Kawa Stream TMDL Project

Watershed influences on Kawa Stream

REPORT INTRO Subwatersheds Runoff Quality Storm Drains

The overall area of the Kawa Stream Watershed (State ID No. 2-11) is 1.56 mi2.1 For purposes of this study, the watershed is divided into 9 subwatersheds, based partly upon the locations of water quality sampling stations, many of which were placed to sample the various branches of the stream near their confluences. In the lower watershed, some subwatersheds are delineated to incorporate slopes on both sides of the central branch of the stream to reflect any and all runoff inputs between selected water quality stations. Table 1 provides information on these subwatersheds or "basins."

Table 1. Subwatershed basin areas and percent land use designation, Kawa Stream watershed, Kane`ohe.
Total Area Land Use (%)
Basin No. (mi2)


forest cemetery residential streets park school golf commercial
1 0.23 56.1 43.7 0.0 0.2 0.0 0.0 0.0 0.0
2 0.08 23.3 75.5 0.3 0.9 0.0 0.0 0.0 0.0
3 0.24 14.2 9.0 57.3 6.52 0.0 13.0 0.0 0.0
4 0.34 61.6 0.0 35.7 2.7 0.0 0.0 0.0 0.0
5 0.12 0.0 0.0 53.4 3.9 0.0 20.5 0.0 22.1
6 0.04 0.0 0.0 83.0 6.0 0.0 11.0 0.0 0.0
7 0.16 15.0 0.0 51.9 5.7 6.0 19.2 0.9 1.3
8 0.11 0.0 0.0 75.2 7.2 0.0 0.0 17.6 0.0
9 0.23 46.9 0.0 17.3 2.1 0.0 0.0 30.7 0.0

ALL 1.56 34.4 11.8 35.9 3.5 0.6 5.9 5.8 1.9

Basin 1 -- Basins 1, 4, 7, and 9 include the forested crest of the watershed bordering the low hills to the east of Pikoiloa. Within Basin 1 this forest (0.13 mi2) and two cemeteries (Veterans and Hawaiian Memorial Park; 0.10 mi2) occupy all of the land. Paved roadways within the cemeteries comprise the only other land use category present. This subwatershed includes the main or central branch of Kawa Stream and other intermittent branches draining forested lands down to the confluence with the upper west branch. WQ Station 08R represents water flowing out of this basin. Stations 009 and 010A are in the basin. Samples from 010A represent mostly cemetery runoff collected in underground drains.

Basin 2 -- Basins 2, 3, and 5 have their upper ends along Kamehameha Highway bordering the Kaneohe Stream watershed. Basin 2 is almost entirely cemetery (Hawaiian Memorial Park; 0.06 mi2), except for the forested gulch (0.02 mi2) of the upper west branch of Kawa Stream. Outflow from this basin is measured at Station 08L (HDOH Sta.8).

Basin 3 -- Basin 3 straddles the central branch, upper middle reach of Kawa Stream between the confluence with the upper west branch and the confluence with the middle east branch. Although some forest area is included (0.03 mi2), the majority of Basin 3 is residential housing and streets (0.13 mi2). The area includes Kaneohe Elementary School (0.07 mi2. Water quality in this segment of Kawa Stream is monitored by Station 007, although Station 016L marks the lowest point in the subwatershed. Runoff water quality during storms was measured at culverts C1, C3,and C4, which all represent residential street drainages.

Basin 4 -- Although one of the larger subwatersheds, the outflow as monitored at Stations 006 and 016R tends to be small. The area is divided between an upper forested portion (0.21 mi2) and a lower developed portion in residences and streets (0.13 mi2).

Basin 5 -- Basin 5 drains through the lower west branch, which arises as drainage culverts at the Windward City Shopping Center and flows through a modified channel at Castle High School. Outflow from this subwatershed is monitored at Station 012L. Station 018 monitors the perennial flow from the culvert outlet for the street drainage system around the shopping center. Land uses are commercial (0.02 mi2), school (0.02 mi2), and residential (0.07 mi2).

Basin 6 -- This subwatershed borders both sides of a short segment of the middle reach of Kawa Stream between stations 06L and 012R. Nearly all of the area is developed as residential neghborhood and streets (0.04 mi2). A small area (<0.01 mi2) of farmland at Castle High School is included.

Basin 7 -- Like the previous subwatershed, this basin consists of two areas bordering a short segment of Kawa Stream, here between stations 012R and 022 (Kaneohe Bay Drive bridge). However, street drains bring runoff in from some distance on both sides: as far away as MacDonalds at Kaneohe Shopping Center to the west and the crest of the eastern hills above Pohi Nani on the other side. Land use areas are park (0.01 mi2), school (0.03 mi2), residential (0.09 mi2), and forest (0.02 mi2).

