Kawa Stream TMDL Project

Rainfall and Stream Flow Considerations

REPORT INTRO Rainfall Stream Flow

    When rain falls within the watershed, some portion of the water soaks into the ground, some is relatively quickly evaporated back into the air, and the remainder, if any, flows into Kawa or one of its tributaries. The water that soaks into the ground may contribute to a body of water within the ground (called an aquifer) and may eventually reach the stream, or may be withdrawn by plants and returned (by evapotranspiration) to the atmosphere. The proportion of a given rainfall that goes into these various compartments of the water cycle can be estimated on an average basis, but will not be exactly the same from rainfall event to rainfall event, or from place to place for that matter. For example, rain falling on a street is most likely to become run-off. However, if the rainfall is not substantial and the pavement is hot, most of the water can evaporate before ever reaching the drainage system and then the stream. Observations made during rainfalls of various intensities have led to the following working values for the relationship between rainfall and run-off in the Kawa watershed:

    • Rainfall under 1 mm (0.04 in) - no appreciable runoff at all, even from streets;
    • Rainfall between 1 mm and 6 mm (0.24 in) - runoff from pavement, but little or no flow of runoff into stream from residential district;
    • Rainfall above 6 mm - runoff in residential area flows from drains into Kawa Stream;
    • Rainfall above 25 mm (1 in) - runoff from forest slope areas; upper intermittent reaches of Kawa Stream could start to flow.
    • Rainfall up to 100 mm (4 in) contributes about 20% of rain to peak flow (Oceanit calculations).
    Note that these values derive from observations on "isolated" downpours. Because rainfall is such a complex phenomenon, we do not have a good estimate of how much time must elapse between one downpour and the next for each to be considered an isolated event. This interval is likely to be several to many hours and for practicality, the values above could be applied to either 12-hour or 24-hour rainfall totals.

    Rainfall has been monitored for the Kawa TMDL project by a gage located near the lower end of Akimala Street, very close to water quality Station 007. This gauge is central in this watershed, which is likely to have a rainfall gradient increasing from northeast to southwest. Data from this gage are summarized in various ways, but the recordings from the tipping-bucket gage provide a record every time 0.01 in (.25 mm) of rain accumulates in the gage.

    We located data from four weather stations (Table 1) in the Kane`ohe area covering at least a portion of the recent several decades that might be applied to measured peak flows in Kawa Stream (see below). Station Kaneohe 838.1 is active and located near the Kaneohe Civic Center; Station Kaneohe Mauka 781 is near Baskerville Spring in the Haiku area; Station Kaneohe Bay MCAS is active and located at Marine Corps Base Hawaii, Kaneohe Bay (MCBH); Station Luluku 781.7 is situated mauka of Likelike Highway near the BWS pumping station at an elevation of 280 feet. Both 781 and 781.7 are in areas thought to be wetter on average than any part of Kawa watershed, whereas MCBH experiences generally less rain than is received on average at the Kawa Watershed. Other weather stations have existed in the area, but provide little or no data applicable to the time frame and location of concern. Station 838.1 (Kaneohe) located approximately 1.8 km (1.1 mi) NNW from the center of Kawa watershed is thought to be most relevant for our purposes.

    Table 1. Rainfall gaging stations in the Kane`ohe Area.
    Name:Luluku ~ 781.7Kaneohe Mauka ~ 781
    Coop ID:515655513113
    lat/long:21 23' N / 157 49' W21 25' N / 157 49' W
    Elevation:85.3 m (279.8 ft)57.9 m (189.9 ft)
    In service:05/01/67 - Present10/01/49 - 06/30/98

    Name:Kaneohe ~ 838.1 MCBH, Kaneohe Bay
    Coop ID: 513117 WBAN: 22519
    Lat/long: 21 25' N / 157 48' W 21 27' N / 157 47' W
    elevation: 18.3 m (60.0 ft) 3.0 m (9.8 ft)
    In service: 03/31/85 - Present 01/01/41 - Present

