MultiGaussian Approach to Assess Block Grade Uncertainty

A MultiGaussian Approach to Assess Block Grade Uncertainty A MultiGaussian Approach to Assess Block Grade Uncertainty Julián M. Ortiz 1, Oy Leuangthong 2, and Clayton V. Deutsch 2 1 Department of Mining Engineering, University of Chile 2 Department of Civil &

More information

A simple non-separable, non-stationary spatiotemporal model for ozone

A simple non-separable, non-stationary spatiotemporal model for ozone Environ Ecol Stat (2009) 16:515 529 DOI 10.1007/s10651-008-0094-8 A simple non-separable, non-stationary spatiotemporal model for ozone Francesca Bruno Peter Guttorp Paul D. Sampson Daniela Cocchi Received:

More information

Advances in Locally Varying Anisotropy With MDS

Advances in Locally Varying Anisotropy With MDS Paper 102, CCG Annual Report 11, 2009 ( 2009) Advances in Locally Varying Anisotropy With MDS J.B. Boisvert and C. V. Deutsch Often, geology displays non-linear features such as veins, channels or folds/faults

More information

Uncertainty in merged radar - rain gauge rainfall products

Uncertainty in merged radar - rain gauge rainfall products Uncertainty in merged radar - rain gauge rainfall products Francesca Cecinati University of Bristol francesca.cecinati@bristol.ac.uk Supervisor: Miguel A. Rico-Ramirez This project has received funding

More information

ENVIRONMENTAL MONITORING Vol. II - Geostatistical Analysis of Monitoring Data - Mark Dowdall, John O Dea GEOSTATISTICAL ANALYSIS OF MONITORING DATA

ENVIRONMENTAL MONITORING Vol. II - Geostatistical Analysis of Monitoring Data - Mark Dowdall, John O Dea GEOSTATISTICAL ANALYSIS OF MONITORING DATA GEOSTATISTICAL ANALYSIS OF MONITORING DATA Mark Dowdall Norwegian Radiation Protection Authority, Environmental Protection Unit, Polar Environmental Centre, Tromso, Norway John O Dea Institute of Technology,

More information

Efficient geostatistical simulation for spatial uncertainty propagation

Efficient geostatistical simulation for spatial uncertainty propagation Efficient geostatistical simulation for spatial uncertainty propagation Stelios Liodakis University of the Aegean University Hill Mytilene, Greece stelioslio@geo.aegean.gr Phaedon Kyriakidis Cyprus University

More information

Introduction to Geostatistics

Introduction to Geostatistics Introduction to Geostatistics Abhi Datta 1, Sudipto Banerjee 2 and Andrew O. Finley 3 July 31, 2017 1 Department of Biostatistics, Bloomberg School of Public Health, Johns Hopkins University, Baltimore,

More information

New Classes of Asymmetric Spatial-Temporal Covariance Models. Man Sik Park and Montserrat Fuentes 1

New Classes of Asymmetric Spatial-Temporal Covariance Models. Man Sik Park and Montserrat Fuentes 1 New Classes of Asymmetric Spatial-Temporal Covariance Models Man Sik Park and Montserrat Fuentes 1 Institute of Statistics Mimeo Series# 2584 SUMMARY Environmental spatial data often show complex spatial-temporal

More information

Promoting Rainwater Harvesting in Caribbean Small Island Developing States Water Availability Mapping for Grenada Preliminary findings

Promoting Rainwater Harvesting in Caribbean Small Island Developing States Water Availability Mapping for Grenada Preliminary findings Promoting Rainwater Harvesting in Caribbean Small Island Developing States Water Availability Mapping for Grenada Preliminary findings National Workshop Pilot Project funded by The United Nations Environment

More information

Sensitivity to the composition of the feature vector and passive simulations

Sensitivity to the composition of the feature vector and passive simulations Sensitivity to the composition of the feature vector and passive simulations Rainfall Generator for the Rhine Basin Jules J. Beersma De Bilt, 2011 KNMI publication 186-VI Rainfall Generator for the Rhine

More information

Gridded monthly temperature fields for Croatia for the period

Gridded monthly temperature fields for Croatia for the period Gridded monthly temperature fields for Croatia for the 1981 2010 period comparison with the similar global and European products Melita Perčec Tadid melita.percec.tadic@cirus.dhz.hr Meteorological and

More information

An Introduction to Spatial Statistics. Chunfeng Huang Department of Statistics, Indiana University

An Introduction to Spatial Statistics. Chunfeng Huang Department of Statistics, Indiana University An Introduction to Spatial Statistics Chunfeng Huang Department of Statistics, Indiana University Microwave Sounding Unit (MSU) Anomalies (Monthly): 1979-2006. Iron Ore (Cressie, 1986) Raw percent data

