1. introduction€¦ · introduction with the rapid and continuous economic growth in china, market...

* Corresponding author address: Christopher Daly, Director, Spatial Climate Analysis Service, Dept. ofGeosciences, Oregon State University, 326 Strand Agricultural Hall, Corvallis, OR 97331; e-mail:[email protected]

JP1.23 DEVELOPMENT OF NEW CLIMATE AND PLANTADAPTATION MAPS FOR CHINA

Christopher Daly *, Wayne Gibson, David Hannaway, and George TaylorOregon State University, Corvallis, OR 97331, USA

1. INTRODUCTION

With the rapid and continuous economicgrowth in China, market demands for improvedforage-livestock systems, urban beautification,and improved environmental protection -including soil conservation and erosion control -have greatly expanded in the last decade.Specifically, the increased demands for animalproducts, beautiful turf for roadsides, lawns, andgolf courses, and reduced soil erosion in theYellow (Huang) and Yangtze (Chang Jiang)River watersheds have contributed to increasedmarket demands for high quality grass seeds.

In 1992 the Oregon Seed Council, aconsortium of over 50 grass seed companies,partnered with Oregon State University toinitiate a USDA Market Access Program (MAP)project for developing the China market forOregon-grown grass seed. Other cooperatorsinclude the Oregon Department of Agriculture,the Chinese Academy of Agricultural Sciences,the Chinese National Meteorological Center,China Agricultural University, NanjingAgricultural University, Wuhan TechnicalUniversity of Surveying and Mapping, JiangsuAcademy of Agricultural Sciences, the YiChangMunicipality, and the Three GorgesDevelopment Corporation.

Effective marketing of our high quality US-grown seeds requires that we be able to identifyall of the areas suitable for using these grasses.Until now that has been impossible on a widescale. The USDA MAP project has conductedfield-based trials in China. However, these havebeen few in number and located mostly in thepopulated eastern lowlands, because of thedifficulty in finding suitable cooperators,establishing and maintaining test plots, andcollecting accurate data on species performanceover a period of years.

Fortunately, GIS technology, combined with

climate mapping expertise, makes it possible tospatially verify and extrapolate the results ofthese field trials to all of China. Climatemapping, in combination with spatial soils dataand climatic tolerances of grass species, can beused to produce very detailed maps of speciesadaptation that can be used to accurately identifysuitable growing areas for effective marketing.

The objectives of the work described here are to:• Obtain and quality-check observed climate

data from Chinese authorities.• Prepare detailed draft maps of mean

monthly minimum/maximum temperatureand precipitation for the People’s Republicof China.

• Use the climate maps and expert estimatesof species climatic tolerances to prepareadaptation maps for several importantOregon-grown grasses.

• Subject draft climate and adaptation maps topeer review by Chinese climatologists,obtain supplementary climate data, andrevise maps as necessary.

This paper describes the methods used toaccomplish the first three objectives for easternChina: obtain climate data; create first-draftclimate maps; and develop grass adaptationmaps.

2. METHODS

2.1 Collection and quality control ofobservational data

In 1998, contact was established with theChinese Climate Data Center (CDC), which is apart of the Chinese National MeteorologicalCenter in Beijing. The CDC is responsible forthe collection, archive, and dissemination ofclimate data for the national climatic network,

Proc., 12th AMS Conf. on Applied Climatology, Amer. Meterological Soc., May 8-11, 2000

which consists of approximately 700 stations. Inexchange for seminars at CDC by Daly andHannaway, and a three-week training course fortwo of their scientists at Spatial ClimateAnalysis Service offices at Oregon StateUniversity, the CDC provided us with meanmonthly minimum and maximum temperatureand precipitation for 679 stations. Thesestations have the longest and most completerecords available.

