
1JULY 2000 NEW ET AL. 2217 Representing Twentieth-Century Space±Time Climate Variability. Part II: Development of 1901±96 Monthly Grids of Terrestrial Surface Climate MARK NEW,* MIKE HULME, AND PHIL JONES Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich, United Kingdom (Manuscript received 21 July 1998, in ®nal form 19 August 1999) ABSTRACT The authors describe the construction of a 0.58 lat±long gridded dataset of monthly terrestrial surface climate for the period of 1901±96. The dataset comprises a suite of seven climate elements: precipitation, mean tem- perature, diurnal temperature range, wet-day frequency, vapor pressure, cloud cover, and ground frost frequency. The spatial coverage extends over all land areas, including oceanic islands but excluding Antarctica. Fields of monthly climate anomalies, relative to the 1961±90 mean, were interpolated from surface climate data. The anomaly grids were then combined with a 1961±90 mean monthly climatology (described in Part I) to arrive at grids of monthly climate over the 96-yr period. The primary variablesÐprecipitation, mean temperature, and diurnal temperature rangeÐwere interpolated directly from station observations. The resulting time series are compared with other coarser-resolution datasets of similar temporal extent. The remaining climatic elements, termed secondary variables, were interpolated from merged datasets comprising station observations and, in regions where there were no station data, synthetic data estimated using predictive relationships with the primary variables. These predictive relationships are described and evaluated. It is argued that this new dataset represents an advance over other products because (i) it has higher spatial resolution than other datasets of similar temporal extent, (ii) it has longer temporal coverage than other products of similar spatial resolution, (iii) it encompasses a more extensive suite of surface climate variables than available elsewhere, and (iv) the construction method ensures that strict temporal ®delity is maintained. The dataset should be of particular relevance to a number of applications in applied climatology, including large-scale biogeo- chemical and hydrological modeling, climate change scenario construction, evaluation of regional climate models, and comparison with satellite products. The dataset is available from the Climatic Research Unit and is currently being updated to 1998. 1. Introduction primary purposes, which are monitoring current climate (and its historic perspective), climate change detection, The description of the mean state and variability of and general circulation model (GCM) evaluation, do not recent climate is important for a number of purposes in necessarily require spatially continuous ®elds or higher global change research. These include monitoring and resolution. detecting climate change, climate model evaluation, cal- There has been a growing demand for datasets with ibration of or merging with satellite data, biogeochem- high spatial (e.g., 0.58 lat±long) and temporal (e.g., ical modeling, and construction of climate change sce- monthly or daily) resolution that are also continuous narios (New et al. 1999). Datasets of surface climate, over the space±time domain of interest. Potential ap- which describe variability in space and time (Hulme plications for such datasets include understanding the 1992; Jones 1994; Easterling et al. 1997), historically role of climate in biogeochemical cycling (Dai and Fung have had incomplete spatial coverage and have been of 1993; Cramer and Fischer 1996), climate change sce- coarse resolution ($2.58 lat±long). This is because their nario construction (Carter et al. 1994; Hulme et al. 1995) and high-resolution climate model evaluation (Chris- tensen et al. 1997). Yet there currently are few datasets that satisfy the requirement of high spatio±temporal res- * Current af®liation: School of Geography and the Environment, University of Oxford, Oxford, United Kingdom. olution. Notable exceptions are the monthly 1971±94 Global Precipitation Climatology Project (GPCP) da- taset (Rudolf et al. 1994; Xie and Arkin 1996; Xie et Corresponding author address: Dr. Mark New, School of Geog- al. 1996; Huffman et al. 1997); the monthly 1900±88, raphy and the Environment, Mans®eld Road, Oxford OX1 3TB, Unit- ed Kingdom. 