Evaluation_NN1 of_IO the_AT IRI-2007_MC model_NN1 options_NN2 for_IF the_AT topside_NN1 electron_NN1 density_NN1 1_MC1 ._. 
Introduction_NN1 The_AT International_JJ Reference_NN1 Ionosphere_NN1 (_( IRI_NP1 )_) is_VBZ the_AT de_JJ21 facto_JJ22 standard_NN1 for_IF ionospheric_JJ parameters_NN2 and_CC is_VBZ widely_RR used_VVN to_TO specify_VVI ionospheric_JJ conditions_NN2 ._. 
For_IF many_DA2 applications_NN2 the_AT Total_JJ Electron_NN1 Content_NN1 (_( TEC_NP1 )_) is_VBZ the_AT parameter_NN1 of_IO interest_NN1 because_CS it_PPH1 determines_VVZ the_AT ionosphere_NN1 's_GE refractive_JJ and_CC retarding_VVG effect_NN1 on_II radio_NN1 signals_NN2 ._. 
TEC_NP1 consists_VVZ of_IO a_AT1 bottomside_NN1 portion_NN1 and_CC a_AT1 topside_NN1 portion_NN1 ._. 
The_AT bottomside_NN1 contributes_VVZ only_RR about_RG 20%_NNU of_IO the_AT TEC_NP1 ._. 
The_AT dominant_JJ contribution_NN1 comes_VVZ form_VV0 the_AT topside_NN1 and_CC the_AT topside_NN1 electron_NN1 density_NN1 profile_NN1 is_VBZ therefore_RR of_IO utmost_JJ importance_NN1 for_IF many_DA2 IRI_JJ users_NN2 ._. 
A_AT1 number_NN1 of_IO studies_NN2 had_VHD noted_VVN discrepancies_NN2 between_II the_AT IRI-2001_MC model_NN1 and_CC measurements_NN2 ,_, especially_RR in_II the_AT upper_JJ topside_NN1 ._. 
At_II issue_NN1 is_VBZ an_AT1 overestimation_NN1 of_IO electron_NN1 densities_NN2 in_II the_AT upper_JJ topside_NN1 that_CST increases_VVZ with_IW altitude_NN1 reaching_VVG about_II a_AT1 factor_NN1 of_IO 3_MC at_II 1000_MC km_NNU above_II the_AT peak_NN1 ._. 
The_AT likely_JJ causes_NN2 of_IO this_DD1 IRI_JJ artifact_NN1 are_VBR the_AT limited_JJ data_NN base_NN1 used_VMK to_TO develop_VVI the_AT original_JJ IRI_JJ model_NN1 ,_, primarily_RR Alouette_NP1 1_MC1 topside_NN1 sounder_NN1 data_NN with_IW some_DD AE-C_JJ ,_, and_CC DE-2_MC in_RR21 situ_RR22 data_NN and_CC typical_JJ profiles_NN2 from_II the_AT Jicamarca_NP1 incoherent_JJ scatter_NN1 radar_NN1 ,_, and_CC a_AT1 fitting_JJ process_NN1 that_CST is_VBZ biased_VVN towards_II F-region_JJ densities_NN2 because_CS they_PPHS2 are_VBR one_MC1 or_CC two_MC orders_NN2 of_IO magnitude_NN1 above_II densities_NN2 in_II the_AT upper_JJ topside_NN1 ._. 
An_AT1 improvement_NN1 of_IO the_AT topside_NN1 electron_NN1 density_NN1 model_NN1 has_VHZ been_VBN a_AT1 high_JJ priority_NN1 for_IF IRI_JJ developers_NN2 and_CC users_NN2 ._. 
Several_DA2 groups_NN2 investigated_VVD model_NN1 corrections_NN2 and_CC new_JJ approaches_NN2 to_TO overcome_VVI this_DD1 shortcoming_NN1 of_IO the_AT IRI-2001_MC model_NN1 ._. 
