The variability of mixed layer depth (MLD) and barrier layer thickness (BLT) has profound implications on energy exchange processes at the air-sea interface. More important is the role of MLD and BLT in the genesis and intensification of weather systems. The physical and chemical changes that take place within these layers have significance on biological productivity of the oceans. In this study, the monthly evolution of MLD and BLT for Indian Ocean was compared using the state-of-art world ocean atlas (WOA) and a recently developed comprehensive ocean atlas [referred to as new climatology (NC)]. The study area comprises the geographical boundaries encompassing 30°N to 60°S and 30°E to 120°E. Qualitative skill assessment of these variables demonstrates that NC is in good agreement with recently reported observational and modelling studies. This brings out the fact that MLD and BLT climatology derived from NC is better than that of WOA.
The quasi-homogeneous region in the surface layer of the ocean is termed as the mixed layer where physical quantities like temperature, salinity, and density are nearly constant with depth. The depth to which this uniformity prevails can be treated as the mixed layer depth (MLD). Proper quantification of MLD is important to understand heat storage capacity of the oceans attributed by the flux exchange at air–sea interface. Owing to high thermal inertia, upper layers of the ocean induces a ventilating effect in land–air–sea interaction processes. Hence proper knowledge in reliable estimates of MLD has consequences in physical, chemical, biological productivity, climate studies, and naval operations in the oceans.
The physical and chemical change which occurs in the mixed layer is very crucial for biological productivity in the world oceans (Kumar and Narvekar, 2005). A better description of MLD on time and space scales can have a significant contribution to understand the climate variability. Hence, to improve our understanding of the climate system in a better perspective, MLD climatology is an essential pre-requisite to the ocean modelling community. This information also supplements to improve existing parameterization schemes in ocean general circulation models (Kara, 2003; de Boyer et al., 2004). Several researchers (Monterey and Levitus, 1997; Kara et al., 2003) have attempted to study MLD climatology of the global oceans using the world ocean atlas (WOA). Studies utilizing Array for Real-time Geostrophic Oceanography (ARGO) data was made by de Boyer et al. (2004) and as such, studies for north Indian Ocean utilizing WOA and ARGO was attempted by Thadathil et al., (2007, 2008).
Past studies have categorized the vertical structure of upper ocean as three layers instead of two, namely (1) well mixed layer (where temperature and salinity are well mixed), (2) intermediate layer or the barrier layer (BL; which prevail in between the mixed layer and thermocline), and (3) deep stratified ocean (Lukas and Lindstrom, 1991; de Boyer et al., 2007). In this context, presence of BL can have significant impact on mixed layer dynamics and heat budget within the MLD (Sprintall and Tomczak, 1992). The impact of barrier layer thickness (BLT) on sea-surface temperature (SST) was investigated by Gregory and McPhaden (2009). Masson et al. (2005) used a coupled model for the south-eastern Arabian Sea (SEAS) where the extent of BL was used as a predictor for the onset of summer monsoon with a 2 month lead-time. In the tropical belt intense precipitation, wind, and circulation patterns along with river runoff can result in the formation of thick and persistent BL (Mignot et al., 2007). For the Indian seas, McPhaden et al. (2009) suggested that high SSTs in Bay of Bengal along with large pool of fresh water cap by river runoff and open ocean precipitation produces ‘BLs’. As such this phenomenon isolates the surface from cooling of sub-surface waters which can energize tropical storms in the Bay of Bengal. For the same region BL persists during late summer and post-monsoon periods. In addition, the associated hydrographical features can have profound implications on the biological productivity (Muraleedharan et al., 2007). Interestingly, BLs are not detected between the latitudinal belts 25° and 45° in both the hemispheres (de Boyer et al., 2007).
The intra-seasonal variability of BL in the North Bay of Bengal during Southwest monsoon period based on temperature and salinity profiles were reported by Vinayachandran et al. (2002). Observational studies by Shenoi et al. (2004) for the SEAS revealed behavioural patterns in the distributions of BLT. The seasonal variability of BLT for Arabian Sea and Bay of Bengal was reported utilizing a comprehensive database by Thadathil et al. (2007, 2008).