Basin 8 -- This subwatershed drains mostly parts of the Puohala neighborhood and upper end of the Bay View Golf Course. Basin 8 encompasses Kawa Stream between stations 022 and 005; essentially the lower reach of the stream. Station 005 monitors water quality in the stream exiting this subwatershed. The land is comprised of residences and streets (0.07 mi2) and golf course (0.02 mi2).

Basin 9 -- This subwatershed encompasses most of Bay View Golf Course (0.07 mi2), but also areas mauka of Kaneohe Bay Drive, including residential development (0.03 mi2) and forested slopes (0.10 mi2). These areas drain through street culverts and intermittent stream channels that enter Kawa Stream at the golf course. Most of the central channel of the stream is estuarine here, with the transition occurring a short ways above Station 004. Stations 004, 003, 002, and 001 monitor conditions in this estuary down to the mouth (opening on Kane`ohe Bay). A freshwater marsh lies along one side of the man-made estuary channel, collecting several drainages coming in from upslope.

Runoff Quality

Two sources contribute water to Kawa Stream and therefore separately influence water quality in the stream: 1) ground water that seeps or flows from springs, and 2) direct run-off from the land at times when rain is falling on the watershed. When rain occurs in the watershed, some portion of the water soaks into the ground, some portion is relatively quickly evaporated back into the air, and the remainder, if any, flows into Kawa or one of its tributaries. The relationships between rainfall and stream flow in this system are discussed beginning with Rainfall.

Both sources of water begin as fairly clean rain drops, but their quality changes as the drops coalesce and flow towards the stream or soak into the earth. Falling droplets loosen particulates on the ground and these can be carried down into the stream if a surface flow develops. Additional particulates may be eroded from the soil surface where streamlets gain force. Water soaking into the soil will not carry particulates with it, but will transport soluble substances downward towards the aquifer. Rainfall on open soil can erode soil particles into the stream raising the turbidity and suspended solids. Both surface flow and ground water movement can carry easily dissolved substances, such as nitrates, into the stream. Increased flow within the stream resulting from rainfall will enhance the ability of the stream to carry particulates, and suspended solids will increase from loose material in the stream bed and eroded from the banks. However, because a heavy rain may contribute a lot more water than substances available to be dissolved from the ground surface, the actual concentration of these in the stream might decrease until run-off subsides.

Although other pollutants such as petroleum hydrocarbons were not considered analytically for any of the samples collected in Kawa Stream TMDL project, a significant observation may require follow-up in the future. When sampling at Station 18 on May 18, 2001, a clearly visible oil sheen and gasoline-like odor was noted in the storm runoff discharging from the culvert outlet. This culvert drains much of the Windward Shopping Center, and the source of the petrochemical contamination may be deposits on the parking-lot surface. The unseasonal mid-May rainfall was heavy and followed after a period of months of no run-off producing rains. Perhaps pollutants accumulated over this period of time on the roadway surfaces, to be flushed off by the downpour.

Rainfall in the Kawa watershed mobilizes most of the dissolved and particulate substances that enter the stream, establishing stream water quality, influencing water quality in the estuary and, ultimately, in Kane`ohe Bay. Rainfall also creates the sudden or flashy high flows (called "freshets") that erode the stream bed and banks, carrying this eroded material eventually into the Bay. Table 2 compares turbidity and TSS for each water quality station, partitioning the events between runoff or freshet ("RNF") and non-runoff or normal flow ("NRM") samples.

Table 2. Summary of suspended particulates measurements in Kawa Stream
comparing all runoff influenced values with all non-run-off valuesA.
Station Turbidity (ntu)
TSS (mg/l)
Sta. 010A 26 (4) ---- 13 (4) ----
Sta. 009 29 (5) ---- 19 (5) ----
Sta. 08R 53 (6) 2 (7) 35 (6) 2 (6)
Sta. 08L 31 (6) 5 (25) 18 (6) 4 (25)
Sta. 007 73 (32) 9 (28) 53 (30) 5 (29)
Sta. 016L 52 (4) 9 (1) 34 (4) 5 (1)
Sta. 016R + 006 32 (7) 8 (24) 19 (7) 5 (26)
Sta. 012R 60 (6) 7 (3) 47 (5) ----
Sta. 012L 43 (7) 7 (3) 39 (6) ----
Sta. 022 47 (4) 9 (2) 30 (4) 6 (2)
Sta. 005 ---- 5 (25) ---- 2 (26)

GRAND MEANSB 50.7 (81) 6.3 (118) 34 (77) 4 (115)

A "RNF" geometric mean values are from samples collected by AECOS/OCEANIT and marked "r" on the data compilation. "NRM" geometric mean values are from all HDOH data plus AECOS data for events not marked "r". Number of data points given in parentheses following geometric mean value in bold.
B Note that one or two locations may dominate these grand mean values.