These_DD2 efforts_NN2 led_VVN to_II the_AT introduction_NN1 of_IO two_MC new_JJ options_NN2 for_IF the_AT electron_NN1 topside_NN1 model_NN1 in_II the_AT latest_JJT version_NN1 of_IO the_AT model_NN1 ,_, IRI-2007_MC ._. 
The_AT first_MD option_NN1 is_VBZ a_AT1 correction_NN1 factor_NN1 for_IF the_AT IRI-2001_MC topside_NN1 model_NN1 that_CST was_VBDZ developed_VVN by_II Bilitza_NP1 based_VVN on_II over_RG 150,000_MC topside_NN1 profiles_NN2 deduced_VVN from_II Alouette_NP1 1_MC1 ,_, 2_MC ,_, and_CC ISIS_NP1 1_MC1 ,_, 2_MC topside_NN1 soundings_NN2 and_CC describes_VVZ variations_NN2 with_IW altitude_NN1 ,_, modified_JJ dip_NN1 latitude_NN1 ,_, and_CC local_JJ time_NNT1 ._. 
With_IW this_DD1 correction_NN1 term_NN1 the_AT IRI_JJ model_NN1 represents_VVZ the_AT topside_NN1 sounder_NN1 data_NN quite_RG well_RR as_CSA shown_VVN in_II the_AT examples_NN2 in_II Fig._NN1 1_MC1 ._. 
The_AT second_MD option_NN1 is_VBZ the_AT NeQuick_NN1 topside_NN1 model_NN1 developed_VVN by_II Radicella_NP1 and_CC Leitinger_NP1 and_CC Coisson_NP1 et_RA21 al_RA22 ._. 
based_VVN on_II ionosonde_NN1 and_CC topside_NN1 sounder_NN1 data_NN ._. 
It_PPH1 is_VBZ the_AT most_RGT mature_JJ of_IO the_AT different_JJ proposals_NN2 for_IF the_AT IRI_JJ topside_NN1 that_CST were_VBDR described_VVN by_II Bilitza_NP1 et_RA21 al_RA22 ._. 
The_AT NeQuick_NN1 topside_NN1 model_NN1 uses_VVZ an_AT1 Epstein-layer_JJ function_NN1 with_IW a_AT1 height-dependent_JJ thickness_NN1 parameter_NN1 and_CC in_II this_DD1 way_NN1 produces_VVZ a_AT1 smooth_JJ transition_NN1 from_II an_AT1 atomic_JJ oxygen_NN1 ionosphere_NN1 near_II the_AT F-peak_NN1 to_II a_AT1 light_JJ ion_NN1 ionosphere_NN1 higher_RRR up_RP ._. 
The_AT model_NN1 parameters_NN2 were_VBDR determined_VVN based_VVN on_II fitting_VVG this_DD1 function_NN1 to_II ISIS_NP1 1_MC1 ,_, 2_MC and_CC Intercosmos_NP1 19_MC topside_NN1 sounder_NN1 profiles_NN2 ._. 
We_PPIS2 will_VM call_VVI the_AT first_MD option_NN1 IRI-2007-cor_NN2 and_CC the_AT second_MD option_NN1 IRI-2007-NeQ_NN2 in_II the_AT remainder_NN1 of_IO the_AT article_NN1 ._. 
In_II this_DD1 study_NN1 we_PPIS2 use_VV0 Alouette_NN1 and_CC ISIS_NP1 topside_NN1 sounder_NN1 data_NN to_TO evaluate_VVI the_AT two_MC new_JJ options_NN2 in_II31 relation_II32 to_II33 IRI-2001_MC and_CC to_TO assess_VVI which_DDQ one_MC1 of_IO the_AT two_MC options_NN2 shows_VVZ better_JJR agreement_NN1 with_IW the_AT data_NN ._. 
We_PPIS2 have_VH0 also_RR included_VVN the_AT model_NN1 of_IO Triskova_NP1 et_RA21 al_RA22 ._. 
