Datacollectedonthesurfaceoftheearthoftenhasspatialinteraction.Inthispaper,anon-isotropicmixingspatialdataprocessisintroduced,andundersuchaspatialstructureanonparametrickernelmethodissuggestedtoestimateaspatialconditionalregression.Undermildregularities,sufficientconditionsarederivedtoensuretheweakconsistencyaswellastheconvergenceratesforthekernelestimator.Ofinterestarethefollowing:(1)Alltheconditionsimposedonthemixingcoefficientandthebandwidtharesimple;(2)Differentlyfromthetimeseriessetting,thebandwidthisfoundtobedependentonthedimensionofthesiteinspaceaswell;(3)Forweakconsistency,themixingcoefficientisallowedtobeunsummableandthetendencyofsamplesizetoinfinitymaybeindifferentmannersalongdifferentdirectioninspace;(4)However,tohaveanoptimalconvergencerate,fasterdecreasingratesofmixingcoefficientandthetendencyofsamplesizetoinfinityalongeachdirectiona