More information

Introduction. Spatial Processes & Spatial Patterns

Introduction. Spatial Processes & Spatial Patterns Introduction Spatial data: set of geo-referenced attribute measurements: each measurement is associated with a location (point) or an entity (area/region/object) in geographical (or other) space; the domain

More information

Kriging Luc Anselin, All Rights Reserved

Kriging Luc Anselin, All Rights Reserved Kriging Luc Anselin Spatial Analysis Laboratory Dept. Agricultural and Consumer Economics University of Illinois, Urbana-Champaign http://sal.agecon.uiuc.edu Outline Principles Kriging Models Spatial Interpolation

More information

ANALYSIS OF DEPTH-AREA-DURATION CURVES OF RAINFALL IN SEMIARID AND ARID REGIONS USING GEOSTATISTICAL METHODS: SIRJAN KAFEH NAMAK WATERSHED, IRAN

ANALYSIS OF DEPTH-AREA-DURATION CURVES OF RAINFALL IN SEMIARID AND ARID REGIONS USING GEOSTATISTICAL METHODS: SIRJAN KAFEH NAMAK WATERSHED, IRAN JOURNAL OF ENVIRONMENTAL HYDROLOGY The Electronic Journal of the International Association for Environmental Hydrology On the World Wide Web at http://www.hydroweb.com VOLUME 14 2006 ANALYSIS OF DEPTH-AREA-DURATION

More information

Time to Depth Conversion and Uncertainty Characterization for SAGD Base of Pay in the McMurray Formation, Alberta, Canada*

Time to Depth Conversion and Uncertainty Characterization for SAGD Base of Pay in the McMurray Formation, Alberta, Canada* Time to Depth Conversion and Uncertainty Characterization for SAGD Base of Pay in the McMurray Formation, Alberta, Canada* Amir H. Hosseini 1, Hong Feng 1, Abu Yousuf 1, and Tony Kay 1 Search and Discovery

More information

Index. Geostatistics for Environmental Scientists, 2nd Edition R. Webster and M. A. Oliver 2007 John Wiley & Sons, Ltd. ISBN:

Index. Geostatistics for Environmental Scientists, 2nd Edition R. Webster and M. A. Oliver 2007 John Wiley & Sons, Ltd. ISBN: Index Akaike information criterion (AIC) 105, 290 analysis of variance 35, 44, 127 132 angular transformation 22 anisotropy 59, 99 affine or geometric 59, 100 101 anisotropy ratio 101 exploring and displaying

More information

Southern California ground motion envelopes over ranges of magnitudes, distances, and site conditions

Southern California ground motion envelopes over ranges of magnitudes, distances, and site conditions 55 Chapter 3 Average properties of Southern California ground motion envelopes over ranges of magnitudes, distances, and site conditions In this chapter, I use the envelope attenuation relationships derived

More information

Traps for the Unwary Subsurface Geoscientist

Traps for the Unwary Subsurface Geoscientist Traps for the Unwary Subsurface Geoscientist ashley.francis@sorviodvnvm.co.uk http://www.sorviodvnvm.co.uk Presented at SEG Development & Production Forum, 24-29 th June 2001, Taos, New Mexico, USA 24-29

More information

Optimizing Thresholds in Truncated Pluri-Gaussian Simulation

Optimizing Thresholds in Truncated Pluri-Gaussian Simulation Optimizing Thresholds in Truncated Pluri-Gaussian Simulation Samaneh Sadeghi and Jeff B. Boisvert Truncated pluri-gaussian simulation (TPGS) is an extension of truncated Gaussian simulation. This method

More information

POPULAR CARTOGRAPHIC AREAL INTERPOLATION METHODS VIEWED FROM A GEOSTATISTICAL PERSPECTIVE

POPULAR CARTOGRAPHIC AREAL INTERPOLATION METHODS VIEWED FROM A GEOSTATISTICAL PERSPECTIVE CO-282 POPULAR CARTOGRAPHIC AREAL INTERPOLATION METHODS VIEWED FROM A GEOSTATISTICAL PERSPECTIVE KYRIAKIDIS P. University of California Santa Barbara, MYTILENE, GREECE ABSTRACT Cartographic areal interpolation

More information

A Geostatistical Approach to Predict the Average Annual Rainfall of Bangladesh

A Geostatistical Approach to Predict the Average Annual Rainfall of Bangladesh Journal of Data Science 14(2016), 149-166 A Geostatistical Approach to Predict the Average Annual Rainfall of Bangladesh Mohammad Samsul Alam 1 and Syed Shahadat Hossain 1 1 Institute of Statistical Research