For the Indian Ocean, there are several reported works on MLD climatology such as Colborn (1975), Hastenrath and Greisher (1989), Rao et al. (1989) and Levitus (1982). On the basis of these reports, MLD is maximum during the southwest and northeast monsoon periods with existence of minimum MLD during the months of April and May in the Arabian Sea and Bay of Bengal. The earlier versions of MLD climatology for Indian Ocean were developed based on historical hydrographical data. However, uncertainty exists with regards to its spatial distribution, resolution as well as the data quality of these observations. The launch of ARGO ocean observing program in the Indian Ocean during the year 2000 had revolutionized the capacity in sub-surface profiling of the oceans. Since the inception of ARGO programme, number of profiling floats in the Indian Ocean has shown an increasing trend. The voluminous amount of information from these floats provides an opportunity to estimate MLD and BLT with better confidence than the earlier works for the Indian Ocean.
In this context, a comprehensive ocean atlas for the Indian Ocean utilizing ARGO data (hereinafter referred to as ‘new climatology’ termed as NC) was developed by Prasad Kumar et al. (2009). The development of this comprehensive atlas utilizes ARGO data for the period 2001–2006, as well as other in situ observations and WOA. The quality checked ARGO data for the Indian Ocean region was obtained from Indian National Centre for Ocean Information Services (INCOIS) an autonomous organization under the Ministry of Earth Sciences, Government of India. In this study, we report on the development of a monthly MLD and BLT climatology for the Indian Ocean. The MLD and BLT obtained from NC has been skill assessed with existing state-of-art climatology and other reported observational and modelling studies for the Indian Ocean.
2. Data and methodology
The investigations carried out in this study are based on two different monthly climatologies for the Indian Ocean viz; (1) NC for the Indian Ocean (Prasad Kumar et al., 2009) and (2) WOA (Antonov et al., 1998; Levitus and Boyer, 1994).
2.1. Overview of the NC
For the Indian Ocean, NC comprising of temperature (T) and salinity (S) fields at standard depths similar to the WOA was developed by Prasad Kumar et al. (2009). This atlas contains three-dimensional ‘T’ and ‘S’ fields up to a maximum depth of 1000 m in a quarter-degree latitude–longitude grid as shown in Figure 1. All available ARGO floats during the period 2001–2006 have been utilized in the preparation of this atlas. A Delaunay-Tesselation based technique using QHull algorithm (Barber et al., 1996) was used to produce uniform ‘T’ and ‘S’ fields from a non-uniform scattered database. For location corresponding to water points, a ‘small-sized neighbourhood box’ was defined to enable the search radius of surrounding ‘in situ’ points from ARGO floats with WOA grid point as centroid of the box. In the presence of an ‘in situ’ point the new database replaces the existing WOA point located at that point of interest. In addition, the entire Indian Ocean domain was sub-divided into pre-defined boxes using the clustered approach (for more details refer to Prasad Kumar et al., 2009). This work also discusses on the robustness of NC based on statistical measures with existing WOA and this provides the confidence for determination of MLD and BLT in the Indian Ocean.
2.2. Criteria for MLD determination
The determination of MLD depends on physical properties of seawater viz; temperature, salinity, and density. An appropriate criterion should be chosen in estimating MLD that depends on the problem being investigated. This study adopts this criteria for MLD estimation based on density variation (Δσt) determined from the corresponding temperature change ΔT (0.8 °C) using the equation of state (Kara et al., 2000). In this criterion a temperature drop of 0.8 °C was optimal to estimate the turbulent mixing penetration. Mathematically, it can be expressed in the form:
The layer extending from surface to the top of the thermocline is defined as the isothermal layer. The vertical extent of isothermal layer can be determined from temperature based criteria, where the temperature decreases to a value of 0.8 °C compared with the surface. The depth of this layer is termed the isothermal layer depth (ILD) (Sprintall and Tomczak, 1992; Kara et al.,2000; Kumar and Narvekar, 2005). During the presence of inversion layers Thadathil et al. (2007) defined ILD as the depth where ‘temperature at the base and top of inversion layer is equal’. They further postulated that the presence of inversion layer in temperature profiles over-estimates the determination of ILD (hereinafter referred to as TILD07). A similar observation was noticed in this study, where at one of the station (Figure 2) presence of inversion layer over-estimated the ILD based on the TILD07 criteria. It is to be noted that using this criterion, the BL is not captured wherein pycnocline depth was larger than the thermocline. As seen from Figure 2, the MLD is 50 m and inversion layer was present in the depth range 20–50 m (ILD is about 30 m). This shows that haline stratification is within the pycnocline layer and as per TILD07 criteria discussed above. On the basis of Kara et al. (2000) formulation the estimated BL was 25 m. The above discussions highlight the validity of Kara formulation to that of TILD07.