(_( TTS_NP1 model_NN1 )_) that_DD1 was_VBDZ developed_VVN based_VVN on_II Atmosphere_NN1 Explorer_NN1 C_ZZ1 ,_, D_ZZ1 ,_, and_CC E_ZZ1 data_NN for_IF low_JJ solar_JJ activities_NN2 and_CC Interkosmos_NP1 24_MC and_CC 24_MC data_NN for_IF high_JJ solar_JJ activity._NNU 2_MC ._. 
Alouette_NN1 and_CC ISIS_NP1 topside_NN1 sounder_NN1 data_NN The_AT Alouette_NN1 1_MC1 ,_, 2_MC and_CC ISIS_NP1 1_MC1 ,_, 2_MC satellites_NN2 were_VBDR ionospheric_JJ observatories_NN2 designed_VVN and_CC operated_VVN jointly_RR by_II the_AT USA_NP1 and_CC Canada_NP1 ._. 
The_AT launch_NN1 dates_NN2 and_CC orbit_NN1 characteristics_NN2 for_IF the_AT four_MC satellites_NN2 are_VBR listed_VVN in_II Table_NN1 1_MC1 ._. 
The_AT primary_JJ instrument_NN1 was_VBDZ the_AT topside_NN1 sounder_NN1 for_IF recording_VVG the_AT electron_NN1 density_NN1 profile_NN1 from_II the_AT satellite_NN1 down_RP to_II the_AT F-peak_NN1 ._. 
But_CCB other_JJ instruments_NN2 were_VBDR included_VVN as_RR21 well_RR22 ._. 
ISIS-2_MC ,_, for_REX21 example_REX22 ,_, carried_VVD a_AT1 sweep-_NN1 and_CC a_AT1 fixed-frequency_JJ ionosonde_NN1 ,_, a_AT1 VLF_NP1 receiver_NN1 ,_, energetic_JJ and_CC soft_JJ particle_NN1 detectors_NN2 ,_, an_AT1 ion_NN1 mass_NN1 spectrometer_NN1 ,_, an_AT1 electrostatic_JJ probe_NN1 ,_, a_AT1 retarding_JJ potential_JJ analyzer_NN1 ,_, a_AT1 beacon_NN1 transmitter_NN1 ,_, a_AT1 cosmic_JJ noise_NN1 experiment_NN1 ,_, and_CC two_MC photometers_NN2 ._. 
NASA_NP1 support_NN1 of_IO the_AT ISIS_NP1 project_NN1 was_VBDZ terminated_VVN on_II October_NPM1 1_MC1 ,_, 1979_MC ._. 
Partial_JJ operations_NN2 were_VBDR continued_VVN by_II the_AT Canadian_JJ project_NN1 team_NN1 until_II March_NPM1 9_MC ,_, 1984_MC and_CC were_VBDR then_RT resumed_VVN in_RP even_RR more_RGR limited_JJ form_NN1 by_II the_AT Japanese_JJ Radio_NN1 Research_NN1 Laboratories_NN2 (_( Kashima_NP1 ground_NN1 station_NN1 )_) from_II August_NPM1 1984_MC to_II January_NPM1 24_MC ,_, 1990_MC ._. 
These_DD2 satellites_NN2 have_VH0 accumulated_VVN a_AT1 large_JJ volume_NN1 of_IO data_NN for_IF the_AT topside_NN1 ionosphere_NN1 ._. 
The_AT inversion_NN1 from_II the_AT recorded_JJ ionogram_NN1 to_II the_AT topside_NN1 electron_NN1 density_NN1 profile_NN1 ,_, however_RR ,_, involves_VVZ a_AT1 tedious_JJ and_CC highly_RR subjective_JJ manual_JJ process_NN1 that_CST could_VM only_RR be_VBI performed_VVN for_IF a_AT1 small_JJ percentage_NN1 of_IO the_AT collected_JJ ionogram_NN1 data_NN ._. 
