Annals of the Association of American Geographers Changes ... · Surface Precipitation Over the Twentieth and Early Twenty-First Century', Annals of the Association of American Geographers,,

PLEASE SCROLL DOWN FOR ARTICLE

This article was downloaded by: [Robeson, Scott M.]On: 12 August 2010Access details: Access Details: [subscription number 925742644]Publisher RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Annals of the Association of American GeographersPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t788352614

Changes in Annual Land-Surface Precipitation Over the Twentieth andEarly Twenty-First CenturyElsa Nickla; Cort J. Willmotta; Kenji Matsuuraa; Scott M. Robesonb

a Center for Climatic Research, Department of Geography, University of Delaware, b Department ofGeography, Indiana University,

First published on: 12 August 2010

To cite this Article Nickl, Elsa , Willmott, Cort J. , Matsuura, Kenji and Robeson, Scott M.(2010) 'Changes in Annual Land-Surface Precipitation Over the Twentieth and Early Twenty-First Century', Annals of the Association of AmericanGeographers,, First published on: 12 August 2010 (iFirst)To link to this Article: DOI: 10.1080/00045608.2010.500241URL: http://dx.doi.org/10.1080/00045608.2010.500241

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial orsystematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply ordistribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae and drug dosesshould be independently verified with primary sources. The publisher shall not be liable for any loss,actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directlyor indirectly in connection with or arising out of the use of this material.

http://www.informaworld.com/smpp/title~content=t788352614

http://dx.doi.org/10.1080/00045608.2010.500241

http://www.informaworld.com/terms-and-conditions-of-access.pdf

Changes in Annual Land-Surface Precipitation Overthe Twentieth and Early Twenty-First Century

Elsa Nickl,∗ Cort J. Willmott,∗ Kenji Matsuura,∗ and Scott M. Robeson†

∗Center for Climatic Research, Department of Geography, University of Delaware†Department of Geography, Indiana University

Time trends in annual land-surface precipitation during the twentieth and early twenty-first centuries and theirspatial patterns are estimated from gridded (at a 0.5◦ × 0.5◦ spatial resolution) rain-gauge-based precipitation datasets available from the Climatic Research Unit (CRU), the Global Precipitation Climatology Centre (GPCC),and at the University of Delaware (UDel). Our analyses of these precipitation data sets make use of spatiallyweighted (geographic) percentiles as well as of join-point and simple linear regression. A consistent increase inannual land-surface-average precipitation (of approximately 0.2 and 0.5 mm/year) occurred during the first half ofthe twentieth century. This increase was followed by nearly a half-century (approximately forty-four years, from1949 through 1993) of decreases in annual land-surface-average precipitation (on the order of 0.3 to 0.6 mm/year).Trends, once again, reversed themselves in the early 1990s and increased (at rates of approximately 0.75 to2.1 mm/year) over the decade from 1992 through 2002. Maps of precipitation change during these alternatingperiods of increasing and decreasing precipitation show considerable spatial variability. Key Words: climatechange, precipitation, rain gauge data.

Las tendencias temporales en precipitacion anual sobre el terreno durante el siglo XX y comienzos del XI ysus patrones espaciales estan calculados a partir de conjuntos de datos de precipitacion basados en registrospluviometricos distribuidos en cuadrıcula (a 0.5˚ x 0.5˚ de resolucion espacial), disponibles en la Unidad deInvestigacion Climatica (CRU, sigla en ingles), el Centro de Climatologıa para la Precipitacion Global (GPCC)y la Universidad de Delaware (UDel). Nuestros analisis de estos conjuntos de datos de precipitacion utilizanpercentiles espacialmente ponderados (geograficos), lo mismo que regresiones join-point y linear simple. Sepresento un consistente incremento de la precipitacion promedio anual sobre el terreno (de aproximadamente0.2 y 0.5 mm/ano) durante la primera mitad del siglo XX. A tal incremento siguio un perıodo de cerca de mediosiglo (aproximadamente cuarenta y cuatro anos, de 1949 a 1993) de disminucion de la precipitacion promedioanual sobre el terreno (del orden de 0.3 a 0.6 mm/ano). Nuevamente las tendencias se reversaron a sı mismas aprincipios de los 1990 y se incrementaron (a tasas de aproximadamente 0.75 a 2.1 mm/ano) en la decada quese extendio de 1992 a 2002. Los mapas del cambio de la precipitacion durante estos perıodos alternantes deincremento y disminucion de la precipitacion muestran considerable variabilidad espacial. Palabras clave: cambioclimatico, precipitacion, datos pluviometricos.