Methods used by the CDC to quality-controlthe station data were unknown. Therefore, wehad to assume that errors existed in the data.Two types of checks were made: metadata andmonthly data. The location and elevationmetadata were checked by plotting each stationon a 30-second digital elevation model (DEM)from the ETOPO-30 global elevation dataset. Ifthe station elevation differed substantially fromthe elevation of DEM pixels in the immediatevicinity, it was assumed that either the DEM wasincorrect or of insufficient resolution, or thereported station location or elevation werefaulty. Many DEM-station elevationdiscrepancies were found. In some regions, suchas the deeply dissected terrain of thesoutheastern Tibetan Plateau, the limitedprecision of the station location (nearest 1 arc-minute) and the inability of the 30-sec DEM toresolve locations at the bottoms of deep canyonswere the culprits. An atlas was used in attemptsto locate the place names and corroborate stationlocations. Stations that could not be found on anatlas and had clearly unreasonable elevationswere brought to the attention of CDC personneland attempts made to resolve or correct thediscrepancies.

As an additional check, the station metadatawere compared to those for the same stations inthe Global Historical Climate Network, preparedand maintained by the US National ClimaticData Center (NCDC). After lengthy discussionswith NCDC, the most common source ofmetadata discrepancies was found to be stationmoves, which caused the GHCN metadata,recorded some time in the past, to differ fromthe Chinese metadata, which reflects only thecurrent location of the station. No stationhistories were available from CDC from whichto track station moves over time.

The mean monthly station data were

checked for reasonableness by applyingASSAY, a version of PRISM (see below) thatperforms jackknife cross-validation on eachstation in the dataset. Stations with largeobservation-prediction errors were flagged. ThePRISM Graphical User Interface was used todisplay climate-elevation scatterplots in thevicinity of the questionable stations to determineif the station “fit in” with the others, or was anobvious outlier. Outliers were removed from thedataset. In general, few monthly outliers werefound, giving us confidence that the Chineseclimate data were of good quality.

2.2 Preparation of Climate Maps

Spatial modeling of the climate data wasperformed at 2.5-minute (~ 4-km) resolution. A2.5-minute DEM was derived from the ETOPO-30 DEM series. GIS was used to preparesupplementary grids used in the interpolation.These included a coastal proximity grid,estimates of wintertime temperature inversionheights, and an effective terrain grid.

The PRISM (Parameter-elevationRegressions on Independent Slopes Model)knowledge-based system (Daly et al., 1992,1994, 1997, in review, Daly and Johnson 1999)was used to grid precipitation and minimum andmaximum temperature over China. PRISMtechnology is being used in several climaticmapping efforts in the US, including a 103-yearclimate time series (Daly et al., 1999), a majorprecipitation mapping effort for NRCS (USDA-NRCS, 1998), an updated climate atlas for theNCDC, and others (e.g., Johnson et al., in press).PRISM uses point data, a DEM, and otherspatial data sets to generate estimates of climaticelements that are gridded and GIS-compatible.

The strong variation of climate withelevation is the main premise underlying themodel formulation. PRISM adopts theassumption that for a localized region, elevationis the most important factor in the distribution oftemperature and precipitation. Observationsfrom many parts of the world show thealtitudinal variations of temperature andprecipitation to approximate a linear form.Available station data often do not span thecomplete range of elevations in an area,


especially in mountainous regions. Therefore,vertical extrapolation is required. This isaccomplished in PRISM at each DEM grid cell(termed the target grid cell) through a moving-window simple linear climate-elevationregression. This regression function serves asthe main predictive equation in the model.

Upon entering the regression function, eachstation is assigned a weight that is based onseveral factors. The combined weight W of astation is a function of the following:

W = f { Wd , Wz , Wc , Wl , Wf , Wp , We }

where Wd , Wz , Wc , Wl , Wf , Wp and We are thedistance, elevation, cluster, vertical layer,topographic facet, coastal proximity, andeffective terrain weights, respectively. A stationis down-weighted when it is relatively distant orat a much different elevation than the target gridcell, or when it is clustered with other stations(which leads to over-representation). Verticallayer weighting is used when the atmosphere islayered, such as in temperature inversions orlow-level moisture conditions. Topographicfacet weighting is used to model rain shadowsand other abrupt climatic shifts caused bytopographic barriers. Coastal proximityweighting allows the reproduction of sharply-defined coastal strips. Effective terrainweighting accounts for the varying ability ofdifferent terrain features to induce orographicprecipitation.