The_AT National_JJ Space_NN1 Science_NN1 Data_NN Center_NN1 (_( NSSDC_NP1 )_) has_VHZ archived_VVN most_DAT of_IO these_DD2 data_NN and_CC has_VHZ made_VVN them_PPHO2 available_JJ on_II its_APPGE ftp_NNU site_NN1 ._. 
The_AT different_JJ available_JJ data_NN sets_NN2 are_VBR listed_VVN in_II Table_NN1 1_MC1 and_CC plotted_VVN in_II Fig._NN1 1_MC1 showing_VVG their_APPGE time_NNT1 and_CC altitude_NN1 coverage_NN1 ._. 
The_AT data_NN stretch_VV0 across_RL almost_RR two_MC solar_JJ cycles_NN2 and_CC go_VV0 up_RG21 to_RG22 3500_MC km_NNU altitude_NN1 ._. 
All_DB except_II the_AT second_MD Alouette_NN1 1_MC1 data_NN set_VVD were_VBDR generated_VVN at_II the_AT Communications_NN2 Research_VV0 Center_NN1 (_( CRC_NP1 )_) in_II Ottawa_NP1 ,_, Canada_NP1 using_VVG manual_JJ scaling_NN1 and_CC the_AT Jackson_NP1 ionogram_NN1 inversion_NN1 program_NN1 ._. 
The_AT responsible_JJ group_NN1 for_IF the_AT second_MD Alouette_NN1 1_MC1 data_NN set_NN1 was_VBDZ the_AT University_NN1 of_IO California_NP1 Los_NP1 Angeles_NP1 (_( UCLA_NP1 )_) Department_NN1 of_IO Meteorology_NN1 using_VVG a_AT1 data_NN analysis_NN1 process_VV0 similar_JJ to_II the_AT CRC_NP1 ._. 
Unfortunately_RR ,_, in_II many_DA2 cases_NN2 the_AT profiles_NN2 do_VD0 not_XX reach_VVI all_DB the_AT way_NN1 down_RP to_II the_AT peak_NN1 ,_, because_II21 of_II22 the_AT difficulties_NN2 in_II scaling_VVG the_AT trace_NN1 in_II the_AT cusp_NN1 region_NN1 near_II the_AT peak_NN1 ._. 
This_DD1 is_VBZ a_AT1 problem_NN1 of_IO the_AT inversion_NN1 technique_NN1 ._. 
Bent_JJ et_RA21 al_RA22 ._. 
tried_VVD to_TO overcome_VVI this_DD1 problem_NN1 by_II using_VVG a_AT1 model_NN1 for_IF the_AT peak_NN1 height_NN1 and_CC by_II assuming_VVG the_AT lowest_JJT profile_NN1 point_NN1 density_NN1 always_RR at_II a_AT1 constant_JJ ratio_NN1 below_II the_AT actual_JJ F2_FO peak_NN1 density_NN1 ._. 
We_PPIS2 have_VH0 used_VVN a_AT1 different_JJ approach_NN1 by_II fitting_VVG a_AT1 Chapman_NP1 layer_NN1 to_II the_AT lower_JJR part_NN1 of_IO the_AT profile_NN1 ._. 
The_AT least-square_JJ fitting_JJ procedure_NN1 finds_VVZ the_AT best_JJT Chapman_NP1 parameters_NN2 for_IF representing_VVG the_AT topside_NN1 F-layer_NN1 ._. 
Profiles_NN2 are_VBR only_RR considered_VVN if_CS they_PPHS2 had_VHD at_RR21 least_RR22 five_MC profile_NN1 points_VVZ in_II the_AT lower_JJR topside_NN1 and_CC if_CS they_PPHS2 could_VM be_VBI fitted_VVN with_IW a_AT1 standard_JJ deviation_NN1 of_IO 1%_NNU or_CC better_RRR ._. 
About_RG 120,000_MC profiles_NN2 remained_VVN after_II this_DD1 fitting_NN1 and_CC selection_NN1 process_NN1 ._. 