Abetter understanding of the spatial andtemporal variability of land-surface precipi-tation is indispensable for advancing climate

change research, as well as for assessing the potentialimpacts of climate change on water resources. Inaddition to climate-model estimates of precipitation,measurement-based fields of precipitation and precipi-tation change are critical for evaluating the vicissitudesof Earth’s climate. Our current measurement-based

knowledge of land-surface precipitation variabilityand change over the last 100-plus years, however, isuncertain, as is evident in the differences betweenavailable gridded land-surface precipitation data sets.The Intergovernmental Panel on Climate Change(IPCC 2007), for example, reports substantial discrep-ancies among trend estimates derived from differentdata sets. According to the IPCC (2007) report, trendsvary by regions and over time, although land-surface

Annals of the Association of American Geographers, 100(4) 2010, pp. 1–11 C© 2010 by Association of American GeographersInitial submission, July 2009; revised submission, October 2009; final acceptance, January 2010

Published by Taylor & Francis, LLC.

Downloaded By: [Robeson, Scott M.] At: 16:38 12 August 2010

2 Nickl et al.

precipitation has generally increased north of 30◦Nover the past century and decreased over much ofthe tropics since the 1970s. Our purpose within thisarticle is to explore, describe, and compare the patternsand historical trends in land-surface precipitationusing three available high-resolution rain-gauge-basedprecipitation data sets.

Available Land-Surface Precipitation DataSets

Estimated land-surface precipitation fields havetended to be of relatively coarse spatial resolution(≥ 2.5◦), although efforts to develop higher resolutionspatial and temporal (e.g., monthly or daily) data setsare increasing. Our analyses of the patterns and trendsin annual land-surface precipitation make use of threerecently available higher resolution land-surface pre-cipitation data sets. Each of these data sets is derivedprimarily from in situ (rain-gauge) observations, has amonthly time step, and is gridded at (spatially inter-polated to) a relatively high (0.5◦) spatial resolution.The data sets are the Climatic Research Unit (CRU)archive (Mitchell and Jones 2005), the Global Precip-itation Climatology Centre (GPCC) archive (Rudolfand Schneider 2005), and the recent University ofDelaware (UDel) data set (Matsuura and Willmott2009). Many of the same observational (rain-gauge)records inform all three of these data sets; however,each data set is not based on exactly the same setof rain-gauge records. It also should be noted thatthe CRU and GPCC archives do not contain esti-mates of precipitation over Antarctica; in turn, Antarc-tica is not included in our analyses of “land-surface”precipitation.

Each of the three precipitation data sets is based on adifferent method of spatial interpolation from the rain-gauge station locations to the nodes of the 0.5◦ grid. TheCRU database is based on an angular-distance weighted(ADW) interpolation method (which bears conceptualsimilarity to the Shepard 1968 and Willmott, Rowe, andPhilpot 1985 approaches). For each grid-node estimate,the ADW method weights each of the eight rain-gaugestation precipitation observations that are nearest to thegrid node by taking into account the distance from thegrid node (using a correlation-decay distance [CDD])and the directional (angular) isolation of each station.Interpolated fields are forced to a climatological meanvalue at grid points where there is no station within theCDD (New, Hulme, and Jones 2000; Mitchell and Jones

2005). As a consequence of this, estimated time seriesover some areas can be invariant for a number of years.