3. Results and discussion
The MLD and BLT was estimated using Kara formulation for the Indian Ocean. In addition, a comparative study was performed to skill assess the performance of WOA and NC. The salient features of MLD and BLT on a monthly scale highlighting various features and its geographical variability is discussed below and summarized in Table 1.
Table 1. Comparison of MLD and BLT for geometrically disconnected regions in the Indian Ocean basin
North Bay of Bengal
Shallow MLD prevails through out the year
From June to March BLs are present; maximum thickness is observed during January–February
Thick BL is noticed in the northwestern Bay of Bengal during most of the period
Central Bay of Bengal
Deep MLDs are observed during January–February and June–September. The MLD ranges between 25 and 50 m for most of the periods
The existence of BL is found during August–March. During March thick BL exist over most of the basin
Existence of deep MLD in NC is noticed during the months of July–August. These features are absent in the WOA
Eastern and western Bay of Bengal
Along east coast of India, MLD ranges between 25 and 50 m. These MLD are shallower during September–December
Presence of thick BL noticed during October–January over this region
Deeper MLD observed during June–July along the east coast of India in WOA. The MLD is shallow in NC and is in good agreement with Thadathil et al. (2007)
Waters surrounding Sri Lanka
During most of the months shallow MLD prevail, and during February deep MLD greater than 50 m is noticed
Thick BL observed during November–January south of Sri Lanka. No seasonal maintenance of BL seen
Significant differences in MLD as well BLT are noticed in both the climatology
SEAS (southeastern Arabian Sea)
During the monsoon and pre-monsoon seasons shallow MLD prevail over this region with MLDs in the range 25–50 m rest of the period.
Mature phase of BLT noticed during January.
In agreement with published literature findings the NC exhibit seasonal evolution of BL over this region, typically over-estimated in WOA climatology.
Northern Arabian Sea
Deeper MLD is noticed during December to January and shallower MLD less than 25 m are observed during May–October
Patches of BL observed only in February
MLDs determined from WOA climatology is over-estimated for this region
Central Arabian Sea
Throughout the year MLD greater than 25 m is noticed. During the months of December–February and June–September deeper MLD is observed in this region
Patches of BL are noticed in the central southwestern Arabian Sea from June–September
During July, MLD greater than 100 m and thick BL in CSWAS are noticed in the NC. These features are not prominent in the WOA
West coast of India
Shallow MLDs are noticed over this region during the monsoon and post-monsoon periods
Evolution of BL noticed during December–March in the SEAS
No BLs observed in this region during September–November in NC with existence of isolated patches in WOA
Central South Indian Ocean
Evolution of MLD noticed from March to November with a wide spread deeper MLD greater than 100 m during July–September.
The region between Madagascar to northwest Australia comprises a patch of thick BL during June–September
The MLD for this region is over-estimated in WOA
Southern region of South Indian Ocean
Relatively shallow MLD is observed during January–March. From June to November deeper MLD greater than 400 m develops having its prominence during July–September
BL observed all throughout the year having a mature phase from April to November
Significant differences in both the climatology are noticed for the south-eastern south Indian Ocean
The variability of MLD with both the climatology (WOA and NC) are shown in Figure 3(a) and (b). During January, deeper MLD was observed in both climatologies for the northern Arabian Sea in concurrence with the findings of Kumar and Narvekar (2005). In the northwestern Arabian Sea shallower MLD are noticed with NC, and these variations are not found in WOA. The northern Bay of Bengal experiences cooling resulting in the formation of haline stratified surface layer inhibiting deepening of MLD due to convective mixing. The findings in WOA and NC (Figure 3(a) and (b)); are in agreement with Thadathil et al. (2007). In WOA deeper MLD is seen in the central Bay of Bengal unlike the case in NC. Shallow MLD during January and February in southern Indian Ocean associated with few patches of BLs and having MLD distributions less than 125 m were noticed in both these climatology as reported by Dong et al. (2008).