The_AT examples_NN2 in_II Fig._NN1 2a_FO and_CC b_ZZ1 include_VV0 the_AT determined_JJ peak_NN1 point_NN1 and_CC show_VV0 that_CST the_AT procedure_NN1 works_VVZ reasonably_RR well_RR ._. 
It_PPH1 is_VBZ worth_II noting_VVG that_CST a_AT1 data_NN save_VV0 and_CC restoration_NN1 effort_NN1 at_II NASA_NP1 's_GE Goddard_NP1 Space_NN1 Flight_NN1 Center_NN1 (_( R._NP1 Benson_NP1 ,_, PI_NN1 )_) was_VBDZ able_JK to_TO digitize_VVI a_AT1 large_JJ number_NN1 of_IO the_AT original_JJ analog_NN1 telemetry_NN1 tapes_NN2 and_CC use_VV0 the_AT Topside_NP1 Ionogram_NP1 Scaler_NP1 with_IW True_JJ Height_NN1 Algorithm_NN1 (_( TOPIST_NP1 )_) program_NN1 developed_VVN at_II the_AT University_NN1 of_IO Massachusetts_NP1 Lowell_NP1 for_IF the_AT automated_JJ processing_NN1 of_IO these_DD2 data_NN ._. 
This_DD1 effort_NN1 will_VM provide_VVI additional_JJ profiles_NN2 for_IF a_AT1 follow-on_JJ evaluation_NN1 study._NNU 3_MC ._. 
Comparing_VVG model_NN1 and_CC data_NN --_JJ topside_NN1 profile_NN1 shape_NN1 In_II IRI_NP1 the_AT electron_NN1 density_NN1 profile_NN1 is_VBZ normalized_VVN to_II the_AT F2_FO peak_NN1 density_NN1 and_CC height_NN1 ._. 
Data-model_NN1 discrepancies_NN2 in_II the_AT absolute_JJ electron_NN1 density_NN1 in_II the_AT topside_NN1 can_VM therefore_RR be_VBI attributed_VVN to_II two_MC error_NN1 sources_NN2 :_: the_AT IRI_JJ peak_NN1 models_NN2 or_CC the_AT IRI_JJ topside_NN1 profile_NN1 model_NN1 ._. 
If_CS we_PPIS2 want_VV0 to_TO evaluate_VVI the_AT reliability_NN1 of_IO the_AT topside_NN1 model_NN1 ,_, we_PPIS2 first_MD have_VH0 to_TO eliminate_VVI the_AT influence_NN1 of_IO the_AT IRI_JJ peak_NN1 models_NN2 ._. 
This_DD1 is_VBZ done_VDN by_II normalizing_VVG the_AT IRI_JJ profile_NN1 with_IW the_AT measured_JJ F-peak_JJ density_NN1 and_CC shifting_VVG it_PPH1 to_II the_AT measured_JJ F-peak_JJ height_NN1 ._. 
Fig._NN1 2a_FO and_CC b_ZZ1 shows_VVZ the_AT resulting_JJ profiles_NN2 clearly_RR illustrating_VVG the_AT overestimation_NN1 of_IO the_AT topside_NN1 gradient_NN1 in_II IRI-2001_MC ._. 
For_IF both_DB2 examples_NN2 the_AT new_JJ IRI-2007_MC topside_NN1 options_NN2 reproduce_VV0 the_AT observed_JJ topside_NN1 shape_NN1 quite_RG well_RR ._. 
The_AT TTS_NP1 model_NN1 was_VBDZ designed_VVN primarily_RR as_II a_AT1 representation_NN1 of_IO the_AT absolute_JJ density_NN1 values_NN2 mostly_RR above_II the_AT F-peak_NN1 and_CC so_RR does_VDZ not_XX include_VVI an_AT1 option_NN1 to_TO normalize_VVI with_IW the_AT measured_JJ peak_NN1 values_NN2 ._. 