The GPCC data set (Rudolf and Schneider 2005)was spatially interpolated with the “SPHEREMAP” in-terpolation tool, which was developed at the Universityof Delaware. It is a spherical adaptation of Shepard’s(1968) empirical weighting scheme, which takes intoaccount the spherical distances from the nearby rain-gauge locations to the grid node (for a limited numberof the nearest stations), the directional distribution ofnearby rain gauges relative to the grid node (to avoidoverweighting of clusters of stations), and spatial gradi-ents within the rain-gauge data field (Willmott, Rowe,and Philpot 1985).

The UDel data set also is based on the Willmott,Rowe, and Philpot (1985) traditional-interpolationprocedure as well as on Climatologically Aided Inter-polation (CAI; Willmott and Robeson 1995; Matsuuraand Willmott 2009), which employs a spatially high-resolution climatology to obtain a monthly precipita-tion difference at each station. These station differencesthen are interpolated to obtain a gridded field using aversion of Shepard’s algorithm (Willmott, Rowe, andPhilpot 1985); finally, each gridded monthly differencefield is added back onto the corresponding monthly cli-matology field to obtain monthly land-surface precip-itation (Matsuura and Willmott 2009). These griddedmonthly values then were integrated over the year toobtain an annual average for each year of interest.

The gridded monthly precipitation values containedwithin the three data sets analyzed here are not ad-justed up for possible rain-gauge undercatch (Legatesand Willmott 1990); therefore, our (and most other)estimates of annual precipitation are likely to be a littlelower than actual precipitation. It is important to note,however, that this should have little or no effect onour estimates of time trends, as long as there was nomeaningful time trend in rain-gauge undercatch.

Geographic Analyses of AnnualLand-Surface Precipitation

Land-surface precipitation change since the be-ginning of the twentieth century is evaluated usingspatially weighted (geographic) percentiles (Willmott,Robeson, and Matsuura 2007) and simple linear andjoin-point regression (Rawlins et al. 2006)—the latteris a type of change-point regression (Draper and Smith1998)—of annual land-surface-average precipitation.An overall assessment of the longer term trends and


Changes in Annual Land-Surface Precipitation Over the Twentieth and Early Twenty-First Century 3

variability apparent within the entire 100-plus years ofrecord—contained in the CRU, GPCC, and UDel datasets—is made first. More detailed analyses of each ofthree major subperiods follow. The subperiods are thefirst half of the twentieth century (1902–1949 herein,for reasons explained later), 1950 through 1993, and1993 through 2002. The year 1993 is a join-pointestimate of the boundary between a long period ofdrying and more recent increases in land-surfaceprecipitation (discussed later). Maps of change ratesfor each subperiod are presented and evaluated.

Because annual land-surface precipitation is intrin-sically a geographic or spatial variable, we take explicitaccount of the sizes (areas) of the grid cells when per-forming our spatial analyses (Willmott, Robeson, andMatsuura 2007). As a consequence, each of our spatialpercentiles has a fundamentally geographic interpreta-tion. Our 75th spatial percentile, for instance, boundsthat 25 percent of the land surface that contains an-nual precipitation values that are greater than the valueof the 75th spatial percentile. Our join-point regres-sion of land-surface-average precipitation is used toidentify those year(s) when major precipitation changeoccurred; that is, since the beginning of the secondhalf of the twentieth century. Join-point regressionfinds one or more “optimal” join-points in a time se-ries by minimizing the sum-of-squared residuals of allpossible join-point regressions. It helped us locate a

year (1993) at or near the transition from more thanforty years of drying to a decade or more of increasingprecipitation.

