Statistics of validation of HelioClim-3 versus ground stations - 11 Brazilian INPE stations

Last update: Jan. 2016

    Generalities - BSRN - Belgium - Brazil INMET - Brazil INPE

Protocol

Date & Authors: Jan. 2016

  • Dr Etienne Wey, Transvalor
  • Pr Lucien Wald, MINES ParisTech
  • Dr Philippe Blanc, MINES ParisTech
  • and Dr Claire Thomas, Transvalor

(A publication is on the way)

The HelioClim-3 radiation values are compared to the 1 minute Global Horizontal Irradiation measurements from the 11 stations of the INPE network.
The measurements, also named observations, were originally quality checked using the procedure described here. As only the GHI component is available in the observations for both datasets, no consistency check is possible via cross-comparison from one component to another. Unfortunately, this protocol turned out to be insufficient for these hourly measurements and that the test to discard the not plausible values ("Extremely Rare Limits" and "Physical Possible Limits") were too permissive; Too many outlyers remained after the QC. Prior the computation of the different quantities to assess the deviation between the satellite estimates and the station measurements as depicted in the protocol of validation, we decided to discard also the measurement values when they are:

  • below 2% * clear sky dry, which corresponds to the ESRA clear sky with no turbidity
  • above 120% * clear sky dry
  • time shift in the data was also observed for several days in the time series. We decided that if more than 10% of the values were located during nightime, the whole day is discarded.

We remind that the other steps are:

  • Set night, sunrise and sunset observation values to zero
  • As soon as the quality check procedure is over, then aggregate the 1 min values to generate the hourly values if at least 85% of the 1-min slots are available. Apply an intelligent mean which takes into account the sun position at each minute.
  • Then the hourly values are aggregated to generate partial daily and monthly sums as depicted in the protocol of validation.

Illustration: list of the INMET and INPE stations used for the validation

List of the 11 stations

Station State Latitude Longitude Altitude (m) Period
CachoeiraPaulista RJ -22.690 -45.006 574 2014-03-01 to 2015-01-31
CampoGrande MS -20.438 -54.538 677 2012-09-01 to 2015-01-31
Chapeco SC -27.080 -52.614 700 2009-01-01 to 2009-12-31
Cuiaba MT -15.555 -56.070 185 2006-01-01 to 2012-12-31
Curitiba PR -25.495 -49.331 891 2014-09-01 to 2015-01-31
Natal RN -5.837 -35.206 58 2009-01-01 to 2015-01-31
Ourinhos SP -22.949 -49.894 446 2006-02-01 to 2011-03-31
Palmas TO -10.178 -48.362 216 2005-04-01 to 2014-06-30
Petrolina PE -9.069 -40.320 387 2007-01-01 to 2014-06-30
Saoluiz MA -2.594 -44.212 40 2009-01-01 to 2014-06-30
Sombrio SC -29.096 -49.813 15 2013-01-01 to 2015-01-31

Quality assessment results of GHI values

HelioClim-3 v4 and v5 -- GHI Quality assessment results vs. measurement stations in Brazil
MONTHLY SUM OF HOURLY IRRADIATION
Station Number of values Mean
- Station -
(kWh/m²)
Bias HC3v4
(kWh/m2)
(relative in %
)
Bias HC3v5
(kWh/m2)
(relative in %
)
RMSE HC3v4
(kWh/m2)
(relative in %
)
RMSE HC3v5
(kWh/m2)
(relative in %
)
Correl. coeff.
HC3v4
Correl. coeff.
HC3v5
CachoeiraPaulista (RJ) 11 147.7 3.9 (2.7%) -1.0 (-0.7%) 4.9 (3.3%) 4.4 (3.0%) 0.996 0.998
CampoGrande (MS) 18 132.0 -1.6 (-1.2%) 0.1 (0.1%) 3.0 (2.3%) 3.4 (2.5%) 0.998 0.997
Chapeco (SC) 12 129.9 5.0 (3.9%) 1.6 (1.2%) 8.4 (6.4%) 2.8 (2.2%) 0.993 0.998
Cuiaba (MT) 84 141.1 1.7 (1.2%) 2.4 (1.7%) 7.8 (5.5%) 7.8 (5.5%) 0.957 0.961
Curitiba (PR) 5 128.1 17.3 (13.5%) 3.2 (2.5%) 19.0 (14.8%) 3.7 (2.9%) 0.999 1.000
Natal (RN) 51 165.8 8.1 (4.9%) 5.9 (3.5%) 9.3 (5.6%) 7.4 (4.5%) 0.996 0.996
Ourinhos (SP) 62 128.6 9.0 (7.0%) 3.8 (3.0%) 10.9 (8.5%) 5.6 (4.4%) 0.988 0.992
Palmas (TO) 72 129.7 0.6 (0.4%) -1.2 (-0.9%) 6.7 (5.2%) 8.1 (6.3%) 0.986 0.980
Petrolina (PE) 89 147.5 6.5 (4.4%) 8.2 (5.5%) 7.5 (5.1%) 9.2 (6.2%) 0.996 0.995
Saoluiz (MA) 47 138.0 3.2 (2.3%) 2.8 (2.0%) 8.3 (6.0%) 7.3 (5.3%) 0.988 0.990
Sombrio (SC) 25 110.9 8.6 (7.8%) 3.8 (3.5%) 13.1 (11.8%) 5.4 (4.9%) 0.990 0.997