But_CCB we_PPIS2 have_VH0 included_VVN the_AT model_NN1 here_RL also_RR because_CS it_PPH1 was_VBDZ proposed_VVN for_IF the_AT representation_NN1 of_IO the_AT total_JJ ion_NN1 density_NN1 in_II31 conjunction_II32 with_II33 the_AT IRI_JJ ion_NN1 composition_NN1 model_NN1 that_CST was_VBDZ developed_VVN by_II the_AT same_DA authors_NN2 ._. 
In_II Fig._NN1 3a-c_FO ,_, we_PPIS2 have_VH0 plotted_VVN the_AT model_NN1 predictions_NN2 versus_II the_AT ISIS-2_MC topside_NN1 sounder_NN1 measurements_NN2 for_IF the_AT three_MC IRI_JJ models_NN2 ._. 
The_AT plot_NN1 includes_VVZ a_AT1 total_NN1 of_IO 223,779_MC data_NN points_NN2 ._. 
If_CS data_NN and_CC model_NN1 agree_VV0 well_RR the_AT data_NN points_NN2 should_VM be_VBI all_RR clustered_VVN around_II the_AT central_JJ diagonal_JJ ._. 
The_AT figure_NN1 for_IF IRI-2001_MC shows_VVZ the_AT noted_JJ problem_NN1 at_II high_JJ altitudes_NN2 where_RRQ the_AT model_NN1 predictions_NN2 become_VV0 almost_RR constant_JJ and_CC exceed_VV0 the_AT sounder_JJR data_NN ._. 
Values_NN2 computed_VVN with_IW the_AT two_MC new_JJ IRI-2007_MC topside_NN1 options_NN2 are_VBR plotted_VVN versus_II the_AT sounder_JJR data_NN in_II Fig._NN1 3b_FO and_CC c_ZZ1 ._. 
Both_DB2 models_NN2 overcome_VV0 the_AT problem_NN1 at_II high_JJ altitudes_NN2 and_CC show_VV0 good_JJ agreement_NN1 with_IW the_AT data_NN across_II the_AT whole_JJ range_NN1 covered_VVN by_II the_AT sounder_JJR data_NN ._. 
IRI-2007-cor_NP1 seems_VVZ to_TO slightly_RR overestimate_VVI the_AT data_NN while_CS the_AT IRI-2007-NeQ_NP1 plot_NN1 indicates_VVZ an_AT1 underestimation_NN1 ._. 
To_TO investigate_VVI this_DD1 further_RRR we_PPIS2 have_VH0 plotted_VVN in_II Fig._NN1 4a-c_FO the_AT data-model_NN1 ratios_NN2 at_II an_AT1 altitude_NN1 of_IO 1000_MC km_NNU above_II the_AT F2_FO peak_NN1 ._. 
The_AT IRI-2001_MC plot_VV0 clearly_RR shows_VVZ the_AT general_JJ overestimation_NN1 reaching_VVG a_AT1 factor_NN1 of_IO 10_MC at_II the_AT magnetic_JJ equator_NN1 and_CC at_II high_JJ latitudes_NN2 ._. 
With_IW the_AT IRI-2007_MC models_NN2 the_AT factor_NN1 has_VHZ been_VBN reduced_VVN to_II 2_MC ._. 
In_II31 addition_II32 to_II33 the_AT general_JJ overestimation_NN1 the_AT plots_NN2 for_IF IRI-2001_MC and_CC IRI-2007-cor_NN2 exhibit_VV0 also_RR a_AT1 distinct_JJ latitudinal_JJ variation_NN1 ._. 
The_AT ratios_NN2 are_VBR lowest_JJT (_( best_JJT agreement_NN1 )_) at_II mid-latitudes_NN2 and_CC reach_VV0 the_AT largest_JJT values_NN2 at_II the_AT magnetic_JJ equator_NN1 and_CC at_II high_JJ latitudes_NN2 ._. 