    There are many ways to look at rainfall records. Figure 1 compares average monthly rainfall amounts recorded over the past 14 years in Kan`eohe (1985 -1999) with the current monthly totals recorded at our gage on Akimala Place. Vertical bars show the range of monthly totals recorded for each month of the year. The logarithmic scale on the vertical axis tends to smooth out differences between wet and dry seasons. However, the higher average rainfall and much greater variability of monthly totals during wet season months (September through April in Kane`ohe) as compared with dry season months (May - August)1 are evident. Also clear, is the fact that the months of the project period (2000-2001) were below average for rainfall in Kane`ohe (comparison of rainfall amounts for months of record show the two locations receive similar amounts of rainfall) -- well below average for some months. February 2000, May 2000,and January 2001 were below the lowest amounts previously recorded for those months in the period of record.

    Sta. 838.1 rainfall record
    Figure 1. Monthly rainfall record for project gage in Kawa watershed compared with 14-year record for Kane`ohe (Sta. 838.1) at Kane`ohe Civic Center.

    Stream Flow

    Estimates of stream flow in Kawa Stream comes from a variety of sources. Although it was a goal of the project to obtain flow measurements coincident with water samples, this proved elusive for a variety of reasons. Repair of a damaged monitoring wier (see Oceanit) provided essentially no storm flow data and only one brief period of continuous water level and water quality measurements before the wier was destroyed by a freshet.

    The only USGS data we could locate for Kawa Stream are for a peak flow gage which was generated the data as shown in Table 2 between 1965 and 1994.

    Table 2. Peak flow discharge in Kawa Stream at USGS Station Number 16265000
    DatePeak Discharge
    (cfs)(ft above datum)
    05/02/65 4750 17.0
    01/27/68 2260 11.3
    02/01/69 5290 17.9
    01/26/70 580 6.37
    04/24/71 758 6.96
    04/14/72 779 7.03
    01/27/74 548 6.26
    05/12/77 557 6.29
    10/30/78 450 6.08
    01/08/80 1930 10.45
    05/07/81 929 7.53
    03/14/82 779 7.03
    10/28/82 785 7.05
    02/14/85 644 6.58
    10/19/85 446 5.92
    12/31/87 1460 9.11
    04/04/89 902 7.44
    01/16/90 440 5.93
    11/13/90 1300 8.66
    02/14/92 398 5.76
    12/30/92 776 7.02
    02/15/94 602 6.44
    10/16/94 539 6.23

    Since historical daily flow data are unavailable for Kawa Stream, a correlation analysis of Kawa Stream peak flow and rainfall from the Kane`ohe rain gage stations (Table 1) was carried out using cumulative rainfall amounts for several days. The results are given in Table 3. Correlation coefficients were low for most of the comparisons. This is not unexpected for several reasons. Firstly, there is considerable variation in rainfall in different parts of Kane`ohe. Secondly, peak flows in a small stream such as Kawa Stream may be caused by different rainfall patterns; e.g., steady rainfall over a several day period or, at the other extreme, torrential rainfall over a several hour period. Thus, the daily (24-hour) rainfall data analyzed here may be too long a time period to discern torrential rainfall patterns. Finally, the NCDC rainfall data are reported for the 24-hour period beginning at 0800 the day before the record date. Thus, rainfall reported on the day in question may have fallen before or after the period of peak flow. Nevertheless, the daily rainfall data for Station Kaneohe Mauka2 781 demonstrate a strong correlation with the Kawa Stream peak flow data (P = <0.0001). Unfortunately, the Kaneohe Mauka 781 station was decommissioned in 1998. If this station had remained active, it might have been possible to get "order of magnitude" estimates of Kawa Stream peak flows for the study period.