In Tables 3 and 4, the same division between run-off and non-run-off influenced events is used to compare means for dissolved inorganic nutrients (nitrate + nitrite and ammonia) and organic nutrients. The dilution of soluble nitrates is evident in this treatment: highest station values occur under conditions of no run-off into Kawa Stream. It is easy to conclude that ground water is the primary source of nitrates to the system2. The middle east branch (Sta.s 016R and 006) appears to be an exception to this pattern. This branch is concrete lined from Sta. 006 to the intermittent reach and perhaps not well connected to the ground water. Flow declines substantially during the dry season. The pattern for ammonia (Table 3) is less clear, although there is a tendency for all stations to show enhanced ammonia values during run-off influenced events.

Table 3. Summary of dissolved nutrients measurements in Kawa Stream
comparing all runoff influenced values with all non-run-off influenced valuesA.
Station Nitrate + nitrite (ug/l)
Ammonia (ug/l)
Sta. 010A 133 (3) ---- 27 (3) ----
Sta. 009 174 (4) ---- 53 (4) ----
Sta. 08R 268 (3) 1046 (8) 5 (3) 2 (7)
Sta. 08L 574 (3) 873 (26) 13 (3) 10 (25)
Sta. 007 ---- 617 (26) ---- 49 (26)
Sta. 016L 432 (2) 963 (1) 41 (2) 21 (1)
Sta. 016R + 006 302 (5) 148 (26) 61 (4) 46 (26)
Sta. 012R 306 (3) 369 (3) 58 (3) 42 (3)
Sta. 012L 665 (5) 3039 (3) 33 (4) 15 (3)
Sta. 022 378 (3) 927 (1) 66 (3) 25 (1)
Sta. 005 ---- 1054 (25) ---- 22 (25)

GRAND MEANSB 325 (31) 585 (119) 35 (29) 23 (117)

A "RNF" geometric mean values are from samples collected by AECOS/OCEANIT and marked "r" on the data compilation. "NRM" geometric mean values are from all HDOH data plus AECOS data for events not marked "r". Number of data points given in parentheses following geometric mean value in bold.
B Note that dry measurements are weighted, in terms of number of analyses, to the HDOH stations in these grand mean values.

Total N values (Table 4) follow the nitrate pattern, which is not surprising given the fact that dissolved nitrate is a significant proportion of almost every Total N measurement. When the nitrate and ammonia values are removed, the remaining total organic nitrogen (TON) is shown to increase during runoff influenced periods, reflecting organic matter brought into the stream by runoff.

Table 4. Summary of total and organic nutrient measurements in Kawa Stream
comparing all runoff influenced values with all non-run-off influenced valuesA.
Station Total N (ug-N/l)
TON (ug-N/l)
Total P (ug-P/l)
Sta. 010A 711 (3) ---- 528 (3) ---- 292 (3) ----
Sta. 009 832 (4) ---- 532 (4) ---- 361 (4) ----
Sta. 08R 599 (3) 1134 (8) 313 (3) 81 (7) 145 (3) 13 (8)
Sta. 08L 883 (3) 1487 (22) 277 (3) 142 (21) 164 (3) 76 (26)
Sta. 007 ---- 954 (25) ---- 220 (25) ---- 26 (24)
Sta. 016L 1426 (2) 981 (2) 849 (2) 352 (2) 362 (2) 57 (2)
Sta. 016R + 006 858 (5) 559 (26) 499 (4) 291 (25) 200 (5) 35 (27)
Sta. 012R 1165 (3) 627 (3) 669 (3) 205 (3) 309 (3) 89 (3)
Sta. 012L 1314 (5) 3243 (3) 584 (4) 4 (3) 208 (5) 65 (3)
Sta. 022 1184 (3) 1120 (1) 649 (3) 168 (1) 259 (3) 57 (1)
Sta. 005 ---- 1417 (25) ---- 271 (25) ---- 36 (25)

GRAND MEANSB 956 (31) 1038 (115) 493 (29) 191 (112) 238 (31) 39 (119)

A "RNF" geometric mean values are from samples collected by AECOS/OCEANIT and marked "r" on the data compilation. "NRM" geometric mean values are from all HDOH data plus AECOS data for events not marked "r". Number of data points given in parentheses following geometric mean value in bold.
B Note that dry measurements are weighted, in terms of number of analyses, to the HDOH stations in these grand mean values.