PRISM was parameterized to produce to themost physically realistic maps by starting withthe settings that gave the lowest cross-validationerror in ASSAY, then modifying these settingsas experience with China’s complex climatepatterns increased over time. Experienced wasgained by consulting publications on Chineseclimate, using the station data themselves (viathe PRISM graphical user interface), and byusing knowledge from other regions, asapplicable. General expectations of climaticpatterns in China were formed, and served as aguide for the evaluation of the PRISM modelingruns. Peer review of the draft maps by Chineseclimatologists will provide another much neededlevel of evaluation.2.3 Preparation of Species Adaptation Maps

First drafts of species adaptation maps wereprepared for three important Oregon-growncool-season grasses: tall fescue, orchardgrass,and perennial ryegrass. These first mapsaccounted only for climatic tolerances and didnot include soils considerations. The climaticconstraints were also simple as a starting point.Each species was assigned approximate valuesof mean January minimum temperature, meanJuly maximum temperature, and mean annualprecipitation that corresponded with well-adapted, moderately adapted, marginallyadapted, and not adapted conditions,respectively. GIS was used to apply theseconstraints to the PRISM climate grids toproduce maps showing climatic adaptation zonesfor each species.

3. RESULTS

Maps of all of mainland China and Taiwanare being produced, but only the eastern half ofmainland China were available at the time ofthis writing. Figure 1 shows mean annualprecipitation. Mean July and mean Januaryprecipitation are shown on the poster only, butare discussed here. Spatial patterns inprecipitation are related primarily to elevationand exposure to the southeast monsoon. On anannual basis, the southeastern quarter of thecountry is the wettest, with amounts in coastalareas and in the mountains exceeding 2000 mmper year. Precipitation gradually decreases asone moves northward and westward. Parts ofInner Mongolia, shielded by several mountainranges from the East China Sea, receives lessthan 100 m per year. Mountainous terrainproduces local maxima up to about 1000-1500 min most areas, except the Tibetan Plateau, whereprecipitation maxima are at much higherelevations.

Summer is the wet season and winter is thedry season across eastern China. Patterns ofJuly precipitation are similar to those of annualprecipitation, with some notable differences.High mountains on the island of Taiwanproduces a distinct rain shadow on the mainlandcoast to the northwest, in Fujian Province. Thedry zone extends west of this area and wellinland into Hunan Province. Further north, thereis a precipitation maximum on the coast in


Shaddong and Jiansu Provinces that is notpresent in the annual map, reflecting a zone ofmoisture transport during mid-summer. EasternChina is very dry in January, when the region isdominated by cold, dry air from the Asianinterior. Most areas receive less than 25 mm,and the maximum is less than 125 mm in thesoutheastern mountains.

July and January mean daily maximumtemperatures are shown on the poster only, butare discussed here. July temperature patterns aredominated by elevation, and show a very stronglapse rate. The coolest temperatures are found atthe highest elevations of the eastern TibetanPlateau, where daily maxima are only 12-16oC.In general, the coastal strip is cooler thanadjacent inland areas. This is most easily seenwhere Shandong Province extends eastward intothe Yellow Sea in the vicinity of Qingdao.Maximum July daily temperatures are highest inthe inland areas of southeastern China, includingthe city of Nanchang, Jiangxi Province, knownas one of the three “stoves” of China. Here, Julytemperatures top out at 34-36oC with highhumidity. Nearby mountains provide mildersummertime temperatures of 24-29oC.

January maximum daily temperature isdominated by latitude, or more specifically,exposure to the air mass associated with theSiberian high pressure center. Maximum dailytemperatures in extreme northern China are -20oC and below, which are colder than thosefound at the highest elevations of the easternTibetan Plateau (-5 to -10oC). On the easternlowlands, temperatures rise steadily as onemoves southward, away from the Siberianinfluence. Hainan Island far to the south andseparated from the mainland by water, retainscomfortable maximum temperatures of 19-22oC,even in January. The effect of elevation is not aspronounced in winter, due to the tendency oflow-level temperature inversions to form in thedry, clear air.

An example of an early draft of a speciesadaptation map for tall fescue is shown in Figure2. The patterns of adaptation are caused bydifferent factors in different locations. Lowwinter temperatures limit the northeastern rangeof tall fescue, while low precipitation limits isnorthwestern range. Low summer temperatureslimit the range on the Tibetan Plateau. In the

southeast interior, it is high summertemperatures that limit the range. The well-adapted zone is extensive; optimumtemperatures and rainfall for tall fescue existacross the central lowlands, then extendsouthwestward, following mild temperatures atmoderate elevations all the way to the southernborder. In addition, many mountainous andcoastal areas in southeastern China show goodadaptation, because of mild summertemperatures at higher elevations.