More information

Kriging of spatial-temporal water vapor data

Kriging of spatial-temporal water vapor data Kriging of spatial-temporal water vapor data Roderik Lindenbergh, Maxim Keshin, Hans van der Marel and Ramon Hanssen Delft Institute of Earth Observation and Space Systems, Delft University of Technology,

More information

CURRENT STATUS OF SCIAMACHY POLARISATION MEASUREMENTS. J.M. Krijger 1 and L.G. Tilstra 2

CURRENT STATUS OF SCIAMACHY POLARISATION MEASUREMENTS. J.M. Krijger 1 and L.G. Tilstra 2 % % CURRENT STATUS OF SCIAMACHY POLARISATION MEASUREMENTS JM Krijger 1 and LG Tilstra 2 1 SRON (National Institute for Space Research), Sorbonnelaan 2, 3584 CA Utrecht, The Netherlands, krijger@sronnl

More information

Spatial Interpolation & Geostatistics

Spatial Interpolation & Geostatistics (Z i Z j ) 2 / 2 Spatial Interpolation & Geostatistics Lag Lag Mean Distance between pairs of points 1 y Kriging Step 1 Describe spatial variation with Semivariogram (Z i Z j ) 2 / 2 Point cloud Map 3

More information

Spatial Data Mining. Regression and Classification Techniques

Spatial Data Mining. Regression and Classification Techniques Spatial Data Mining Regression and Classification Techniques 1 Spatial Regression and Classisfication Discrete class labels (left) vs. continues quantities (right) measured at locations (2D for geographic

More information

Practical 12: Geostatistics

Practical 12: Geostatistics Practical 12: Geostatistics This practical will introduce basic tools for geostatistics in R. You may need first to install and load a few packages. The packages sp and lattice contain useful function

More information

Nonstationary models for exploring and mapping monthly precipitation in the United Kingdom

Nonstationary models for exploring and mapping monthly precipitation in the United Kingdom INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 3: 39 45 (21) Published online 16 March 29 in Wiley InterScience (www.interscience.wiley.com) DOI: 1.12/joc.1892 Nonstationary models for exploring

More information

Stanford Exploration Project, Report 105, September 5, 2000, pages 41 53

Stanford Exploration Project, Report 105, September 5, 2000, pages 41 53 Stanford Exploration Project, Report 105, September 5, 2000, pages 41 53 40 Stanford Exploration Project, Report 105, September 5, 2000, pages 41 53 Short Note Multiple realizations using standard inversion

More information

Chapter 2 Agro-meteorological Observatory

Chapter 2 Agro-meteorological Observatory Chapter 2 Agro-meteorological Observatory Abstract A Meteorological observatory is an area where all the weather instruments and structures are installed. The chapter gives a description of a meteorological

More information

Quantifying uncertainty of geological 3D layer models, constructed with a-priori

Quantifying uncertainty of geological 3D layer models, constructed with a-priori Quantifying uncertainty of geological 3D layer models, constructed with a-priori geological expertise Jan Gunnink, Denise Maljers 2 and Jan Hummelman 2, TNO Built Environment and Geosciences Geological

More information

Analysis of Radar-Rainfall Uncertainties and effects on Hydrologic Applications. Emad Habib, Ph.D., P.E. University of Louisiana at Lafayette

Analysis of Radar-Rainfall Uncertainties and effects on Hydrologic Applications. Emad Habib, Ph.D., P.E. University of Louisiana at Lafayette Analysis of Radar-Rainfall Uncertainties and effects on Hydrologic Applications Emad Habib, Ph.D., P.E. University of Louisiana at Lafayette Motivation Rainfall is a process with significant variability

More information

Exploring the World of Ordinary Kriging. Dennis J. J. Walvoort. Wageningen University & Research Center Wageningen, The Netherlands

Exploring the World of Ordinary Kriging. Dennis J. J. Walvoort. Wageningen University & Research Center Wageningen, The Netherlands Exploring the World of Ordinary Kriging Wageningen University & Research Center Wageningen, The Netherlands July 2004 (version 0.2) What is? What is it about? Potential Users a computer program for exploring

More information

USING GEOSTATISTICS TO DESCRIBE COMPLEX A PRIORI INFORMATION FOR INVERSE PROBLEMS THOMAS M. HANSEN 1,2, KLAUS MOSEGAARD 2 and KNUD S.

