PERFORMANCE OF THE HE ALGORITHM DURING THE CENTRAL AMERICAN RAINY SEASON OF 2001

1
PERFORMANCE OF THE H-E ALGORITHMDURING THE
CENTRAL AMERICAN RAINYSEASON OF 2001
2

• 1.- Introduction
• 2.- Methodology
• 2.1 Hydro-Estimator
• 2.2 Comparisons
• 3.- Data set
• 4.- Results and discussion
• 5. Future work

3
Introduction

• GOES digital data available since July 2001
• Rain gauge network has been decreasing over the
  years and radar data is not available over the
  area
• Previous results for Keith showed the best
  results for the H-E
• The performance of the H-E was evaluated during
  the CA rainy season to use the estimates for
  flash flood warnings in the near future.

4
Methodology H-E

• 1.- variant of the A-E (Vicente et al. 1998)
• 2.- cloud growth rate correction factor is not
  used
• 3.- uses a modified temperature gradient
  correction
• 4.- same temperature/rain rate relationship and
  Z(Mean-T)/SD
• 5.- moisture correction factor PW and RH

5
Comparisons

• Satellite estimates versus daily and 6-h rain
  gauge measurements
• Rain gauge image obtained from the rain gauges
• Each pixel of the rain gauge image is compared
  with the 9 pixels surrounding it from the
  corresponding satellite image (BEV comparisons)

6
Statistical measures used in the comparisons
7
Data set

• Orographic image at 10-km res.
• From 1515 UTC 20 August to 1445 UTC 22 August,
  from 1515 UTC 8 October to 1445 10 October, from
  1315 UTC July 17 to 1245 UTC July 18, 2001
  (Chantal, Iris, tropical wave)
• a.- 10.7 um TB
• b.- corresponding grid files from the Eta
  model u, v, PW and RH.

8
Data set (cont.)

• H-E estimates for days between May and Spetember
  2001 (26 mm/day or more)
• Daily rain gauge data from Belize and the Yucatan
  Peninsula from their Met Services.
• Daily rain gauge data from the IMN
• Rain gauge data every 15 min from rain gauge
  networks over Guatemala, Honduras, El Salvador
  and Nicaragua.

9
CHANTAL
Satellite estimates
Rain gauge values
10
TROPICAL WAVE
Satellite estimates
Rain gauge values
11
(No Transcript)
12
Comments to the results

• Better results for Chantal compared with those
  obtained for Keith. A HSS of 0.71 and 0.81 for a
  threshold of 26 and 52 mm/day were obtained.
  Only stations over Belize and the Yucatan
  Peninsula were used.
• Due to the few raingauges available over Belize,
  in order to gain comparable points in the
  analysis, the comparisons for hurricane Iris
  included stations in the Yucatan peninsula, where
  warmer IR temperatures were observed. This fact
  could have an important influence on the results
  obtained since the center of hurricane Iris was
  over Belize

13
Comments to the results (cont.)

• For the tropical wave case, the results too were
  most likely influenced by the location of the
  gauges over Costa Rica most stations are over
  the middle of the country where mainly light rain
  was observed.
• The 6-h analysis for tropical storm Chantal
  included stations over Guatemala, Honduras, El
  Salvador and Nicaragua. The 6-h totals were
  limited to fixed periods at 0, 6, 12, and 18 UTC.
  Besides, stations were not located close to the
  center of the storm which can have an importan
  influence on the poorly results obtained for
  these data.

14
Comments to the results (cont.)

• The rain gauge network used in each case seemed
  to limit and influence the results obtained in
  each case.
• Results from daily comparisons from May through
  September included stations over Costa Rica and
  Honduras mainly.
• The results obtained from rainy days with 26
  mm/day or more, show poor results as compared
  with those obtained for Chantal and the tropical
  wave case. Further classification (as a funcion
  of cloud size/pattern/temperature) within
  tropical systems restrict the comparisons because
  of the sparsity of rain gauges. However, the use
  of TRMM data could solve this problem in the
  future.

15
Future work

• Include an orographic map with a higher
  resolution
• Adapt the cloud top temperature-rain rate curve
  to different type of cloud systems based on case
  studies over the region.
• Validation of the blended GOES/microwave (Turk
  et.al 1998) over CA is underway at the University
  of Costa Rica. This is an algorithm which give
  more direct precipitation measurements
• Validation for shorter periods of time in order
  to use the results for flash flood warnings.

16
Future work

• The use of precipitation algorithms is very new
  in Central America. The use of these products by
  the Central American countries can help to focus
  future work in problems detected on the daily use
  basis.
• Two versions of the H-E are running at this
  moment in the server at the IMN, the one from
  this study and a new one which has been used for
  Florida. Comparisons of the two versions are
  underway at the University of Costa Rica.