Figure 3(c) and (d) illustrates BLT determined using the WOA and NC. For the northeastern Arabian Sea in NC, BLT varied between 30 and 50 m north of Gujarat which was absent in WOA. This shows that NC reproduces features of thick BL as reported by Thadathil et al. (2008). In the near vicinity of Somalia coast and central Arabian Sea, isolated patches of BL were noticed in the NC and such features are absent in WOA. In region south of Sri Lanka thick BL exceeding 50 m are also noticed in both the climatology (Durand et al., 2007). Between 40°S and 50°S thicker BL ranging between 20 and 150 m are observed in NC, whereas thinner BL are seen in WOA.
In the Arabian Sea, north of Oman (∼22°N, 60°E) shallower MLD are noticed in NC (Figure 4(b)), whereas deeper MLD are observed in WOA (Figure 4(a)). Climatologically a sub-tropical anti-cyclonic gyre (SAG) develop during this period over the Bay of Bengal resulting in formation of low saline water trap in the central and southern bay transported from northern head Bay region. The role of SAG lies in deepening isothermal layer and shoaling the MLD resulting in larger BLT. In horizontal scales larger spread of shallower MLD (Figure 4(a) and (b)) occurs over the central head Bay of Bengal (Thadathil et al., 2007). South of Sri Lanka, MLD ranging between 50 and 75 m was noticed only in NC. In the southwestern Bay of Bengal deeper MLD ranging between 50 and 75 m are observed in WOA, whereas shallower MLD of 25–50 m are noticed in NC. For the Southern Ocean there is no significant change in MLD as compared with previous month. A point of observation is that spatial distribution of BLT (Figure 4(c) and (d)) is sparse as compared with January.
In the west coast of India, north of Mumbai shallower MLD and thick BL of 30–40 m are observed in NC (Figure 4(b) and (d)) and these features are absent in WOA (Figure 4(a) and (c)). Also existence of isolated patches in BLT for the central Arabian Sea are noticed in NC which is absent in WOA. It is to be mentioned that these findings are not documented in earlier published works.
The comparison of both the climatology (Figure 5(a) and (b)) show persistence of shallower MLD over Arabian Sea as well as the Bay of Bengal. These features are in good agreement with Thadathil et al. (2007, 2008). For the Southern Ocean in both the climatology (Figure 5(a) and (b)) evolution and deepening of MLD at geographically disconnected locations are seen. They develop prominently into deeper MLD patches in the subsequent months.
In the SEAS during February and March both the climatology shows weakening of BLT off Kerala coast. The notable feature in NC is shallow BLT (Figure 5(d)) in the range between 10 and 20 m whereas relatively thicker BLT in WOA (Figure 5(c)). In NC (Figure 5(d)) shallow BLT was noticed along the east coast of India. In addition to above there are signatures of thick BL (∼75 m) in the central Bay. These variations are not found in WOA (Figure 5(c)). A comparison of Figure 5(c) and (d) shows that in NC, thicker BL ranging between 40 and 50 m is noticed in southern Sumatra coast and this being absent in WOA.
In the Arabian Sea location off Mogadishu (Somalia) deeper MLD (50–75 m) was noticed in WOA (Figure 6(a)) as seen in NC (Figure 6(b)) with a smaller spatial coverage. The density of ARGO floats is substantially larger in this area as seen from Figure 1. The MLD noticed from NC highlighting the above mentioned feature is smaller as compared with WOA. The reasons for this however need to be investigated in detail and is beyond the scope of this work. The other notable feature in WOA is the presence of deeper MLD (∼75 m) located between 80°E and 90°E along the equatorial Indian Ocean, and this is absent in NC. For the Southern Ocean evolution of MLD develops from March and this subsequently spreads over larger area during April. This feature is noted in both the climatology (Figure 6(a) and (b)).