USING GEOSTATISTICS TO DESCRIBE COMPLEX A PRIORI INFORMATION FOR INVERSE PROBLEMS THOMAS M. HANSEN 1,2, KLAUS MOSEGAARD 2 and KNUD S. USING GEOSTATISTICS TO DESCRIBE COMPLEX A PRIORI INFORMATION FOR INVERSE PROBLEMS THOMAS M. HANSEN 1,2, KLAUS MOSEGAARD 2 and KNUD S. CORDUA 1 1 Institute of Geography & Geology, University of Copenhagen,

More information

THE SAMPLE AREA OF TIME DOMAIN REFLECTOMETRY PROBES IN PROXIMITY TO SHARP DIELECTRIC PERMITTIVITY BOUNDARIES

THE SAMPLE AREA OF TIME DOMAIN REFLECTOMETRY PROBES IN PROXIMITY TO SHARP DIELECTRIC PERMITTIVITY BOUNDARIES THE SAMPLE AREA OF TIME DOMAIN REFLECTOMETRY PROBES IN PROXIMITY TO SHARP DIELECTRIC PERMITTIVITY BOUNDARIES Paul A. Ferré^, Henrik H. Nissen*, Per Moldrup* and John H. Knight# *Department of Environmental

More information

Multiple realizations using standard inversion techniques a

Multiple realizations using standard inversion techniques a Multiple realizations using standard inversion techniques a a Published in SEP report, 105, 67-78, (2000) Robert G Clapp 1 INTRODUCTION When solving a missing data problem, geophysicists and geostatisticians

More information

Observed and Predicted Daily Wind Travels and Wind Speeds in Western Iraq

Observed and Predicted Daily Wind Travels and Wind Speeds in Western Iraq International Journal of Science and Engineering Investigations vol., issue, April ISSN: - Observed and Predicted Daily Wind Travels and Wind Speeds in Western Iraq Ahmed Hasson, Farhan Khammas, Department

More information

Multivariate autoregressive modelling and conditional simulation of precipitation time series for urban water models

Multivariate autoregressive modelling and conditional simulation of precipitation time series for urban water models European Water 57: 299-306, 2017. 2017 E.W. Publications Multivariate autoregressive modelling and conditional simulation of precipitation time series for urban water models J.A. Torres-Matallana 1,3*,

More information

Time-lapse filtering and improved repeatability with automatic factorial co-kriging. Thierry Coléou CGG Reservoir Services Massy

Time-lapse filtering and improved repeatability with automatic factorial co-kriging. Thierry Coléou CGG Reservoir Services Massy Time-lapse filtering and improved repeatability with automatic factorial co-kriging. Thierry Coléou CGG Reservoir Services Massy 1 Outline Introduction Variogram and Autocorrelation Factorial Kriging Factorial

More information

Multiple realizations: Model variance and data uncertainty

Multiple realizations: Model variance and data uncertainty Stanford Exploration Project, Report 108, April 29, 2001, pages 1?? Multiple realizations: Model variance and data uncertainty Robert G. Clapp 1 ABSTRACT Geophysicists typically produce a single model,

More information

Introduction to Spatial Data and Models

Introduction to Spatial Data and Models Introduction to Spatial Data and Models Sudipto Banerjee 1 and Andrew O. Finley 2 1 Department of Forestry & Department of Geography, Michigan State University, Lansing Michigan, U.S.A. 2 Biostatistics,

More information

GENERALIZED LINEAR MODELING APPROACH TO STOCHASTIC WEATHER GENERATORS

GENERALIZED LINEAR MODELING APPROACH TO STOCHASTIC WEATHER GENERATORS GENERALIZED LINEAR MODELING APPROACH TO STOCHASTIC WEATHER GENERATORS Rick Katz Institute for Study of Society and Environment National Center for Atmospheric Research Boulder, CO USA Joint work with Eva

More information

11/8/2018. Spatial Interpolation & Geostatistics. Kriging Step 1

11/8/2018. Spatial Interpolation & Geostatistics. Kriging Step 1 (Z i Z j ) 2 / 2 (Z i Zj) 2 / 2 Semivariance y 11/8/2018 Spatial Interpolation & Geostatistics Kriging Step 1 Describe spatial variation with Semivariogram Lag Distance between pairs of points Lag Mean

More information

chapter 12 MORE MATRIX ALGEBRA 12.1 Systems of Linear Equations GOALS

chapter 12 MORE MATRIX ALGEBRA 12.1 Systems of Linear Equations GOALS chapter MORE MATRIX ALGEBRA GOALS In Chapter we studied matrix operations and the algebra of sets and logic. We also made note of the strong resemblance of matrix algebra to elementary algebra. The reader

More information

2.6 Two-dimensional continuous interpolation 3: Kriging - introduction to geostatistics. References - geostatistics. References geostatistics (cntd.

2.6 Two-dimensional continuous interpolation 3: Kriging - introduction to geostatistics. References - geostatistics. References geostatistics (cntd. .6 Two-dimensional continuous interpolation 3: Kriging - introduction to geostatistics Spline interpolation was originally developed or image processing. In GIS, it is mainly used in visualization o spatial

More information