Land-Surface Precipitation over theTwentieth and Early Twenty-FirstCenturies

Using the CRU, GPCC, and UDel data sets, timeseries of the annual spatial means as well as of the 5th,25th, 75th, and 95th spatial percentiles are estimatedand plotted for each data set (Figure 1). All of the datasets suggest that long-term land-surface average (rain-gauge-caught) precipitation is about 716 mm/year. Thethree data set average 5th, 25th, 75th, and 95th spatialpercentiles are approximately 58, 234, 936, and 2,264mm/year, which illustrates the right (positive) skew thattypifies precipitation distributions. Over the entire pe-riod of record, but especially during the first half of thetwentieth century, the three traces of the spatial meansand spatial percentiles are remarkably similar, with a fewexceptions. Prior to the 1930s, for example, the lower25th spatial percentile of the GPCC trace is noticeablybelow the CRU and UDel traces, whereas during thesecond half of the twentieth century, the UDel tracebecomes the lowest estimate of the higher spatial per-centiles and the spatial mean. Variability, both within

Figure 1. Time series (1901–2008)of annual land-surface precipitation(mm) estimated from the Climatic Re-search Unit (CRU), Global Precipita-tion Climatology Centre (GPCC), andUniversity of Delaware (UDel) datasets. Annual spatial means as well asthe 5th, 25th, 75th, and 95th spa-tial percentiles are plotted for eachof the three data sets. Open trian-gles are used to represent CRU data,open squares indicate values derivedfrom GPCC data, and blackened circlesshow estimates made from the UDeldata set. There are five clusters of threecurves each. The top set of three curvescontains the estimates of the 95th spa-tial percentile of annual precipitation,the next set (moving down the graph)represents the 75th spatial percentiles,then comes the spatial means, then the25th spatial percentiles, and the last(bottom) set indicates the 5th spatialpercentiles.


4 Nickl et al.

Figure 2. Three simple linear regressions of annual land-surface-average precipitation (over the period 1902–1949) are estimatedfrom the Climatic Research Unit (CRU), Global Precipitation Cli-matology Centre (GPCC), and University of Delaware (UDel) datasets and plotted. The values of annual land-surface-average precipi-tation over the period 1902 to 1949 also are plotted. Open trianglesare used to represent CRU data, open squares indicate values de-rived from GPCC data, and blackened circles show estimates madefrom the UDel data set.

and among the three traces, tends to increase after the1970s. This appears to arise from increasing variabilityin the maxima, as suggested in the traces of the 95thspatial percentile.

There are several reasons why the UDel annual av-erage precipitation measures fall increasingly below thecomparable CRU and GPCC measures, as the presentis approached (Figure 1). The UDel data set containsrecords from the Global Surface Summary of the Day(GSOD) archive, which improves the station-networkcoverage of the land surface, especially during the re-cent past and over the more harsh (drier and ruggedregions) of the land surface. It follows that improvedrepresentation of drier regions would lessen (relativelyspeaking) the UDel estimates of land-surface-averageprecipitation. It also is true, however, that the GSODarchive contains a variety of unrealistic extreme val-ues, including unbelievably long strings of zeros anda limited number of unusually high values. AlthoughMatsuura and Willmott (2009) attempted to filter outonly the truly unrealistic daily and monthly precip-itation values from the GSOD station records, anyerroneous zero value that was missed or an incor-rectly removed maximum would tend to produce anunderestimate.

Land-Surface Precipitation over the First Half of theTwentieth Century

It is clear that there was a general increase in land-surface-average precipitation during the first half of thetwentieth century (1902–1949; Figure 2). Ignoring theestimated precipitation values for the year 1901, owingto probable station-network biases (Willmott, Robeson,and Feddema 1994), all three traces indicate an increasein the spatial mean, at estimated rates of between 0.2and 0.5 mm/year. The spatial distribution of changeover this period, however, is quite variable (Figure 3).Although there are many areas over which relativelysmall decreases in annual precipitation can be seen (e.g.,over large areas of North America, North Africa, andAustralia), these are outweighed by increases elsewhere.Large increases are evident over portions of the Amazonbasin (especially within the GPCC and UDel fields),over the Maritime Continent (especially within theCRU and GPCC data sets), and, to a lesser extent, overnorthern India (within all three data sets). Spatiallyextensive but relatively small increases also appear over



Figure 3. Spatial distribution of timetrends (mm/year) in annual land-surface precipitation (1902–1949) ap-parent within the Climatic ResearchUnit (CRU), Global Precipitation Cli-matology Centre (GPCC), and Uni-versity of Delaware (UDel) data sets.