This_DD1 again_RT might_VM be_VBI due_II21 to_II22 deficiencies_NN2 in_II the_AT least-square_JJ fitting_NN1 process_VV0 that_DD1 was_VBDZ used_VVN to_TO generate_VVI the_AT IRI-2001_MC topside_NN1 model_NN1 ._. 
Insufficient_JJ weighting_NN1 of_IO the_AT low_JJ topside_NN1 densities_NN2 could_VM result_VVI in_II the_AT latitudinal_JJ variations_NN2 being_VBG dominated_VVN by_II the_AT behavior_NN1 at_II F-region_JJ heights_NN2 which_DDQ is_VBZ quite_RG different_JJ from_II the_AT variation_NN1 in_II the_AT topside_NN1 ._. 
The_AT typical_JJ equatorial_JJ anomaly_NN1 crests_NN2 on_II both_DB2 sides_NN2 of_IO the_AT equator_NN1 observed_VVN at_II F-region_JJ heights_NN2 move_VV0 closer_RRR towards_II the_AT magnetic_JJ equator_NN1 with_IW increasing_JJ height_NN1 until_CS they_PPHS2 merge_VV0 into_II a_AT1 single_JJ peak_NN1 at_II about_RG 1000_MC km_NNU above_II the_AT peak_NN1 ._. 
This_DD1 characteristic_JJ behavior_NN1 is_VBZ well_RR reproduced_VVN by_II the_AT IRI-2007-cor_NP1 model_NN1 as_CSA can_VM be_VBI seen_VVN in_II the_AT latitude-height_JJ contour_NN1 map_NN1 shown_VVN in_II Fig._NN1 5b_FO ._. 
The_AT old_JJ IRI-2001_MC model_NN1 produces_VVZ unrealistically_RR steep_JJ profiles_NN2 in_II the_AT upper_JJ topside_NN1 whereas_CS the_AT NeQuick_NN1 option_NN1 shows_VVZ the_AT typical_JJ double-hump_JJ structure_NN1 of_IO the_AT EA_NP1 extending_VVG up_RG21 to_RG22 1500_MC km_NNU altitude_NN1 and_CC not_XX merging_VVG to_II a_AT1 single_JJ peak_NN1 as_CSA expected._NNU 4_MC ._. 
Comparing_VVG model_NN1 and_CC data_NN --_JJ overall_NN1 result_VV0 So_RG far_RR our_APPGE comparisons_NN2 were_VBDR based_VVN primarily_RR on_II the_AT ISIS-2_MC data_NN set_VV0 because_CS it_PPH1 provides_VVZ the_AT best_JJT coverage_NN1 in_II altitude_NN1 and_CC latitude_NN1 ._. 
The_AT other_JJ 5_MC data_NN sets_VVZ in_II Table_NN1 1_MC1 produce_VV0 similar_JJ results_NN2 and_CC similar_JJ plots_NN2 to_II Fig._NN1 3_MC and_CC Fig._NN1 4_MC ._. 
To_TO get_VVI an_AT1 overall_JJ estimation_NN1 of_IO the_AT data-model_NN1 discrepancies_NN2 we_PPIS2 have_VH0 computed_VVN the_AT mean_JJ and_CC standard_JJ deviation_NN1 of_IO the_AT relative_JJ deviation_NN1 between_II data_NN and_CC models_NN2 for_IF all_DB six_MC data_NN sets_VVZ separately_RR ._. 
The_AT results_NN2 are_VBR listed_VVN in_II Table_NN1 2_MC ._. 
The_AT percentage_NN1 deviation_NN1 is_VBZ computed_VVN as_II ._. 
In_II all_DB cases_NN2 we_PPIS2 get_VV0 a_AT1 negative_JJ PD_NP1 indicating_VVG that_CST on_II average_NN1 all_DB 4_MC models_NN2 overestimate_VV0 the_AT sounder_JJR data_NN ._. 