The variability of BLT in both the climatology are shown in Figure 6(c) and (d). A common feature observed is the presence of BLT ranging between 30 and 40 m off Maharashtra in the west coast of India. Near south of Oman, isolated patches of BLT (∼10 m) were observed only in NC. Durand et al. (2004) reported BLT excess of 20 m in the SEAS. A comparison between WOA and NC shows its presence only in NC off Kerala coast. Along east of Sri Lanka presence of thick BLT (ranging between 30 and 40 m) is observed in both the climatology. The thick BLT was noticed only during the month of April dissipating further in subsequent months (∼10 m). This could be attributed to freshwater influx received from the northern Bay of Bengal.
The distribution of MLD for Arabian Sea show no significant difference compared with the previous month, whereas relatively deepened MLD is found over central Bay of Bengal. In addition, an enhanced tongue of MLD formation is noticed in the region west off Sumatra. Subsequently this feature extends on a basin wide scale till north Arabian Sea which is prominent till the month of August (Figure 10(a) and (b)). This tongue further dissipates during subsequent months. There is a sporadic deepening of MLD (about 175–200 m) in the Southern Ocean which is seen as isolated patches in both WOA and NC (Figure 7(a) and (b)). Earlier studies have reported deeper MLDs in this region during June to July, but the genesis of their formation occurs during this month. These findings are well supported with the observational evidence of Dong et al. (2008).
In the Arabian Sea (west of Somalia) from NC (Figure 7(d)) one can find prominent differences in BLT with that of WOA (Figure 7(c)). The thick BLT which appears during the month of April seen in both climatology located east of Sri Lanka disappears during May. Differences between the two climatology was also noticed for the Java Sea, such as thick BL and shallow MLD in NC, whereas thin BL and deeper MLD in WOA.
In the central Arabian Sea patches of MLD ranging between 50 and 75 m are noticed in the NC (Figure 8(b)) as reported by Kumar and Narvekar (2005) and absent in WOA (Figure 8(a)). Deepening of MLD in southern Indian Ocean is noticed in both the climatology (Dong et al., 2008). Deeper MLD's in excess of 400 m (Figure 8(a) and (b)) are noticed in this month. The spatial extent of these deeper MLD is found to exist from June till November. In the region off Madagascar (∼20°S) as winter prevails deeper MLD (∼125 m) is noticed during June in NC whereas this feature in WOA appears from July onwards.
In the coastal regions of northeastern head Bay of Bengal (extending until coastal limits of Myanmar and Thailand), deeper BLT ranging between 30 and 40 m are observed in both WOA and NC (Figure 8(c) and (d)). The formation of such thick BLT in the head Bay region initiates during this month and extends for a longer period of time (Thadathil et al., 2007).
In the NC (Figure 9(b)) for location corresponding to the southcentral Arabian Sea, MLD greater than 100 m are noticed as shown by Kumar and Narvekar (2005) and is absent in WOA (Figure 9(a)). At central Bay, patches of deeper MLD (∼75 m) are observed in NC, whereas these features are absent in WOA. The deepening of MLD could be attributed to intrusion of southwest monsoon current over this region (Vinayachandran et al., 1999) which brings high saline waters from Arabian Sea. The westward spread of thick BLT (Figure 9(c) and (d)) ranging between 30 and 50 m are observed in both the climatology over the Bay of Bengal region (Thadathil et al., 2007).
Deepening of MLD in south-central Arabian Sea is seen in both the climatology. Deeper MLD was found in NC (Figure 10(b)) compared with WOA (Figure 10(a)). A common feature in both climatology (Figure 10(a) and (b)) show the existence of shallow MLD in northern head Bay and relatively deeper MLD in the southwestern Bay.