6 Nickl et al.

Figure 4. Three join-point and three sim-ple linear regressions of annual land-surface-average precipitation (over theperiod 1950–2002) are estimated fromthe Climatic Research Unit (CRU),Global Precipitation Climatology Cen-tre (GPCC), and University of Delware(UDel) data sets and plotted. The valuesof annual land-surface-average precipita-tion over the period 1950 through 2008are also plotted. Open triangles are usedto represent CRU data, open squares indi-cate values derived from GPCC data, andblackened circles show estimates madefrom the UDel data set.

much of the Arctic land surface and subtropical Africa.Annual land-surface-average precipitation during thefirst half of the twentieth century and into the 1950swas relatively stable (exhibiting limited variability) butincreasing consistently.

Land-Surface Precipitation Since the Middle of theTwentieth Century

Graphs of the spatial means of annual land-surfaceprecipitation from 1949 forward (especially in Figure 4)





8 Nickl et al.




indicate a dramatic decrease in land-surface precipita-tion as well as gradually increasing divergence amongthe three data sets. It is not entirely clear when thedrying began but, on inspection (Figure 4), probablyduring the mid- to late 1950s. It is less clear exactlywhen it ended; therefore, we use join-point regressionto help us make a determination. A single-point, join-point linear regression was fit to each of the three tracesof the annual land-surface means over the period from1949 through 2002 (Figure 4), and each suggests thatthe decrease ended during the early 1990s, most likelyin 1993. Our regressions were applied only through2002 because that is the last year for which CRU datawere available. For comparison purposes, we also esti-mate and plot (Figure 4) a simple linear regression overthe entire period from 1949 through 2002. It describesa general drying over the entire fifty-three-year period.

The reduction in land-surface-average precipitationfrom 1949 through 1993 was substantial, on the or-der of 0.3 to 0.6 mm/year or about 14 to 26 mm overthe entire forty-four-year period. The steepest declinewas estimated from the UDel data (Figure 4). To as-sess the spatial distribution of this land-surface drying,we estimate the linear time trend over the period from1950 through 1993 at each land-surface grid node, andwe map the rates of change (mm/year) obtained fromeach of the three data sets (Figure 5). Drying dominatesAfrica, especially over sub-Saharan Africa, as well asover the Maritime Continent, Southeast Asia, and thenorthwest Amazon basin. The strongest drying oversub-Saharan Africa and the northwest Amazon basin isestimated from the CRU data set. Areas wherein slightincreases appear include North America, southeasternSouth America, western Australia, eastern Europe, andcentral Asia, although the signal over eastern Europeand central Asia is weaker within the UDel data set.

Reasons for the persistent, forty-four-year dryingthat occurred over extensive reaches of Earth’s landsurface, and especially over the lower latitudes, are notcompletely understood, but several contributing fac-tors have been identified for the years after 1949 (e.g.,see Marengo 2004; Hoerling et al. 2006). Marengo(2004), for example, associated circulation features typ-ical of strong El Nino years with drier conditions overthe northern Amazon basin region, while Aldrian andDjamil (2008) attributed decreases in precipitation overparts of the Maritime Continent (during the periodfrom 1955–2005) to a waning of monsoonal dominance.Droughts also tend to develop over the Maritime Conti-nent during strong El Nino years. The drying over muchof Africa has been tied to relatively warm sea-surface

temperatures atop the tropical oceans, including theIndian Ocean (Hoerling et al. 2006). Our sense is thatthese and perhaps other factors, such as deforestation—especially within the Amazon region—and global dim-ming, contributed to the drying.