4. CONCLUSIONS

With the rapid and continuous economicgrowth in China, market demands for grass seedfor forage-livestock systems, urbanbeautification, and improved environmentalprotection have greatly expanded in the lastdecade. In 1992 the Oregon Seed Councilpartnered with Oregon State University toinitiate a USDA Market Access Program projectfor developing the China market for Oregon-grown grass seed. Effective marketing of ourhigh quality US-grown seeds requires that webe able to identify all of the areas suitable forusing these grasses. Until now that has beenimpossible on a wide scale. Fortunately, GIStechnology, combined with climate mappingexpertise, spatial soils data and climatictolerances of grass species, can be used toproduce very detailed maps of speciesadaptation that can be used to accurately identifysuitable growing areas for effective marketing.

The objectives of the work described herewere to: (1) obtain and quality-check observedclimate data from Chinese authorities; (2)prepare detailed draft maps of mean monthlyminimum/maximum temperature andprecipitation; (3) use the climate maps andexpert estimates of species climatic tolerances toprepare adaptation maps for several importantOregon-grown grasses; and (4) subject draftclimate and adaptation maps to peer review byChinese scientists, obtain supplementary climatedata, and revise maps as necessary. This paperdescribed the methods used to accomplish thefirst three objectives for eastern China.

Further work in climate mapping willinvolve expanding the climate mapping to all ofChina, attempting to obtain more climate data,


and subject the maps to peer review. Furtheractivities in species adaptation mapping includeusing expert opinion to better quantify theclimatic tolerances of grass species. The rangeand types of adaptation maps will also beexpanded to include soils tolerances, andmanagement prescriptions such as irrigation,fertilization, and intercropping with other cropspecies at specified times of the year.

5. REFERENCES

Daly, C., W. P. Gibson, G.H. Taylor, G. L.Johnson, P.Pasteris. In review. Towards aknowledge-based approach to the statisticalmapping of climate. Climate Research.

Daly, C. and G.L. Johnson. 1999. PRISMspatial climate layers: their development anduse. Short Course on Topics in AppliedClimatology, 79th Annual Meeting of theAmerican Meteorological Society, 10-15January, Dallas, TX. 49 pp.http://www.ocs.orst.edu/prism/prisguid.pdf.

Daly, C., T.G.F. Kittel, A. McNab, J.A. Royle,W.P. Gibson, T. Parzybok, N. Rosenbloom,G.H. Taylor, and H. Fisher. 1999.Development of a 102-year high-resolutionclimate data set for the conterminous UnitedStates. In: Proceedings, 10th Symposium onGlobal Change Studies, 79th Annual

Meeting of the American MeteorologicalSociety, 10-15 January, Dallas, TX, 480-483.

Daly, C., R.P. Neilson, and D.L. Phillips. 1994.A statistical-topographic model for mappingclimatological precipitation overmountainous terrain. Journal of AppliedMeteorology 33:140-158.

Daly, C., G.H. Taylor, W. P. Gibson, T.W.Parzybok, G. L. Johnson, P. Pasteris. Inreview. High-quality spatial climate datasets for the United States. AppliedEngineering in Agriculture.

Daly, C., G.H. Taylor, and W.P. Gibson. 1997.The PRISM approach to mappingprecipitation and temperature. In: Proc.,10th AMS Conf. on Applied Climatology,Amer. Meteorological Soc., Reno, NV, Oct.20-23, 10-12.

Johnson, G.L., C. Daly, C.L. Hanson, Y.Y. Luand G.H. Taylor. In press. Spatialvariability and interpolation of stochasticweather simulation model parameters.Journal of Applied Meteorology.

USDA-NRCS. 1998. PRISM Climate MappingProject--Precipitation. Mean monthly andannual precipitation digital files for thecontinental U.S. USDA-NRCS NationalCartography and Geospatial Center, Ft.Worth TX. December, CD-ROM.


1. introduction€¦ · introduction with the rapid and continuous economic growth in china, market...

Documents