2.58 lat±long precipitation dataset of Dai et al. (1997a, E-mail: [email protected] hereinafter DAI); and the 0.58 lat±long daily dataset q 2000 American Meteorological Society Unauthenticated | Downloaded 09/23/21 03:45 PM UTC 2218 JOURNAL OF CLIMATE VOLUME 13 being developed by Piper and Stewart (1996, hereinafter cient for the derivation of anomaly ®elds directly from PS). However, these products either cover relatively station data. We therefore used empirical relationships short periods (1970s±presentÐGPCP, PS), are limited to derive synthetic anomalies from the gridded anom- to precipitation (GPCP, PS, DAI) and maximum and alies of primary variables and merge these with station minimum temperature (PS), do not include an elevation anomalies of secondary variables over regions where dependence in their interpolation schemes (GPCP, PS, such data were available. The merged anomalies were DAI), or have a relatively coarse resolution (DAI). A then combined with the 1961±90 normal grids men- further limitation is that GPCP and PS interpolate di- tioned above, thereby standardizing the anomalies rectly from station time series: their methodology has against high-resolution observed data. This approach is to overcome dif®culties in interpolating monthly climate described in more detail in the second half of the paper, over complex terrain and they cannot make use of the along with an evaluation of the various empirical re- more extensive network of station climatological nor- lationships. We end the paper with a discussion of the mals to de®ne a mean climatology (see below). merits and limitations of this new dataset and our con- In this paper, we describe the construction of a new clusions. dataset of monthly surface climate over global land ar- eas, excluding Antarctica, for the period of 1901±96. The dataset is gridded at 0.58 lat±long resolution and 2. Primary variables comprises a suite of seven variables, namely, precipi- a. Datasets tation, wet-day frequency, mean temperature, diurnal temperature range, vapor pressure, cloud cover, and Three global station datasets compiled by the Cli- ground frost frequency. matic Research Unit (CRU) form the basis for the con- In constructing the monthly grids, we used an ``anom- struction of the gridded anomalies of primary variables. aly'' approach, which attempts to maximize available The precipitation (Eischeid et al. 1991; Hulme 1994, station data in space and time (New et al. 1999). In this updated) and mean temperature (Jones 1994, updated) technique, grids of monthly anomalies relative to a stan- station data have been compiled by the CRU over the dard normal period (in our case, 1961±90) were ®rst last 20 yr. The diurnal temperature range dataset is based derived. The anomaly grids were then combined with a on the Global Historical Climatology Network (GHCN) high-resolution mean monthly climatology to arrive at maximum and minimum temperature data (Easterling et ®elds of estimated monthly surface climate. We used al. 1997) but has been updated for more recent years the 0.58 lat±long 1961±90 climatology described in a by CRU and enhanced with additional station data ob- companion paper (New et al. 1999) for this purpose. tained by the CRU and the U.K. Meteorological Of®ce The advantage of this approach is that the number of (Horton 1995, updated). The original data have been archived and easily obtainable station normals is far subjected to comprehensive quality control over the greater than that of station time series, particularly as years, as described by the above authors. Updates for one goes back in time. Using as many stations as pos- more recent years and additional station data collated sible to generate the mean ®elds together with an explicit by the CRU have also been checked for homogeneity treatment of elevation dependency maximizes the rep- and outliers. resentation of spatial variability in mean climate. The CRU precipitation data have not been corrected Monthly anomalies, on the other hand, tend to be more for gauge biases, the most signi®cant of which is un- a function of large-scale circulation patterns and rela- dercatch of solid precipitation in colder areas. Under- tively independent of physiographic control. Therefore, catch also varies with gauge type, so periodic instrument a comparatively less extensive network can be used to changes can therefore result in inhomogeneties in the describe the month-to-month departures from the mean records. The correction of individual records requires climate. detailed local meteorological and station metainforma- We have divided the seven climatic elements into two tion, which are not readily available. groups, primary and secondary variables. The former, The station networks for all three variables exhibit a comprising precipitation, mean temperature, and diurnal gradual increase in the total number of stations from temperature range, was considered to have suf®cient 1901 to about 1980, after which the numbers decline station coverage to attempt the derivation of grids di- (Figs. 1±3). The recent reduction in station numbers is rectly from station anomalies for the entire period of primarily
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages22 Page
-
File Size-