As_CSA expected_VVN the_AT largest_JJT discrepancies_NN2 are_VBR found_VVN with_IW the_AT IRI-2001_MC and_CC TTS_NP1 model_NN1 because_CS IRI-2001_MC misrepresent_NN1 the_AT upper_JJ topside_NN1 and_CC TTS_NP1 is_VBZ not_XX normalized_VVN to_II the_AT F-peak_NN1 ._. 
The_AT two_MC new_JJ IRI-2007_MC options_NN2 produce_VV0 significant_JJ improvements_NN2 over_II the_AT older_JJR model_NN1 reducing_VVG the_AT data-model_NN1 difference_NN1 by_II a_AT1 factor_NN1 of_IO up_RG21 to_RG22 10_MC ._. 
For_IF all_DB six_MC data_NN sets_VVZ best_JJT results_NN2 are_VBR obtained_VVN with_IW IRI-2007-NeQ_NN2 showing_VVG at_II times_NNT2 a_AT1 factor_NN1 of_IO 2_MC better_JJR representation_NN1 than_CSN IRI-2007-cor_NN2 ._. 
Taking_VVG all_DB Alouette_NN1 and_CC ISIS_NP1 data_NN together_RL we_PPIS2 find_VV0 that_CST the_AT 165%_NNU overestimation_NN1 of_IO the_AT data_NN with_IW IRI-2001_MC is_VBZ reduced_VVN to_II 46%_NNU with_IW IRI-2007-cor_NN2 and_CC to_II 24%_NNU with_IW IRI-2007-NeQ_NN2 ._. 
The_AT new_JJ topside_NN1 models_NN2 also_RR significantly_RR reduce_VV0 the_AT scatter_NN1 of_IO difference_NN1 values_NN2 as_CSA shown_VVN by_II the_AT decrease_NN1 of_IO the_AT standard_JJ deviation_NN1 by_II a_AT1 factor_NN1 of_IO 2-5._MCMC 6_MC ._. 
Summary_NN1 and_CC conclusions_NN2 IRI-2007_MC introduces_VVZ two_MC new_JJ model_NN1 options_NN2 for_IF the_AT topside_NN1 electron_NN1 density_NN1 ,_, a_AT1 correction_NN1 of_IO IRI-2001_MC and_CC the_AT NeQuick_NN1 model_NN1 ._. 
In_II our_APPGE study_NN1 we_PPIS2 have_VH0 evaluated_VVN these_DD2 new_JJ options_NN2 with_IW six_MC data_NN sets_NN2 of_IO topside_NN1 profiles_NN2 deduced_VVN from_II Alouette_NP1 1_MC1 ,_, 2_MC and_CC ISIS_NP1 1_MC1 ,_, 2_MC topside_NN1 sounder_NN1 measurements_NN2 ._. 
The_AT total_JJ number_NN1 of_IO profiles_NN2 used_VVN for_IF our_APPGE study_NN1 is_VBZ 120,059_MC ._. 
It_PPH1 is_VBZ important_JJ to_TO note_VVI that_CST our_APPGE evaluation_NN1 is_VBZ done_VDN after_II the_AT influence_NN1 of_IO the_AT F-peak_JJ models_NN2 has_VHZ been_VBN eliminated_VVN ._. 
This_DD1 was_VBDZ done_VDN by_II normalizing_VVG the_AT model_NN1 profile_NN1 with_IW the_AT measured_JJ peak_NN1 density_NN1 and_CC height_NN1 ._. 
Our_APPGE main_JJ results_NN2 are_VBR :_: --_NN1 The_AT new_JJ options_NN2 result_VV0 in_II a_AT1 significant_JJ improvement_NN1 over_RG IRI-2001_MC ._. 
While_CS IRI-2001_MC overestimated_VVD the_AT data_NN by_II an_AT1 overall_JJ average_NN1 of_IO 165%_NNU ,_, IRI-2007-cor_NP1 reduced_VVD this_DD1 number_NN1 to_II 46%_NNU and_CC IRI-2007-NeQ_NN2 to_II 24%_NNU ._. 