Land-surface-average precipitation trends then re-versed themselves, ostensibly beginning in 1993, andmarked increases ensued over the decade from 1992through 2002. Our second set of join-point regressionlines (fit to the 1993–2002 land-surface-average precip-itation values but constrained by the 1993 join point)suggest increases that are even more dramatic thanthe forty-four years of drying that occurred previously(Figure 4). These increases are on the order of 0.75 to2.1 mm/year but, of course, the reduced length of record(ten years) and nontrivial year-to-year variability arguefor cautious interpretation of these increases.

Once again, to assess the spatial distribution of theseincreases in yearly average precipitation, we estimatethe linear time trend over the period from 1993 through2002 at each land-surface grid node, and we map therates of change (mm/year; Figure 6). All three data setsshow significant drying over much of North Americaand south Asia as well as over much of North Africaand parts of South America. In contrast, substantialincreases in annual average precipitation can be seenover parts of South America (especially over southernSouth America), southern Africa, northern Australia,the Maritime Continent, and parts of Southeast Asia.Regional dissimilarities in these recent trends emergefrom the three data sets (compare the three geographiesof trend estimates for South America, for example)and they underscore the need for more research intothe finer scale nature of land-surface precipitationvariability and change. Nonetheless, judging from themost recent values of estimated land-surface-averageprecipitation (2003–2008) obtained from the GPCCand UDel data (Figures 1 and 4), it seems likely thatland-surface-average precipitation has continued toincrease.

Summary and Conclusions

Our assessment of the changing patterns and trendsin annual land-surface precipitation is based on threerecently available monthly land-surface precipitation(rain-gauge-based) data sets. The data sets were as-sembled at the CRU, the GPCC, and UDel. Each ofthese data sets was gridded at (spatially interpolatedto) a 0.5◦ × 0.5◦ spatial resolution and we temporally


10 Nickl et al.

integrated the monthly grid-node values to obtain theannual grid-node totals.

Our analyses of land-surface precipitation changemade use of spatially weighted (geographic) percentiles,simple linear regression, and join-point regression. Allof the precipitation data suggest that long-term averageland-surface precipitation (rain-gauge-caught) is about716 mm/year. There appears to have been a modestbut consistent increase in land-surface-average precip-itation (between 0.2 and 0.5 mm/year, approximately)during the first half of the twentieth century. Thiswas followed by nontrivial reductions in annual land-surface-average precipitation (on the order of 0.3 to 0.6mm/year), which were estimated over the period from1949 through 1993. Trends in land-surface-average pre-cipitation then reversed themselves again in the early1990s and increased (at rates of approximately 0.75 to2.1 mm/year) over the decade from 1992 through 2002.Recent data also suggest that land-surface-average pre-cipitation has continued to increase until the presentday.

The spatial distribution of the time trends in pre-cipitation during these alternating periods of increas-ing and decreasing precipitation was explored withmaps of the estimated trends, which showed consid-erable spatial variability. Over the ten years from 1992through 2002, for instance, there was marked dryingover much of North America and south Asia as wellas over much of North Africa and parts of SouthAmerica; at the same time, sizable increases in an-nual average precipitation occurred over parts of SouthAmerica, southern Africa, northern Australia, the Mar-itime Continent, and Southeast Asia. A number of re-gional differences in trends estimated from the threedata sets were apparent as well, especially in recentyears.

Our findings argue for additional research into as-sessing the regional differences among the three sets ofestimated trends that emerged from the three data sets;that is, into the finer scale spatial variability of land-surface precipitation. There also is a need to resolve theintra-annual variability of precipitation change. Ourhope is that in situ–based precipitation fields, such asthese, also will be used to evaluate and improve climate-model estimates of precipitation.

Acknowledgments

Portions of this work were made possible by NASAGrant NNG06GB54G to the Institute of Global Envi-

ronment and Society and we are most grateful for thissupport.