    Table 3. Correlation coefficients between peak flow (cfs) in Kawa Stream and cumulative rainfall from selected weather stations for a several day period.
    	Correlation Coefficient (r)
    Station     	0-day	1-day	2-day	3-day
    Kaneohe 838.1	0.35	0.50	0.32	0.28
    Kaneohe Bay MCAS	0.37	0.34	0.65	0.47
    Luluku 781.7	0.31	0.23	-0.02	0.02
    Kaneohe Mauka 781	0.74	0.76	0.72	0.71

    Two gaging stations were set up on Kawa Stream by a consultant to Bay View Golf Course (Nance, 1999). These gages recorded stream flows throughout most of 1997 and 1998. The upper gage was located just downstream of Kaneohe Bay Drive (below Sta.022); the lower gage at Station 005. All of the raw data have not been made available, but the graphical presentation in the prepared report suggest the "average" flows shown in Table 4. Differences between the gages were too small to discern from the graphical presentation, however differences between years demonstrate the difficulty of establishing typical values for rainfall or stream flow.

    Table 4. Averaged flows for Kawa Stream from two years of stream flow gage records.
    Wet Season2.570.80.8524.1
    Dry Season1.542.50.25.7

    Because instantaneous stream flow estimates were obtained only occasionally in 1999-2001 and not continuously anywhere, another approach to estimating pollutant loadings (= parameter concentration times flow volume) had to be found. Selected peak flow discharges recorded in lower Kawa Stream (Nance, 1999) were compared. Although the largest storm recorded was a 4.3 inch rainfall, the relationship between rainfall and runoff (peak-flow volume) was found to be quite linear. Only about 20% of the rainfall appeared in the peak flow discharge of Kawa Stream for the storms that were monitored. This value would likely increase in larger storms (Oceanit, 2001).

    The stream flow data collected in 1997-98 can be compared with rainfall from the same period (Kaneohe Civic Center gage) to roughly estimate stream flow and mass balance conditions given only rainfall data in (for example) the year 2000. Table 5 illustrates these calculations. Known or measured quantities are given in bold in the table. For each of the years 1997 and 1998, the total rainfall (in millimeters) is multiplied by the watershed area (4.04 km2) to give a volume of water received (here, in millions of liters). Of this volume, 20% is conservatively estimated to discharge as freshet flow (direct runoff). The average baseflow for each season is used to estimate total baseflow volume by assuming six months of wet season and six months of dry season flow each year. Averages for the years 1997-98 are used to calculate values for the year 2000 based upon the recorded rainfall.

    Table 5. Kawa Stream flow and mass balance calculations (after Oceanit, 2001).

    Annual rainfall (mm) 16926401166905
    Annual rainfall volume6834258647103656
    Stream base flow, wet season (cfs)1.750.851.301.01
    Base flow volume, wet seasonA 784381582452
    Stream base flow, dry season (cfs)1.580.20.890.69
    Base flow volume, dry seasonA70489396308
    Peak flows volume (20% of rainfall)1367517942731
    Losses (evaporation & other)3980159927902165
    Annual stream discharge285498719201491

    Units are millions of liters unless indicated otherwise.
    AAssumes 183 wet season days and 182 dry season days

    The difference obtained by subtracting base flow amounts and freshet flow amount from the annual total received is ascribed to "losses", such as evaporation and evapotranspiration. These calculated losses amount to between 3 and 8 cm per month, which are reasonable estimates considering pan evaporation alone could be 2-3 cm per month in Kane`ohe. Of the total amount of water discharged from Kawa Stream, the average partitioning between base flow and freshet flow is 51% and 49%, respectively. The annual discharge volumes in Table 5 may be used to calculate loading of various substances based upon average concentrations in Kawa Stream under various conditions of wet and dry season base flow, and freshet flow.

    Page Footnotes:
    1 -- Statewide, the dry season is "officially" from May through October.
    2 -- This rain gage is in a wetter part of Kane`ohe than the Kawa watershed, so in general, the amount of rain recorded cannot be applied to rainfall around Kawa. However, the correlation analysis suggests that when heavy rainfall occurs at Kaneohe Mauka, heavy rainfall is also occurring at Kawa. The approximate location of the former gage can be seen from the latter in a SW direction (mauka).

    January 4, 2002 — Webmaster