Satellite support for CCE-LTER

Mati Kahru, mkahru@ucsd.edu

Click here for JPEG images

Click here for image data in HDF

Abstract

This document describes the products and analysis of satellite data provided in support of the CalCOFI-LTER cruises. The chlorophyll maps are merged from SeaWiFS (1997-present), Aqua-MODIS (2002-present), and Terra-MODIS (2007-present) data and SST from Terra-MODIS and Aqua-MODIS data. Terra-MODIS Chl-a data is used with lower priority, i.e. only when data from other sensors is not available. Other products will be added in the future.

Methods

· Methods for creating the full resolution satellite time series are described in Full-resolution satellite time series of the California Current area. In short, chlorophyll-a (CHL-A) datasets are created from SeaWiFS (1997-present), Aqua-MODIS (2002-present) and Terra-MODIS (2007-present) level-2 data. For the overlapping years CHL-A is a merged product from all available sensors. SST is merged from Aqua-MODIS and Terra-MODIS level-2 data.

· For each CalCOFI-LTER cruise the corresponding IEH file is downloaded from the CalCOFI website. The station and depth-related data is separated (with a utility iehconv.exe) and ingested into an Access database (spg2001.mdb). A list of stations with respective longitude, latitude, date, time is created with a database query (WIM_Lon_Lat_from_CalStations). Another query is used to make a similar table with the addition of surface chlorophyll-a. Composite images of CHL-A and SST are created from daily mapped datasets (with a utility wam_composite_list.exe) for the time period of the cruise ± 1 day and saved in the HDF format. For color coding of CHL-A I use the full range of 0.05-10.0 mg m^-3 (pixel values 48-200) and lookup table chl1_white_end.lut. For color coding of SST I use the full range of 10.05-35.25 ˚C (pixel values 87-255) and lookup table sst_kuring.lut. Locations of the standard CalCOFI stations are overlaid

· The following is an example of a cruise CHLA image (for CAL0404). This image is 8 times reduced from the original full-size image of 3840 x 3805 pixels. Because of the reduced size the station locations and many details are not visible. The full-size JPEG can be seen by clicking on the image below. For 2006 and later the reduced images are in the PNG format and 4 times reduced from the full size image.

Fig. 1. SeaWiFS-Aqua-MODIS merged CHL for CalCOFI-LTER 0404.

· The following is an example of a cruise SST image (for CAL0404). This image is 8 times reduced from the original full-size image of 3840 x 3805 pixels. Because of the reduced size the station locations and many details are not visible. The full-size JPEG can be seen by clicking on the image below.

Fig. 2. Aqua-MODIS SST for CalCOFI-LTER 0404.

· Match-up analysis was performed between in situ near-surface (average of samples in the top 5 m) and daily mapped CHL images (merged between SeaWiFS and Aqua-MODIS). Using match-up criteria of less than 12 hours time difference and at least 3 valid pixels in the 3 x 3 pixel window centered at the nearest pixel to the in situ station produced 93 coincident measurements. A link to the match-up file is provided here. The match-up file is a comma separated (CSV) text file with the following columns: Longitude,Latitude,DateGMT,TimeGMT,Cruise,Station,AvgOfChla,Image,Timediff_Hr,VarName,PointValue,Nin,Nout,Min,Max,Mean,StDev,Median. 11 of the match-up points looked as obvious outliers (marked with # in the first character position) and were excluded from Fig. 3. Most column names should be obvious; less obvious are: Nin = number of valid pixels, Nout = number of invalid pixels. The statistics and the pixel counts are given for the 3 x 3 pixel window centered at the pixel nearest to the in situ station.

· Below is a plot of the satellite-detected CHL versus in situ chlorophyll. At in situ Chl-a < 0.5 mg m^-3the two estimates agree very well but at in situ Chl-a > 1 mg m^-3 it is obvious that satellite estimates are significantly below in situ values. Satellite estimates of the two highest in situ values less than 50% of the in situ values. A number of factors can contribute to this discrepancy. It is possible that the satellite algorithm is somewhat biased but such bias can also be produced by the different foot-prints of the two sampling methods.

Fig. 3. Satellite-detected CHL versus in situ CalCOFI-LTER chlorophyll (mean for the top 5 m samples). CalCOFI-LTER cruises 0401, 0404, 0407 and 0411 are included. The satellite CHL is estimated as the median in a 3 x 3 pixel window centered at the in situ station. The vertical bars show the min-max range of valid pixels in the 3 x 3 pixel window. The dashed lines show (from top to bottom) one-to-one agreement, 2 times below and 3 times below in situ values.