References

Aldrian, E., and Y. S. Djamil. 2008. Spatio-temporal climaticchange of rainfall in East Java Indonesia. InternationalJournal of Climatology 28:435–48.

Draper, N. R., and H. Smith. 1998. Applied regression analysis.3rd ed. New York: Wiley.

Hoerling, M., J. Hurrell, J. Eischeid, and A. Phillips. 2006.Detection and attribution of twentieth-century northernand southern African rainfall change. Journal of Climate19 (16): 3989–4008.

Intergovernmental Panel on Climate Change (IPCC). 2007.Climate change 2007: The physical science basis. Contribu-tion of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change, ed. S.Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis,K. B. Averyt, M. Tignor, and H. L. Miller, 1–996. NewYork: Cambridge University Press.

Legates, D. R., and C. J. Willmott. 1990. Mean seasonal andspatial variability in gauge-corrected, global precipita-tion. International Journal of Climatology 10:111–27.

Marengo, J. A. 2004. Interdecadal variability and trends ofrainfall across the Amazon basin. Theoretical and AppliedClimatology 78:79–96.

Matsuura, K., and C. J. Willmott. 2009. Terrestrial pre-cipitation: 1900–2008 gridded monthly time series (Ver-sion 2.01). Newark: Center for Climatic Research,Department of Geography, University of Delaware.http://climate.geog.udel.edu/∼climate/ (last accessed 15July 2009).

Mitchell, T. D., and P. D. Jones. 2005. An improved methodof constructing a database of monthly climate observa-tions and associated high-resolution grids. InternationalJournal of Climatology 25:693–712.

New, M., M. Hulme, and P. Jones. 2000. Representingtwentieth-century space-time climate variability: Part II.Development of 1901–96 monthly grids of terrestrial sur-face climate. Journal of Climate 13: 2217–38.

Rawlins, M. A., C. J. Willmott, A. Shiklomanov, E. Lin-der, S. Frolking, R. B. Lammers, and C. J. Vorosmarty.2006. Evaluation of trends in derived snowfall and rain-fall across Eurasia and linkages with discharge to theArctic Ocean. Geophysical Research Letters 33:L07403.

Rudolf, B., and U. Schneider. 2005. Calculation of grid-ded precipitation data for the global land-surface us-ing in-situ gauge observations. In Proceedings of the 2ndWorkshop of the International Precipitation Working Group,231–47. Germany: Global Precipitation ClimatologyCentre.

Shepard, D. 1968. A two-dimensional interpolation functionfor irregularly-spaced data. In Proceedings, 1968 ACMNational Conference, ed. R. B. Blue and A. M. Rosen-berg, 517–23. New York: Association for ComputingMachinery.

Willmott, C. J., and S. M. Robeson. 1995. Climatologi-cally aided interpolation (CAI) of terrestrial air tem-perature. International Journal of Climatology 15 (2):221–29.



Willmott, C. J., S. M. Robeson, and J. J. Feddema. 1994.Estimating continental and terrestrial precipitation av-erages from rain-gauge networks. International Journal ofClimatology 14 (4): 403–14.

Willmott, C. J., S. M. Robeson, and K. Matsuura. 2007. Ge-ographic box plots. Physical Geography 28 (4): 331–44.

Willmott, C. J., C. M. Rowe, and W. D. Philpot. 1985.Small-scale climate maps: A sensitivity analysis of somecommon assumptions associated with grid-point interpo-lation and contouring. The American Cartographer 12:5–16.

Correspondence: Department of Geography, University of Delaware, Newark, DE 19716, e-mail: [email protected] (Nickl); [email protected](Willmott); [email protected] (Matsuura); Department of Geography, Indiana University, Bloomington, IN 47405; e-mail: [email protected] (Robeson).


Annals of the Association of American Geographers Changes ... · Surface Precipitation Over the Twentieth and Early Twenty-First Century', Annals of the Association of American Geographers,,

Documents