Total P in Kawa Stream is nearly always enhanced by freshets (see Figure 1), indicating a strong correlation between phosphates and particulates loading. Soluble, inorganic phosphates (called orthophosphates or ortho-P) are infrequently measured, although presumably comprise a significant portion of the non-runoff influenced Total P values, as shown by historical data for this stream. The historical data suggest no significant seasonal variation in ortho-P. Soluble phosphates tend to bind with the iron-rich soil particles, and thus are not readily transported to the ground water. Ortho-P is not nearly as abundant in springs and seeps as soluble nitrates, even where use of both in yard fertilizers might be considerable.

Suburban Storm Drains

During the study, samples were collected from several storm drains in the residential area of the watershed (see data) to compare street runoff quality with stream water quality. These sample sites, labeled C1 through C5, all represent runoff from residential surfaces that input to the stream whenever sufficient rainfall occurs. A few other sample locations also represent runoff inputs through drains, but appear to be perennial flowing (e.g., Sta. 18) and are thus presumed to be tapping local springs.

The samples of outflow from drains that are normally dry have been combined into a single data set which is summarized in Table 5. Because all of these samples represent run-off from the residential land use type, combining results from different drains should not obscure patterns in the data. Temporal variables such as intensity of rainfall, preceding period of no rainfall, and timing relative to peak run-off may be more influentual than location in these types of samples.

Table 5. Summary statistics for all samples collected from street drains flowing into Kawa Stream during rain storms (1999-2001).
uS/cm ntu mg/l ug-N/l ug-N/l ug-N/l ug-N/l ug-P/l

meansA 49.9 39.7 28 308 73 1056 1499 473
Std. Dev. 28.5 9.1-173 4-192 167-568 39-136 924-1207 1311-1713 282-794
count 5 29 26 6 6 6 6 5

A Means are geometric means for all parameters except conductivity, which is an arithmetic mean

For what are essentially street run-off samples, the mean conductivity is clearly less than is found in the middle reach of the stream during non-runoff influenced periods (see data), and below most measurements made in the stream at times influenced by run-off (i.e., freshets). Both turbidity and TSS means are comparable to stream values obtained during freshets, perhaps a bit less if comparison is made with station means in the same area (the middle reach). Ammonia is somewhat elevated (Table 3) and nitrate + nitrite is slightly depressed when compared with stream stations in the middle reach (Table 3). Mean concentrations of each of these parameters indicate that during freshets Kawa Stream water quality in the middle reach is similar to street run-off -- perhaps not a very surprising result as the volume of run-off quickly overwhelms base flow at such times. However, it was the impression of the sampler, when collecting storm samples at Sta. 007 and C3 (entering just downstream of Sta.007) that the stream was nearly always more turbid than the street run-off. Samples for turbidity and TSS from these two locations were collected more often than from any other single location, and samples were frequently collected close together in time (see data). The comparable geomeans were: turbidity = 37.8 (C3) and 71 (007) ntu; and TSS = 28.6 (C3) and 44 (007) mg/l. Thus, the stream was more turbid on average than the run-off entering through this drain.

It is interesting that the total N and total P means for these run-off samples exceed nearly every stream station sample mean for the data set influenced by runoff (Table 4). TON (TN minus nitrate, nitrite, and ammonia) at just over 1000 ug-N/l is high compared with freshet stream samples (Table 4). Since TSS is lower on average, either the suspended matter coming off the residential land is richer in organic matter or the run-off contains proportionately more dissolved organics than the stream (freshet) samples.

Page Footnotes:
1 -- The area of Kawa watershed is 1,336 acres or 2.086 mi2 according to Geographic Decision Systems (1994). The difference appears to reflect their inclusion within Kawa of the several small watersheds (e.g., Kokokahi) along Kaneohe Bay Drive to beyond the Saddle Road. These watersheds drain directly into Kane`ohe Bay and are technically not part of Kawa watershed. This source presents the areas of State Land Use Districts within the larger watershed as 73% Urban and 27% Conservation.
2 -- Although no wells were sampled for this study, a sample from a seep near Sta. 007 provided evidence of a high nitrate ground water concentration within the suburban district.

June 4, 2001 — Webmaster