Deeper MLD (∼100 m) at northwest Australia is noticed in WOA which is relatively shallow in NC (∼70 m). This is in agreement with report of Qu and Meyers (2005). One can find isolated patches of BL in the Central South-Western Arabian Sea (CSWAS) only in the NC (Figure 10(d)). Presence of deeper BLT between Madagascar and Australia (Figure 10(c) and (d)) is evident in both the climatology as reported by Mignot et al. (2007).
In southcentral Arabian Sea shoaling of deeper MLD was noticed this month in both the climatology compared with July to August (Figure 11(a) and (b)). The estimated MLD is less than 100 m as shown by Kumar and Narvekar (2005) which was found more prominent in NC (Figure 11(b)). In NC deeper MLD (>50 m) is noticed over the central Bay, and this is absent in WOA (Figure 11(a)).
Isolated patches of thick BL in CSWAS during August noticed in NC have dissipated during this month (Figure 11(c) and (d)). Along the east coast of Africa (∼15°S) patches of BL (<50 m) are observed in NC, whereas WOA show no existence of BLT.
The migration of thick MLD (about 100 m during September) in the central Arabian Sea was found to weaken and migrate further south during October (Figure 12(a) and (b)) propagating towards the Bay of Bengal basin in November. These features are noticed in both the climatology. The east coast of India off Visakhapatnam (Figure 12(c) and (d)) shows existence of thick BLT (∼75 m) in both the climatology. The existence of such thick BLT appears only during October (Vinayachandran et al., 2005) in both the climatology found to strengthen thereafter.
In the northern and central Arabian Sea, isolated patches of deeper MLD are noticed in NC (Figure 13(b)) and absent in WOA (Figure 13(a)). These patches are found to develop in the following months in WOA climatology. The evolution of shallow MLD (less than 25 m) south of Sri Lanka appears in both climatology (Figure 13(a) and (b)) during November extending till January (Rao et al., 2008).
It is evident from Figure 13(c) and (d) that during November, in both the climatology a thin BL appears in the SEAS and BL which completely vanishes in the CSWAS (Thadathil et al., 2008). A close look from both climatology reveal the existence of relatively deep and large spatial coverage of BL only in NC (Figure 13(d)) off Visakhapatnam.
In the central and northwestern Arabian Sea, MLD ranging between 50 and 100 m is observed in NC whereas for WOA this ranges between 50 and 75 m (Figure 14(a) and (b)). In the southern Indian Ocean both the climatology reveal shoaling of MLD (<200 m). In Bay of Bengal most of the region has thicker BL except in the south central head Bay region (Figure 14(c) and (d)). The thickest BL is found in the northeast region in both the climatology adjoining Myanmar coast. The non-existence of BL in the south central Bay of Bengal could be attributed due to presence of cyclonic gyre (Thadathil et al., 2007).
A comprehensive ocean atlas of MLD and BLT is very crucial to understand complex oceanic processes and for the benefit of scientific community involved in ocean modelling studies. The state-of-art WOA is being widely used as the first guess by ocean modelers, due to lack of choice in opting for a better climatology. A recently developed comprehensive ocean atlas for the Indian Ocean having high spatial resolutions utilizing ARGO and other quality checked in situ data was used in this study. The new ocean atlas provided an opportunity to investigate and compare the variability of MLD and BLT over the Indian Ocean basin on a monthly basis to that of WOA. The monthly evolution of MLD and BLT in both the climatology was then evaluated with reported observational and modelling studies. A critical qualitative evaluation was then conducted with the computed MLD and BLT. The overall comparison show that the new ocean atlas performs better to that of WOA. Based on this study, utilization of new comprehensive ocean atlas for the Indian Ocean could be recommended for understanding the oceanic processes in a better perspective.
We express our gratitude to Indian National Centre for Ocean Information Services (INCOIS) and ARGO community, for providing ARGO data. Mr. Naresh Krishna Vissa, would like to acknowledge the Council of Scientific and Industrial Research (CSIR), New Delhi, for funding his PhD research work at IIT Kharagpur. The authors would like to express their gratitude to anonymous reviewers for their constructive suggestions in improving the quality of manuscript.