Top of page

HelioClim-3 v4 and v5 -- GHI Quality assessment results vs. measurement stations in Brazil
DAILY SUM OF HOURLY IRRADIATION
Station Number of values Mean
- Station -
(Wh/m²)
Bias HC3v4
(Wh/m2)
(relative in %
)
Bias HC3v5
(Wh/m2)
(relative in %
)
RMSE HC3v4
(Wh/m2)
(relative in %
)
RMSE HC3v5
(Wh/m2)
(relative in %
)
Correl. coeff.
HC3v4
Correl. coeff.
HC3v5
CachoeiraPaulista (RJ) 315 5159.3 137.0 (2.7%) -34.3 (-0.7%) 418.2 (8.1%) 371.3 (7.2%) 0.978 0.982
CampoGrande (MS) 447 5315.2 -63.3 (-1.2%) 5.6 (0.1%) 533.0 (10.0%) 545.2 (10.3%) 0.951 0.953
Chapeco (SC) 332 4694.6 181.2 (3.9%) 57.9 (1.2%) 531.1 (11.3%) 410.6 (8.7%) 0.972 0.979
Cuiaba (MT) 2349 5045.2 60.2 (1.2%) 84.5 (1.7%) 616.8 (12.2%) 632.4 (12.5%) 0.905 0.910
Curitiba (PR) 118 5429.1 732.7 (13.5%) 137.6 (2.5%) 904.9 (16.7%) 554.6 (10.2%) 0.960 0.962
Natal (RN) 1394 6064.1 296.3 (4.9%) 214.5 (3.5%) 458.1 (7.6%) 405.8 (6.7%) 0.966 0.967
Ourinhos (SP) 1635 4877.1 340.7 (7.0%) 146.0 (3.0%) 561.8 (11.5%) 422.1 (8.7%) 0.964 0.969
Palmas (TO) 1766 5288.9 23.4 (0.4%) -49.8 (-0.9%) 681.6 (12.9%) 715.0 (13.5%) 0.857 0.854
Petrolina (PE) 2435 5392.7 235.8 (4.4%) 298.7 (5.5%) 457.6 (8.5%) 501.3 (9.3%) 0.960 0.961
Saoluiz (MA) 1324 4899.2 114.6 (2.3%) 99.4 (2.0%) 512.6 (10.5%) 497.0 (10.1%) 0.935 0.937
Sombrio (SC) 646 4290.7 333.3 (7.8%) 148.9 (3.5%) 622.0 (14.5%) 391.9 (9.1%) 0.971 0.984

Top of page

HelioClim-3 v4 and v5 -- GHI Quality assessment results vs. measurement stations in Brazil
HOURLY IRRADIATION
Station Number of values Mean
- Station -
(Wh/m²)
Bias HC3v4
(Wh/m2)
(relative in %
)
Bias HC3v5
(Wh/m2)
(relative in %
)
RMSE HC3v4
(Wh/m2)
(relative in %
)
RMSE HC3v5
(Wh/m2)
(relative in %
)
Correl. coeff.
HC3v4
Correl. coeff.
HC3v5
CachoeiraPaulista (RJ) 3375 481.5 12.8 (2.7%) -3.2 (-0.7%) 83.4 (17.3%) 79.9 (16.6%) 0.961 0.963
CampoGrande (MS) 4937 481.2 -5.7 (-1.2%) 0.5 (0.1%) 111.0 (23.1%) 111.6 (23.2%) 0.931 0.933
Chapeco (SC) 3548 439.3 17.0 (3.9%) 5.4 (1.2%) 99.6 (22.7%) 92.3 (21.0%) 0.945 0.949
Cuiaba (MT) 25346 467.6 5.6 (1.2%) 7.8 (1.7%) 125.8 (26.9%) 125.8 (26.9%) 0.913 0.916
Curitiba (PR) 1369 468.0 63.2 (13.5%) 11.9 (2.5%) 140.2 (30.0%) 117.2 (25.1%) 0.922 0.924
Natal (RN) 15558 543.3 26.5 (4.9%) 19.2 (3.5%) 74.4 (13.7%) 71.7 (13.2%) 0.976 0.976
Ourinhos (SP) 17013 468.7 32.7 (7.0%) 14.0 (3.0%) 95.8 (20.4%) 86.2 (18.4%) 0.948 0.951
Palmas (TO) 19212 486.2 2.1 (0.4%) -4.6 (-0.9%) 136.7 (28.1%) 136.8 (28.1%) 0.889 0.889
Petrolina (PE) 27292 481.1 21.0 (4.4%) 26.6 (5.5%) 85.5 (17.8%) 88.4 (18.4%) 0.962 0.962
Saoluiz (MA) 14652 442.7 10.4 (2.3%) 9.0 (2.0%) 111.2 (25.1%) 110.5 (25.0%) 0.922 0.923
Sombrio (SC) 6697 413.9 32.2 (7.8%) 14.4 (3.5%) 100.8 (24.4%) 86.2 (20.8%) 0.943 0.951

Top of page